Liquid Chromatography. Fundamentals and Instrumentation [Volume 1, 3 ed.] 0323999687


140 57 35MB

English Pages 846 Year 2023

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Front Matter
Chapter-1---Milestones-in-the-development-of-liquid-ch_2023_Liquid-Chromatog.pdf
Milestones in the development of liquid chromatography
Introduction
Developments before 1960
HPLC at the beginning
HPLC theory and practice
New HPLC modes and techniques
Selection of conditions for the control of selectivity
Columns
Particles and column packing
Stationary phases and selectivity
Equipment
Detectors
Apologies and Acknowledgments
References
Further reading
Chapter-2---Kinetic-theories-of-liquid-chromatograp_2023_Liquid-Chromatograp.pdf
Kinetic theories of liquid chromatography
Introduction
Macroscopic kinetic theories
Lumped kinetic model
The van Deemter plate height equation
General rate model
General rate model for monolith columns
General rate model for core-shell particles
Moment analysis
Lumped pore diffusion model
The equivalence of the macroscopic kinetic models
Kinetic theory of non-linear chromatography
Microscopic kinetic theories
Stochastic model
Stochastic-dispersive model
First-passage time
The Giddings plate height equation
Monte Carlo simulations of non-linear chromatography
Comparison of the microscopic and the macroscopic kinetic models
References
Chapter-3---Column-technology-for-liquid-chromatogr_2023_Liquid-Chromatograp.pdf
Column technology for liquid chromatography
Introduction
Column hardware and design
Column packing
Parameters/characteristics of LC columns
Retention and retention volume
Retention factor (k)
Selectivity or separation factor (α)
Efficiency (N) and plate height (H)
Resolution (Rs)
Peak symmetry
Hydrodynamic parameters-van Deemter equation
Column testing and evaluation
Engelhardt column test
Tanaka column test
Other column test methods
Conclusions and perspectives
References
Chapter-4---General-instrumentation-in-HPLC-_2023_Liquid-Chromatography.pdf
General instrumentation in HPLC*
Introduction
Instrumental set-up
Mobile-phase/solvent reservoir
Solvent delivery system
Sample introduction device
Pre-column apparatus
Column
Post-column apparatus
Detector(s)
Data collection and output
Post-detection eluent processing
Connective tubing and fittings
Related HPLC techniques
Further reading
Chapter-5---Liquid-solid-chromatography_2023_Liquid-Chromatography.pdf
Liquid-solid chromatography
Introduction
Retention and separation
The retention process (``mechanism´´)
Solute and solvent localization
Selectivity
Method development
Thin-layer chromatography
Selection of the mobile phase
Example of method development
Problems with the use of normal-phase chromatography
References
Chapter-6---Reversed-phase-liquid-chromatography_2023_Liquid-Chromatography.pdf
Reversed-phase liquid chromatography
Introduction
Parameters that affect retention
System properties
Surface excess adsorption
Interphase model
Formal mechanistic retention models
Semi-empirical retention models
Solvent strength
Exothermodynamic relationships
Temperature and Pressure
Temperature
Pressure
Linear free energy relationships
Solvation parameter model
System maps and analysis of system constants
Anomalies in system maps
Hydrophobic-subtraction model
References
Chapter-7---Secondary-chemical-equilibria-in-reversed-ph_2023_Liquid-Chromat.pdf
Secondary chemical equilibria in reversed-phase liquid chromatography
Introduction
Use of acid-base secondary equilibria
Changes in retention with pH
Buffers and measurement of pH
Ion-interaction chromatography
Retention mechanism
Common reagents and operational modes
Separation of inorganic anions
The silanol effect and its suppression with amine compounds
Use of perfluorinated carboxylate anions and chaotropic ions
Use of ionic liquids
Measurement of the enhancement of column performance using additives
Micellar liquid chromatography
An additional secondary equilibrium in the mobile phase
Hybrid micellar liquid chromatography
Microemulsion liquid chromatography
Metal complexation
Determination of metal ions
Determination of organic compounds
Redox reactions
Acknowledgment
References
Chapter-8---Ultrafast-high-performance-liquid-chroma_2023_Liquid-Chromatogra.pdf
Ultrafast high-performance liquid chromatography
Introduction
High-efficiency high-speed separations
Sub-2-μm fully porous packing materials
Superficially porous particles
High-throughput HPLC analysis
Short highly efficient HPLC columns with MS detection
Monolithic HPLC columns
Applications of ultrafast high-performance or high-throughput liquid chromatography
High-resolution applications in the analysis of pharmaceutical compounds and drugs of abuse
Two-dimensional chromatographic separations
High-throughput analysis with simplified sample preparation
Challenges with performing ultrafast high-performance HPLC separations
Conclusions
References
Chapter-9---Nano-liquid-chromatography_2023_Liquid-Chromatography.pdf
Nano-liquid chromatography
Introduction
Features of microfluidic analytical techniques
Improving sensitivity by reducing the chromatographic dilution
Efficiency and extracolumn band broadening
SPs and capillary column preparation
SPs used in nano-LC
Capillary column preparation
Instrumentation
Microfluidic pump systems
Nano-volume injection
Detectors
Hyphenation of nano-LC with mass spectrometry
Some selected applications
Proteins and peptide analysis
Food analysis
Environmental analysis
Pharmaceutical and clinical analysis
Legal and forensic analysis
Conclusions
References
Chapter-10---Hydrophilic-interaction-liquid-chromato_2023_Liquid-Chromatogra.pdf
Hydrophilic interaction liquid chromatography
Introduction
Principles of HILIC
Thermodynamics of adsorption
Adsorption kinetics
Stationary and mobile phases commonly employed in HILIC
Stationary phases
Silica gel
Chemically bonded phases
Ion exchange and zwitterionic stationary phase
Hydrophilic macromolecules bonded phases
Surface-confined ionic liquids stationary phases
Mobile phases
Applications
References
Chapter-11---Mobile-phase-selection-in-liquid-chromat_2023_Liquid-Chromatogr.pdf
Mobile phase selection in liquid chromatography*
Elution strength
Columns and solvents in RPLC, NPLC, and HILIC
Elution strength assessment
The Hildebrand solubility parameter and other global polarity estimators
Global polarity for solvent mixtures
Field of application of the chromatographic modes deduced from the Schoenmakers rule
Isoeluotropic mixtures
Solvent-selectivity triangles
The Snyders solvent-selectivity triangle
Prediction of the character of solvent mixtures
A solvatochromic triangle of solvent selectivity
Other solvent descriptors and alternative diagrams for classification and comparison of solvents
Practical guidelines for the optimization of mobile phase composition
Selecting the chromatographic mode
Description of retention using the modifier content as a factor
Systematic trial-and-error optimization of mobile phase composition in isocratic elution
Systematic trial-and-error optimization of mobile phase composition in gradient elution
Computer-assisted interpretive optimization
Use of combined mobile phases or gradients to achieve complete resolution
Additional considerations for the selection of solvents
Acknowledgments
References
Chapter-12---Co-solvents-and-mobile-phase-additives-_2023_Liquid-Chromatogra.pdf
Co-solvents and mobile phase additives in HPLC
Introduction
Fluorinated ion-pairing agents: Carboxylic acids, amines, and alcohols
Ionic liquids
Deep eutectic solvents (DESs)
Chaotropic anions
Kosmotropic ions
Surfactant additives
Conclusions
References
Chapter-13---Method-development-in-liquid-chromatog_2023_Liquid-Chromatograp.pdf
Method development in liquid chromatography
Introduction
Goals
A structured approach to method development
Column plate number, N: Term i of Eq. (13.1)
Retention factor, k: Term ii of Eq. (13.1)
Selectivity, α: Term iii of Eq. (13.1)
Gradient elution
Method development in practice
Resolution-modeling software
Priority of column screening
HPLC vs UHPLC
A systematic plan
Prevalidation
Validation
Documentation
Summary
References
Further reading
Chapter-14---Physicochemical-property-determinations-by_2023_Liquid-Chromato.pdf
Physicochemical property determinations by liquid chromatography
Introduction
Solvation properties determined from retention
Partition constants by liquid-liquid chromatography
Compound descriptors by reversed-phase liquid chromatography
Acid dissociation constants by reversed-phase liquid chromatography
Apparent formation constants by reversed-phase and ion-exchange chromatography
Interactions at solid interfaces determined from retention
Protein binding constants by affinity chromatography
Soil-water distribution constant by soil column liquid chromatography
Inverse liquid chromatography
Properties inferred from retention correlation models
Octanol-water partition constant by reversed-phase liquid chromatography
Soil-water distribution constants by reversed-phase liquid chromatography
Biopartition constants by reversed-phase liquid chromatography
Membrane permeation by biomimetic liquid chromatography
Immobilized artificial membranes
Micellar and microemulsion liquid chromatography
Physical properties determined by kinetic measurements
Molecular diffusion and mass transfer coefficients
Pore size distribution by inverse size-exclusion chromatography
References
Chapter-15---Theory-and-practice-of-gradient-elution-li_2023_Liquid-Chromato.pdf
Theory and practice of gradient elution liquid chromatography
Introduction
The effects of experimental conditions on separation
Gradient and isocratic separation compared
The effect of gradient conditions
The effect of column conditions
The effect of other conditions on selectivity
Method development
Problems associated with gradient elution
References
Chapter-16---Fundamentals-of-enantioselective-liquid-c_2023_Liquid-Chromatog.pdf
Fundamentals of enantioselective liquid chromatography
Introduction
Chiral selector
Inert carrier
Mobile phase
Thermodynamics of a separation process
Kinetics of a separation process
Enantioselective recognition mechanisms
Experimental techniques
Computation/molecular modeling
Conclusions and future trends
References
Chapter-17---Hydrophobic-interaction-chromatograph_2023_Liquid-Chromatograph.pdf
Hydrophobic interaction chromatography
Introduction
Historical perspective
Operating principles of HIC
Operation of HIC
Mobile phase
Elution gradient
Solute properties
Stationary phase
Detector
Applications of HIC
Solute purification
Protein refolding
Solute characterization
HIC as a part of multi-dimensional separation platforms
Challenges and limitations
Conclusions
Acknowledgments
References
Chapter-18---Ion-chromatography_2023_Liquid-Chromatography.pdf
Ion chromatography
Introduction
Definitions
History
Basic principles and separation modes
Ion-exchange chromatography
Ion-exclusion chromatography
Chelation ion chromatography
Zwitterionic ion chromatography
Eluents for ion chromatography
Typical eluents for anion exchange
Typical eluents for cation exchange
Instrumentation
Ion chromatography columns
Anion-exchange columns
Cation exchange columns
Eluent generators
Detection in ion chromatography
Conductimetric detection
Non-suppressed conductivity
Suppressed conductivity
Electrochemical detection
Charge detector
Amperometry
Spectroscopic detection
Photometric detection
Post-column reaction detection
Mass spectrometry
Two-dimensional ion chromatography
Applications
Industrial applications
Environmental applications
References
Chapter-19---Size-exclusion-chromatography_2023_Liquid-Chromatography.pdf
Size-exclusion chromatography
Introduction
Historical background
Retention in size-exclusion chromatography
A size-exclusion process
An entropy-controlled process
An equilibrium process
Band broadening in size-exclusion chromatography
Extra-column effects
Resolution in size-exclusion chromatography
Size-exclusion chromatography enters the modern era: The determination of absolute molar mass
Universal calibration and on-line Viscometry
Static light scattering detection
Size-exclusion chromatography today: Multidetector separations, physicochemical characterization, two-dimensional t ...
Conclusions
Disclaimer
References
Chapter-20---Affinity-chromatography_2023_Liquid-Chromatography.pdf
Affinity chromatography
Introduction
Basic components of affinity chromatography
Bioaffinity chromatography
Immunoaffinity chromatography
Dye-ligand and biomimetic affinity chromatography
Immobilized metal-ion affinity chromatography
Analytical affinity chromatography
Miscellaneous methods and newer developments
Acknowledgment
References
Chapter-21---Multidimensional-liquid-chromatograph_2023_Liquid-Chromatograph.pdf
Multidimensional liquid chromatography
Introduction
Fundamentals
Instrumental set-up and data analysis
Conclusions and future perspectives
References
Chapter-22---Process-concepts-in-preparative-chromat_2023_Liquid-Chromatogra.pdf
Process concepts in preparative chromatography
Introduction
Classical isocratic discontinuous elution chromatography
Mathematical modeling and typical effects
Other discontinuous operating concepts
Gradient chromatography
Recycling techniques
Closed-loop recycling chromatography
Steady-state recycling chromatography
Continuous simulated moving bed (SMB) chromatography
Classical SMB operating concept
Improved SMB operating concepts
Optimization and concept comparison
Conclusions
Acknowledgments
References
Chapter-23---Modeling-of-preparative-liquid-chromato_2023_Liquid-Chromatogra.pdf
Modeling of preparative liquid chromatography
Introduction
Column model
The equilibrium-dispersive model
Adsorption model
Band shape dependence on adsorption
Adsorption isotherms
The Langmuir adsorption isotherm
Determination of adsorption data
Frontal analysis
The inverse method
Process optimization of preparative chromatography
Empirical optimization
Numerical optimization
General procedures
Numerical injection-volume optimization
Numerical full optimization
Important operational conditions
Holdup volume
Injection profiles
Modeling additives
Modeling ion-pair reagent additives
Modeling gradient elution
Case example
Acknowledgments
References
Chapter-24---Capillary-electrochromatography_2023_Liquid-Chromatography.pdf
Capillary electrochromatography
Introduction
Principles of capillary electrochromatography
Instrumentation
Injection
Stationary phases
Packed columns: Particle-packed columns
Packed columns: In situ formed monolithic columns
Open-tubular columns
Detection
Mass spectrometry
Miniaturized systems
Applications
References
Chapter-25---Miniaturization-and-microchips_2023_Liquid-Chromatography.pdf
Miniaturization and microchips
Introduction
Compact solvent delivery systems
Aspects of sample injection in miniaturized HPLC
Microchips
Microchip materials and microfabrication technologies
Hard polymers
Soft polymers
Silicon-based materials
Separation channels and beds
Analyte injection
Two-dimensional LC
Microfluidic LC coupled to mass spectrometry
Electrochemical and optical detection in miniaturized and chip-based setups
Electrochemical detection
Fluorescence detection
Absorption detection
Refraction detection
Chip-based HPLC instruments
Portable HPLC
Acknowledgments
References
Chapter-26---Mass-spectrometric-detection--instrumentati_2023_Liquid-Chromat.pdf
Mass spectrometric detection, instrumentation, and ionization methods
Introduction
Ionization techniques
Ionization under vacuum conditions
Electron ionization (EI)
Matrix-assisted laser desorption ionization (MALDI)
Atmospheric pressure ionization (API)
Electrospray ionization (ESI)
Atmospheric pressure chemical ionization APCI
Atmospheric pressure photoionization (APPI)
Ambient ionization mass spectrometry
Mass analyzers
Low-resolution mass analyzers
Quadrupole
Ion trap (IT)
High-resolution mass analyzers
Time-of-flight (TOF)
Fourier transform mass analyzers
Fourier-transformed ion cyclotron resonance (FTICR-MS)
Orbitrap
Tandem mass spectrometry (MS/MS)
Acknowledgments
References
Chapter-27---Identification-and-quantitation-in-liquid-ch_2023_Liquid-Chroma.pdf
Identification and quantitation in liquid chromatography-mass spectrometry
Sample collection
Sample preparation
Calibration and validation
LC-MS analysis
Data analysis and reporting
Matrix effects on quantification
Additional strategies for quantification
Matrix match calibration
Echo peak calibration
Post-column internal standard infusion
One standard per substance class
Metabolic isotopic labelling
Isobaric tags for relative and absolute quantitation (iTRAQ)
References
Chapter-28---Advanced-IR-and-Raman-detectors-for-identif_2023_Liquid-Chromat.pdf
Advanced IR and Raman detectors for identification and quantification
Introduction
Off-line hyphenation
On-line hyphenation
Conclusions
References
Chapter-29---Recent-advances-in-nuclear-magnetic-resonance-spectr_2023_Liqui.pdf
Recent advances in nuclear magnetic resonance spectroscopy detection compatible with on-flow operational regi ...
Introduction
General features of NMR-hyphenated systems
On-flow LC-NMR operation mode
Strategies to improve on-flow LC-NMR limitations
LC-NMR improvements by decoupling LC and NMR
Stop-flow mode
Loop-storage mode
Solid-phase extraction (SPE)
Droplet evaporation (EV)
LC-NMR improvements by using NMR microdetectors
General features of NMR microdetectors
Microsaddle detectors
Solenoid microcoil detectors
Planar spiral microcoil detectors
Stripline detectors
Parallel NMR detection
LC-NMR improvements by using faster NMR experiments
Miscellanea
Other examples
Reaction monitoring
Conclusions
Acknowledgments
References
Chapter-30---Prediction-of-retention-in-liquid-chroma_2023_Liquid-Chromatogr.pdf
Prediction of retention in liquid chromatography
Introduction
Methodology and goals of QSRR studies
Mathematical models
Structural parameters used in QSRR
Retention prediction
3D-QSRR
Applications of QSSR in ``OMICS´´
Application of QSRR in other techniques
3S-Similarity, selectivity, and specificity
Characterization of stationary phases
Quantitative retention biological activity relationships
Conclusions and future perspectives
Acknowledgment
References
Chapter-31---Validation-of-liquid-chromatographic-me_2023_Liquid-Chromatogra.pdf
Validation of liquid chromatographic methods
Discussion
Traditional method validation
Enhanced approaches
The analytical target profile (ATP)
Example ATP
Technique selection and method development
Analytical method risk assessment
Develop understanding and identify operating conditions
Method validation
Method control strategy
Life cycle management
Conclusion
References
Index_2023_Liquid-Chromatography.pdf
Index
Recommend Papers

Liquid Chromatography. Fundamentals and Instrumentation [Volume 1, 3 ed.]
 0323999687

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Liquid Chromatography Fundamentals and Instrumentation

Handbooks in Separation Science

Liquid Chromatography Fundamentals and Instrumentation Volume 1 Third Edition Series Editor Colin F. Poole Edited by Salvatore Fanali Scientific Board of the Ph.D. School in Nanoscience and Advanced Technologies, University of Verona, Verona, Italy

Bezhan Chankvetadze Institute of Physical and Analytical Chemistry, School of Exact and Natural Sciences, Tbilisi State University, Tbilisi, Georgia

Paul R. Haddad Australian Centre for Research on Separation Science, School of Natural Sciences, University of Tasmania, Hobart, TAS, Australia

Colin F. Poole Department of Chemistry, Wayne State University, Detroit, MI, United States

Marja-Liisa Riekkola Department of Chemistry, University of Helsinki, Helsinki, Finland

CHAPTER

Milestones in the development of liquid chromatography

1

Lloyd R. Snyder† and John W. Dolan LC Resources, McMinnville, OR, United States

The importance of liquid chromatography (LC), and especially high-performance LC (HPLC), in today’s world hardly needs stating. It is the most widely used technique for the analysis of chemical mixtures and has contributed in a major way to science (especially the biological sciences) and everyday laboratory practice. LC is primarily a practical technique, so our story is limited to those innovations that contributed significantly to its present use in “working” laboratories. In reflecting on the history of LC, it appears to us that only a few “essential” actors exist in this drama: single individuals whose absence might have delayed the technique by more than a year or two. Thus, the development of the present-day LC has largely been an evolutionary, rather than a revolutionary, process. Furthermore, many important innovations within the past 50 years have occurred within industrial research and development (R&D) groups, where it is often not possible to assign credit to a single person for a final product. Finally, every attempt at history suffers from incomplete and conflicting accounts of who did what and when. In the present discussion, we try to emphasize “what” and “when” rather than “who.”

1.1 Introduction Several important innovations in the history of LC have been reviewed by Ettre [1]: • • • • • •

Invention of chromatography in the early 1900s [2]. Invention of partition and paper chromatography in the early 1940s [3]. Development of ion-exchange chromatography (IEC) [4] and the amino acid analyzer during the 1950s [5]. Invention of gel filtration chromatography in the late 1950s [6,7]. Development of the gel permeation chromatography (GPC) in the early 1960s [8,9]. Development of HPLC in the mid-1960s [8,10–17].



Deceased.

Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00001-1 Copyright # 2023 Elsevier Inc. All rights reserved.

1

2

CHAPTER 1 Milestones in the development of liquid chromatography

The present chapter emphasizes the work on HPLC, while noting major, prior contributions that made this technique possible. Most advances in HPLC can be organized as follows: • •



Development or application of the basic theory combined with empirical observations of the separation process. Invention of new chromatographic modes (e.g., ion-pair chromatography, hydrophilic-interaction chromatography (HILIC)) and the development of HPLC columns for new applications (chiral separation, large biomolecules). Development and improvement of equipment and columns.

1.1.1 Developments before 1960 A good account of the discovery of chromatography by Tswett is given in Refs. [2] and [3,pp. 4–6]. Despite a few subsequent applications of chromatography in other laboratories [3,pp. 7–9], this technique became generally accepted only after its reintroduction in 1931 by Kuhn et al. [4]. The invention of liquid partition chromatography was reported by Martin and Synge in 1941, followed soon after by its extension to paper chromatography in 1944 and the first application of two-dimensional chromatography [5]. Significant work on ion-exchange separation began in the 1930s, with the subsequent development and application of IEC for separation of the rare earths and transuranium elements [6]. The extension of IEC to organic compounds was next, implemented by Cohn and Samuelson [3,pp. 17–21]. By 1958, Moore, Stein, and Spackman reported the automatic analysis of amino acid samples by means of IEC [7]. Their system was a precursor of HPLC that incorporated automatic pumping, efficient IEC columns, and continuous colorimetric detection. This system was later improved and commercialized as the Beckman-Spinco model 120B amino acid analyzer. Yet another major development, in the later 1950s, was the invention of gel filtration [8,9] for the separation of large biomolecules by molecular size; this was followed a few years later by the development of GPC for similar separation of synthetic polymers [10]. The latter then led to the development of a commercial GPC system by Waters Associates (the GPC-100 [11]), which would morph into an early commercial HPLC system (the ALC-100).

1.1.2 HPLC at the beginning Prior to the development of the first HPLC systems, gas chromatography (GC) provided an example of what HPLC might be capable of: automation, speed, and detection sensitivity, as well as the separation of higher boiling and thermally unstable compounds. By the early 1960s, the automation of LC had been demonstrated for amino acid analysis and GPC (Section 1.1.2). By then, it was appreciated that smaller particles in well-packed beds could increase both separation speed and efficiency.

1.2 HPLC theory and practice

Quick separations with small-particle columns also require higher pressures to pump the mobile phase through the column. Detectors that could be used with LC for most samples presented a major challenge at this time and for several years thereafter (Section 1.5). Before 1965, the possibility of using HPLC for separating samples other than amino acids or polymers had undoubtedly occurred to many people. However, the exploitation of this idea required its reduction to practice, followed by the production of commercially available equipment, as in the case of the amino acid analyzer and the GPC (Section 1.1.2). Viewed in these terms, HPLC was first reduced to practice in 1964 in the United States under the direction of Csaba Horva´th [12] and in the Netherlands by Josef Huber (see Ref. [13, pp. 159–166 and 209–217]). The system developed by Horva´th was subsequently the basis for the LCS 1000 Nucleic Acid Analyzer sold by Picker Nuclear (later acquired by Varian), and contributed to the first general-purpose HPLC (Waters ALC-100) [11]. Jack Kirkland had visited Huber’s laboratory in 1964 and subsequently began an HPLC program at DuPont [14–17], which culminated in the Model 820 at about the same time as the ALC100. By 1970, sales of HPLC systems were dominated by Waters Associates and Du Pont. Superficially porous Zipax [17] was initially the most popular column packing. In our opinion, Horva´th, Huber, and Kirkland can be considered the “fathers” of HPLC. Some closely related work at that time by others [18–24] is also relevant to the origin of HPLC. For a description of the columns, equipment, and practice at that time, see Ref. [25].

1.2 HPLC theory and practice Separation as a function of experimental conditions was understood in general terms for GC, and similar principles were expected to apply to HPLC. Resolution, Rs, can be described by the Purnell equation [26] in terms of the plate number, N, separation factor, α, and the retention factor, k:  Rs ¼ 0:25 N 0:5 ½ðα  1Þ=α½k=ðk + 1Þ

(1.1)

A semiquantitative understanding of column efficiency (plate number, N) existed prior to 1965, based on the further development of the van Deemter equation for GC [27] and its extension to LC by Giddings [18]. Later work resulted in the highly useful and widely applied Knox equation [28]: h ¼ Av0:33 + B=v + Cv,

(1.2)

where the reduced plate height, h, is related to the reduced velocity, v, of the mobile phase. The later development of “Poppe” (or “kinetic”) plots further advanced our understanding and use of column efficiency [29]. For further details on Eq. (1.2) and values of N, see Chapter 2 and [30]. When developing an HPLC procedure, the main challenge has been the selection of separation conditions for the control of peak spacing, that is, “optimum” values of α (Section 1.2.2).

3

4

CHAPTER 1 Milestones in the development of liquid chromatography

Basic theory played an important role in the development of HPLC, but its implementation was primarily the result of (a) the introduction of new separation modes or techniques (Section 1.2.1), (b) a better understanding of how best to vary conditions for a satisfactory separation (Section 1.2.2), and (c) improved columns (Section 1.3) and hardware (Sections 1.4 and 1.5).

1.2.1 New HPLC modes and techniques Many of the separation “modes” used today in HPLC were described prior to 1960; for example, reversed-phase chromatography (RPC) was first used by Martin in 1950 [31]. Although a few names are often associated with the rapid development of RPC over the past four decades (e.g., Horva´th, Kirkland, Knox), the present dominance of this technique can be attributed to the efforts of numerous practitioners in both industry and universities, as well as its inherent advantages. A history of the development of gradient elution is provided in Refs. [32–34]; the group of Tiselius is generally given priority for its first implementation and theoretical treatment in the 1950s. A practical understanding of gradient elution has since been facilitated by the linear-solvent-strength model [34]. The technique of ion-pair chromatography became a useful supplement to RPC for the separation of ionizable compounds that were often poorly retained by RPC. Schill and Knox are usually associated with the introduction of this mode [35,36]. Another technique for the separation of more polar compounds, such as sugars (which have poor retention in RPC), was used in the 1970s and later improved by Alpert [37] to become HILIC. Separations of large biomolecules by HPLC required the development of suitable columns, featuring rigid, large-pore particles, and less-hydrophobic stationary phases. Whereas RPC has been used for separating proteins, these large, hydrophilic compounds are more often separated by gel filtration or ion exchange. Beginning around 1975, Chang et al. [38], Kato et al. [39], and others pioneered the development of columns for bio-HPLC. The first such column (SynChropak GPC100) was sold in 1978. Chiral separation was another area that awaited the development of suitable enantiospecific columns (Pirkle, Davankov, Okamoto, Armstrong, and others; see Ref. [30] for details). Nothing more will be discussed about the latter columns for chiral separation, and very little about columns for biochromatography.

1.2.2 Selection of conditions for the control of selectivity •

By the 1960s, it was appreciated that values of α for different pairs of solutes could be affected by the column, the mobile phase, and the temperature. There are many different columns, and the mobile phase may vary according to the concentrations of its constituent solvents, various buffers, and additives, as well as the pH. There are thus a very large number of possible combinations of these separation conditions; only gradually was it learned the conditions most useful for controlling α and the best way to minimize the number of necessary

1.3 Columns

• • •



experiments for a successful separation. As the result of much practical experience and a few systematic studies (e.g., Refs. [40–43]), successful strategies for optimizing selectivity are now available for different samples (Chapters 14 and 15, and Ref. [30]). Some noteworthy developments between 1970 and 2010 include: The introduction of resolution maps (for α or Rs as a function of temperature for IEC [44]). Development of computer-assisted mapping of α in RPC as a function of mobile phase mixtures of acetonitrile, methanol, and tetrahydrofuran [45]. Development of computer-assisted mapping of Rs in either isocratic or gradient RPC as a function of (a) simultaneous change in two or more conditions that affect α and (b) all conditions that affect N (e.g., DryLab [30], Chapter 11). Development of a reliable procedure for characterizing column selectivity [46] and its use in various practical applications [47].

1.3 Columns The development of HPLC depended on new columns, which in turn required new particles, new stationary phases (particle coatings), and improved procedures for packing the column. For details on column developments before 1994, see the review of Majors [48]. Table 1.1 summarizes several of these column innovations, with the dates of their introduction into the marketplace.

1.3.1 Particles and column packing In 1939, Martin noted that (a) small particles in well-packed beds would be needed for increased separation efficiency and (b) such columns would require higher pressures to operate. Prior to 1960, particles for chromatography usually had diameters 100 μm and consisted of either polymeric spheres or irregular silica (prepared by sieving crushed silica). Polymeric materials typically gave lower plate numbers, so that inorganic oxides (mainly silica) became preferred for HPLC columns. HPLC columns at first used coated glass beads (e.g., Pellosil) or beads coated with a porous layer of silica (e.g., Zipax). The thickness of the stationary phase was only a fraction of the particle diameter, thereby reducing the diffusion distance within the stationary phase and yielding higher values of N, but with decreased loadability. Smaller, fully porous particles with a narrow size range were expected to provide more efficient columns (as well as superior loadability), but such particles could not be produced by sieving. The introduction of air classification for particle sizing overcame this difficulty, and around 1970, Merck offered irregular silica in diameters of 5 and 10 μm. Whereas larger particles were easily packed by tapping the column until the bed settled, packing particles of 10 μm diameter (or smaller) required a different approach. The first published procedure for reproducibly packing HPLC columns with particle diameters 10 μm used a balanced-density approach [49], similar to

5

6

CHAPTER 1 Milestones in the development of liquid chromatography

Table 1.1 Some highlights in commercial HPLC column development. Datea

Column

Description

Company

1967 1969 1971 1972 1972 1973 1978

Pellosil Zipax MicroPak Zorbax Permaphase μBondapak C18 SynChropak GPC100 Rx-silica StableBond Hypercarb ZirChrom PBD SilicaRod XTerra Rapid resolution Acquity Halo

Pellicular ion exchange (40 μm) Porous layer silica (40 μm) Irregular porous silica (5–10 μm) Spherical porous silica (7 μm) Silane phase (7 μm) Silane phase (10 μm) Gel filtration column

Northgate DuPont Varian DuPont DuPont Waters SynChrom

Type-B silica Stable bonded phases Porous graphitic carbon Zirconia particles Monolith Hybrid particles (3–5 μm) Porous silica (1.8 μm) Porous silica (1.7 μm) Superficially porous particles (2.7 μm)

DuPont DuPont Hypersil Keystoneb Merck Waters Agilent Waters AMTc

1988 1989 1994 1996 2000 1999 2003 2004 2007 a

Date of commercial introduction. Produced by Zirchrom, distributed by Keystone Scientific. Advanced Materials Technology. Adapted from Majors RE. Twenty-five years of HPLC column development—a commercial perspective. LCGC N Am 1994;12:508. b c

that used earlier for GPC columns; the resulting MicroPak columns were offered for sale by Varian in 1971. Over the years, many procedures have been described for packing particles with diameters 10 μm [30]; in practice, each new particle requires customized conditions for best results. About the same time, spherical particles with 7 μm diameters were produced by DuPont and sold under the name Zorbax. The latter particles were prepared by the aggregation of colloidal particles, much as popcorn balls are assembled from individual kernels of popcorn. Other procedures have been developed for the manufacture of porous, spherical particles [50]. Today, most analytical columns use spherical particles. The silica used for HPLC columns before 1988 was usually derived from natural sources (water-glass solution) contaminated by various metals (e.g., iron, aluminum) that can increase its acidity. This often results in tailing peaks for basic compounds, poor column reproducibility, and other problems. These difficulties were largely overcome by preparing silica particles from the hydrolysis and polycondensation of pure tetraethoxysilane (TEOS). The latter, less-acidic silica, is now referred to as type-B silica, in contrast to the older, less-pure, and more-acidic type-A silica. The first widely used columns based on type-B silica were introduced by DuPont in 1988, and today most analytical columns use type-B silica.

1.3 Columns

One disadvantage of bonded silica is that it is unstable at both low and high pH. Particles of porous, spherical, graphitic carbon [51] provided one answer to this problem; the resulting Hypercarb columns were offered in 1994 by Hypersil. Another approach to pH-stable particles is the use of porous, spherical zirconia instead of silica [52]. The first widely distributed column of this type (ZirChrom PBD) was introduced in 1996. Similar, but less dramatic improvements in column stability can be achieved by changes in bonding chemistry (Section 1.3.2). A quite different approach to HPLC columns was the development of the socalled monoliths. Monolithic columns are cast as a porous, rigid cylinder by in situ polymerization and can be made from either silica or polymer. Several groups have contributed to their development, as reviewed in Ref. [53]. One advantage of monolithic columns is their greater permeability, allowing separations at lower pressures. Although the first monolithic column (SilicaRod, Merck) was introduced for sale in 2000, the use of these columns (as of 2016) has remained somewhat limited. Hybrid particles are formed of an organic/inorganic structure, based on the reaction of TEOS and an alkyl triethoxysilane [54]. The resulting particles are more stable than their silica counterparts, with reduced acidity. Hybrid columns are therefore well suited for the separation of basic samples. The first hybrid column (XTerra, Waters) was offered for sale in 1999. From 1970 to 1990, the preferred particle size gradually evolved from 10 μm to about 3 μm with a resulting improvement in separation speed. The use of very small particles requires an increase in the maximum operating pressure of the equipment (>6000 psi). Equipment for separations at much higher pressures was first described by Rogers et al. [55], later perfected by MacNair et al. [56], and eventually commercialized by Waters (ultra-HPLC or UPLC) in 2004. This enabled the use of 1.8-μm (Rapid Resolution, Agilent, 2003) and 1.7-μm fully porous particles (Acquity, Waters, 2004). A little later, the so-called core-shell (or superficially porous) particles (Halo, AMT, 2007) were introduced for use at pressures 6000 psi, but with similar performance as for smaller-particle columns and ultrahigh-pressure LC (UHPLC) systems. The latter, superficially porous particles, consist of a nonporous core coated with a 0.5-μm layer of porous silica (total diameter of 2.7 μm). Smallerdiameter superficially porous particles (SPP) are also available today. Surfacecoated particles of various kinds have played an important role in HPLC since the beginning; see the review of [57].

1.3.2 Stationary phases and selectivity Three stationary phases were used with surface-coated particles at the beginning of HPLC: an attached polymeric layer for ion exchange, a mechanically held liquid (liquid-liquid partition), and bare silica (adsorption). Although liquid-liquid partition was initially the most popular one, the mechanically held liquid proved unstable and operationally inconvenient. Various attempts were made to permanently bond an organic layer to a silica particle, with eventual success using a silicone polymer

7

8

CHAPTER 1 Milestones in the development of liquid chromatography

as the stationary phase (marketed in 1972 as Permaphase; DuPont). Subsequently, organosilanes were used as reactants; for example, Cl  SiðCH3 Þ2 C18 + ≡ SiOH , ≡ SiO  SiðCH3 Þ2 C18

(1.3)

(e.g., μBondapak, Waters, 1973). The latter approach has since been preferred for the preparation of RPC and other columns. One shortcoming of the original silane phases is their instability at both low and high pH levels, which can limit their application. The development of sterically protected stationary phases provided stability at a low pH [58]; the first column of this kind (StableBond, DuPont) was introduced in 1989. These columns use hindered silanes, where the methyl groups in the silane of Eq. (1.3) are replaced by bulky groups such as i-butyl. Many other silanes have found use for HPLC stationary phases [30], for various purposes. The C18 group used initially has been followed by other ligands to yield phenyl, cyano, embedded polar group, and other columns. These column types can result in large differences in selectivity (Section 1.2.2), as well as provide other advantages (e.g., avoidance of stationary-phase dewetting with highly aqueous mobile phases). For a review of column packings, including both the particle and stationary phases, see Refs. [50, 59].

1.4 Equipment Many of the innovations in HPLC equipment remain as proprietary company secrets or are buried in the patent literature. Most of the dates listed here are estimates based on personal knowledge, interviews, and (limited) patent information. Additional information on instrumentation prior to 1980 can be found in Ref. [60]. Early HPLC systems contained most of the same components used today, but with less-sophisticated execution. A reservoir, pump, injector, column, detector, and data collection system were required. Reservoirs most commonly consisted of laboratory glassware. Microfiltration of the mobile phase (solvents) was used. Solvent outgassing was not a big problem until on-line mixing and gradient elution became more important. At first, solvents were degassed prior to use by means of vacuum or heating, followed in the late 1970s by the more reliable helium sparging (Spectra-Physics [61]), and later in-line vacuum degassing. Although complete HPLC systems were available from some vendors, many early workers built their own HPLC systems or purchased components and assembled HPLC systems from the “best” available modules. In the late 1960s and early 1970s, the Milton-Roy Mini-Pump formed the core of most home-built systems, as well as some commercial units. This was a single piston pump with the stroke length adjusted to control the flow rate. It used a mechanical pressure gauge, whose Bourdon tube also acted as a pulse damper. Additional pulse dampening might be added with gas ballasts or additional Bourdon tubes. The Waters M-6000 (dual-reciprocating-piston) pump was introduced in 1972 [59,62] that rapidly became the gold standard. With two variable-speed, reciprocating pistons operating out of phase,

1.4 Equipment

pulsations and system volume were reduced, with piston volumes in the 100-μL range. Nester-Faust (later acquired by Perkin Elmer) and Varian took a different approach, with large (250–1000 mL) pistons, one for each solvent, driven by a screw drive. Depending on the solvent volume used per run, the pistons might need to be refilled after each run. Syringe pumps dropped from the market soon after a publication pointed out serious limitations in their use [63]: mobile phase compressibility could result in poor reproducibility of retention times, especially in gradient elution. About 1976, Altex (later purchased by Beckman) introduced the Model 110A pump with a fast-refill feature [64]; its single piston could spend more than 50% of its duty cycle delivering solvent, thus reducing pump pulsations. A year or two earlier, Altex had introduced the Model 100 pump for operation at 10,000 psi, far exceeding the 6000 psi upper limits of other systems and foreshadowing today’s UHPLC systems. Another innovation in the late 1970s was Perkin Elmer’s tandem-piston or accumulator-piston pump, where the first piston feeds solvent to the second piston. This design reduced both pulsations and the number of required check valves from four to three (or, in some cases, two) and improved reliability— this feature is still common in many of today’s pumps. The earliest gradients were generated with a single pump; the weak solvent A was placed in a beaker on a stir plate, and the strong solvent B was added to the beaker (by siphoning) as its contents were delivered to the pump. More reliable, on-line mixing was subsequently accomplished by controlling the relative flow rate of two solvents, each pumped separately by a dedicated pump to a high-pressure mixer. In the late 1970s, Spectra-Physics introduced a three-solvent, low-pressure mixing system [65] coupled with its patented helium degasser [61]. This was the first practical low-pressure mixing system that allowed the simultaneous use of up to three solvents. At about the same time, an alternative multisolvent design was featured in Varian’s 5000 pumping system [66]; this design introduced each of the three solvents directly into the pump head, minimizing bubble problems. These pumps also feature active inlet check valves, an innovation that eliminated most problems associated with these valves. As patents expired and licensing agreements were reached, the best of these features—low-pressure mixing, accumulator-piston pumps, and active check valves—became standard features on pumps from many manufacturers. Sample introduction was a challenge for early HPLC users. For lower-pressure operation, a septum-type injector could be used, but these tended to leak and were difficult to use; stop-flow injection represented an alternative. The Waters U6K [67] was introduced in 1973, and allowed convenient and reliable injection into the flowing stream. The U6K also included an innovative flow-bypass channel that reduced the pressure shock when the valve was cycled, a common source of column collapse with the less-robust columns in use at that time. At about the same time, Valco introduced six-port rotary injectors, a design that is now the industry standard. Rheodyne was formed soon after, and its injection valve [68] eventually became the industry leader. An automated valve inevitably led to an autosampler. One of the first widely used autosamplers, the Micromeritics model [69], used sealed vial cap that acted as a syringe plunger to deliver the sample. A needle was inserted through the cap, and the

9

10

CHAPTER 1 Milestones in the development of liquid chromatography

cap was depressed so as to displace sample into the needle and sample loop. Several companies later developed autosamplers that used a motorized syringe to draw samples from a vial into the loop of a fixed-loop injector. This and a needle-in-loop design [70] are the most common autosampler configurations today. Although some of the early instruments (e.g., DuPont) included column ovens, many users did not consider column ovens necessary; others used hot water baths, heat tapes, or retired GC ovens to control the column temperature. Today’s ovens are designed specifically for HPLC, with either a resistance or Peltier heater both to maintain column temperature and preheat the mobile phase. Data were first collected on strip-chart recorders; every user had a favorite technique to keep the paper from jamming or pens from clogging, either of which could mean loss of data. Manual quantification was the rule, using either peak height or area—the latter by triangulation, planimetry, or cutting out the peak from the paper and weighing it. Disk integrators were an innovation of the 1960s for GC and later provided some automation to the LC data collection process. Spectra-Physics and Hewlett-Packard pioneered digital integrators in the mid-1970s, which revolutionized HPLC data collection. In the late 1970s, Nelson Analytical introduced software that could integrate peaks automatically, even with drifting baselines; this software became part of data systems sold by many manufacturers. With the introduction of the IBM Personal Computer in 1981, the death knell rang for stand-alone integrators. Peak integration and system control gradually migrated to this now universally accepted computing platform. A recent advance in instrumentation is the development of UHPLC systems that exceed the traditional 6000 psi/400 bar pressure barrier; some systems (e.g., Shimadzu’s Nexera) offer pressures up to 19,000 psi/1300 bar. Waters introduced the UHPLC system in 2004, and many other manufacturers now supply competitive instrumentation. For optimal performance, these systems use sub-2-μm particles in short, narrow-diameter columns, which generate peaks of very small volume. Chromatographic band broadening due to noncolumn sources (especially injection, plumbing, detector configuration, and data processing) then became critical. The small volume design of UHPLC systems (and many newer HPLC systems) reduce, but do not eliminate, the band broadening problems that restricted the acceptance of 1.0 > 1.0 > 1.0 0.35 > 1.0 > 1.0 > 1.0 > 1.0 > 1.0 > 1.0

Presumed similar to ethyl ether > 1.0 > 1.0 > 1.0 0.20 0.53 > 1.0 0.70 0.65 0.44 > 1.0

> 1.0 > 1.0 0.46 0.07 0.23 > 1.0 0.54 0.35 0.20 0.60

> 1.0 > 1.0 0.27 0.03 0.10 > 1.0 0.45 0.15 0.11 0.40

> 1.0 > 1.0 0.18 0.02 0.04 0.09 0.28 0.07 0.05 0.21

0.25 > 1.0 0.10 0.01 0.02 0.00 0.10 0.03 0.03 0.18

0.08 0.10 0.05 0.00 0.01 0.00 0.05 0.01 0.02 0.09

a

Values for the pure solvent [2], derived as described in Ref. [1]. Immiscible with hexane. c Nonbasic localizing. d Nonlocalizing. e Basic localizing. f Easily oxidized and therefore less useful in practice. g Very strong (localizing), proton-donor solvent; classification as “basic” or “nonbasic” may not be relevant. b

commonly n-hexane). Several solvents that can be used for NPC are listed in Table 5.1, with some useful properties.

5.3.3 Example of method development An assay was required for samples containing the polar drug paclitaxel in a mixture with a more hydrophobic polymer (poly[sebacic-recinoleic ester-anhydride]) [13]. Because the drug is more polar than the polymer, an assay by RPC would have required either prior separation of the polymer from the drug (because of very late elution of the polymer in RPC) or gradient elution. For this reason, NPC separation was explored, so that the polymer would leave the column before the drug, thereby precluding a need for either sample pretreatment or gradient elution. Initial studies were carried out by means of TLC with silica plates. The use of 100% methylene chloride yielded RF ¼ 0.00 for the drug, so the stronger B solvent methanol (MeOH, ε0 ¼ 0.70) was investigated, in a mixture with methylene chloride. Mobile phases composed of 2–5% MeOH-CH2Cl2 appeared promising, and 1.5% MeOH-CH2Cl2 provided the acceptable HPLC separation of Figure 5.6A, with UV detection at 240 nm.

5.4 Problems with the use of normal-phase chromatography

Paclitaxel

Fresh sample Polymer

(A) Degraded polymer

(B) Degraded sample

0

(C)

2

4

6

8

10

Time (min)

FIGURE 5.6 NPC assay of paclitaxel in the presence of a polymeric additive. Conditions: 250  4.0-mm silica column (5-μm particles); 1.5% methanol-methylene chloride; 25°C; 1 mL/min. (A) fresh sample; (B) degraded sample of polymer (stored at pH 7.4 and 37°C for 60 days); (C) degraded sample of paclitaxel plus polymer. Adapted from Ref. [13].

The method of Figure 5.6A was also intended for use with thermally stressed samples, so the experiments of Figure 5.6B and C were carried out. It appears that thermal degradation of the polymer, Figure 5.6B, does not result in the formation of peaks that overlap the paclitaxel peak and thereby compromise its quantification. For other details of this NPC method development, see Ref. [13].

5.4 Problems with the use of normal-phase chromatography Various problems can occur when carrying out NPC separation with silica as column packing: • • •

Poor separation reproducibility (including extreme sensitivity to the mobile phase water content); Solvent demixing; and Slow column equilibration when changing the mobile phase.

85

86

CHAPTER 5 Liquid–solid chromatography

Water is a very strong solvent for NPC, and traces of water in the mobile phase can markedly affect sample retention, especially for weaker mobile phases (e.g., ε < 0.3). Therefore, as ambient humidity increases, the concentration of water in a solvent can increase, leading to a decrease in sample retention. This problem can be minimized by maintaining a constant water concentration in the mobile phase. A mobile phase with fixed water content can be prepared by blending specified volumes of the mobile phase that is either water-free or water-saturated. For example, mixing equal volumes of the two mobile phases result in 50% water saturation. For further details, see Refs. [1,3]. Solvent demixing refers to the preferential retention of the B solvent with a higher polarity and its removal from the mobile phase. This process is more important for TLC, where its effects can be minimized by a preequilibration of the plate in a mobile phase-saturated atmosphere, prior to adding the sample. While demixing can also occur in HPLC operation, the passage of a suitable volume of mobile phase through the column eliminates this problem. Column equilibration after a change of the B solvent may require a longer time than that in RPC, so it is advisable to confirm that retention does not change with time by repeated injections of the sample. These problems are usually of minor importance and should not discourage the use of NPC for the right application. For further details on NPC, see Ref. [4].

References [1] Snyder LR. Principles of adsorption chromatography: the separation of nonionic organic compounds. New York, NY: Marcel Dekker; 1968. [2] Snyder LR. In: Horva´th C, editor. High-performance liquid chromatography: advances and perspectives, vol. 3. New York, NY: Academic Press; 1983. p. 157. [3] Snyder LR. Solvent selectivity in normal-phase TLC. J Planar Chromatogr 2008; 21:315. [4] Snyder LR, Kirkland JJ, Dolan JW. Introduction to modern liquid chromatography. 3rd ed. New York, NY: Wiley-Interscience; 2010 [chapter 8]. [5] Truedsson L-A, Smith BEF. Study of retention behaviour of primary, secondary and tertiary anilines in normal- and reversed-phase liquid chromatography. J Chromatogr 1981;214:291. [6] Karger BL, Snyder LR, Horva´th C. An introduction to separation science. New York, NY: Wiley-Interscience; 1973 [chapter 19]. [7] Snyder LR, Poppe H. The mechanism of solute retention in liquid-solid chromatography and the role of the mobile phase in affecting separation. Competition vs ‘Sorption’. J Chromatogr 1980;184:363. [8] Meyer VR, Palamareva MD. New graph of binary mixture solvent strength in adsorption liquid chromatography. J Chromatogr 1993;641:391. [9] Soczewinski E. Solvent composition effects in thin-layer chromatography systems of the type silica gel-electron donor solvent. Anal Chem 1969;41:179.

References

[10] Soczewinski E, Dzido T, Gołkiewicz W. Comparison of high-performance liquid chromatographic and thin-layer chromatographic data obtained with various types of silica. J Chromatogr 1977;131:408. [11] Glajch JL, Kirkland JJ, Snyder LR. Practical optimization of solvent selectivity in liquidsolid chromatography using a mixture-design statistical technique. J Chromatogr 1982;238:269. [12] Fried B, Sherma J. Thin layer chromatography. 4th ed. New York, NY: Marcel Dekker; 1999. [13] Vaisman B, Shikanov A, Domb AJ. Normal phase high performance liquid chromatography for determination of paclitaxel incorporated in a lipophilic polymer matrix. J Chromatogr A 2005;1064:85. [14] Solvent guide. Muskegon, MI: Burdick & Jackson Labs; 1980.

87

CHAPTER

Reversed-phase liquid chromatography

6 Colin F. Poole

Department of Chemistry, Wayne State University, Detroit, MI, United States

6.1 Introduction Reversed-phase liquid chromatography (RPLC) is by far the most popular liquid chromatographic technique based on annual citations, easily eclipsing all other types of liquid chromatography. It could rightly claim to be the normal mode of liquid chromatography from the perspective of contemporary use. From a historical context, liquid–solid chromatography (LSC) was well established by the 1970s and was considered the normal mode of liquid chromatography before RPLC existed and is still sometimes referred to as normal-phase chromatography. To draw attention to the new mode of liquid chromatography and out of deference to LSC, it became known as reversed-phase liquid chromatography reflecting the fact that LSC typically employed inorganic oxide stationary phases, designated as a polar stationary phase, and mixtures of organic solvents as the mobile phase, designated as a low-polarity phase, while RPLC typically used chemically bonded inorganic oxide stationary phases, designated a low-polarity stationary phase, and aqueous-organic solvent mixtures as a mobile phase, designated as a polar phase. This reversal in phase polarity was incorporated in naming the new technique supported by the observation that for varied compounds, those compounds considered polar tend to be located at the end of the chromatogram in LSC and at the beginning for RPLC, a general reversal in elution order. The prominence of RPLC is a result of its exceptional range of applications spanning a wide range of molecular size, polarity, and ionicity; its versatility; and the availability of relatively straightforward albeit semi-empirical approaches to method development [1]. Simultaneous separation of neutral and ionic compounds is possible and the rapid equilibrium of the stationary phase with changes in mobile phase composition facilitates applications requiring gradient elution. The inclusion of additives in the mobile phase allows the exploitation of secondary chemical equilibria to control retention and selectivity by ion suppression, ion-pair formation, micellar distribution, and various complexation mechanisms. To allow a general focus on the Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00015-1 Copyright # 2023 Elsevier Inc. All rights reserved.

89

90

CHAPTER 6 Reversed-phase liquid chromatography

broader aspects of RPLC, secondary chemical equilibria are not discussed further. This topic is deferred to a companion contribution (Chapter 7) to facilitate a thorough discussion.

6.2 Parameters that affect retention Often a broad definition of RPLC is cited that includes all separations that employ a polar mobile phase and a less polar stationary phase. This statement includes, but fails to emphasize, the key feature of separations by RPLC, which is that the mobile phase is always an aqueous solution in which the volume fraction of water is sufficient to establish control over the retention mechanism. While it would be incorrect to describe water as a unique solvent, compared with typical organic solvents, it certainly qualifies as an extreme solvent, on account of its relatively high cohesive energy and hydrogen-bond acidity, and to a lesser extent, dipolarity. These properties explain the general features of separation by RPLC that include a general increase in retention with compound size, and for compounds of similar size, lower retention for polar compounds capable of hydrogen-bonding and dipole-type interactions with water. Water is a weak solvent at moderate temperatures, and to achieve elution of most compounds in RPLC, it is typically diluted with a miscible organic solvent. The organic solvent serves to increase the solvent strength of the mobile phase and adjust selectivity by its capability to reduce the cohesive energy of the mobile phase and through its participation in additional solute-solvent interactions that are different in intensity from those for water. Thus, RPLC is usually performed using binary or higher-order mobile phase compositions in which the strength-adjusting solvent is water and the strong solvent is one or more water-miscible organic solvents, such as methanol, acetonitrile, or tetrahydrofuran.

6.2.1 System properties The retention mechanism in RPLC is difficult to unravel because of the heterogeneous nature of the interphase region (active stationary phase) and its variation with mobile phase composition; the microscopic heterogeneity of water-organic solvent mixtures (mobile phase); the difficulty of defining thermodynamically meaningful phase boundaries (unknown phase ratio); and a general uncertainty concerning the mechanism of the distribution process [2]. In fact, whether retention occurs by partitioning between phases or adsorption at their interface has been an open question since the beginning of RPLC [3–5]. A partition mechanism implies that the solute is transferred to the stationary phase with full insertion into the stationary phase volume. The adsorption mechanism views transfer as a competitive displacement process occurring at an interface between the mobile and stationary phases, which occurs with a displacement of an equal volume of mobile phase molecules to accommodate the solute at the surface of the interface. The two models represent extreme possibilities. Real separations are unlikely so well defined for varied

6.2 Parameters that affect retention

compounds. A mixed retention mechanism in which adsorption and partition simultaneously affect retention is the normal situation. Nonlinear chromatography and molecular dynamic simulations have shed some light on this problem where previous studies employing linear chromatography could not. The heterogeneity of the retention mechanism can only be characterized by measurements at concentrations approaching the saturation of different sorption sites. This requires high sample concentrations to saturate high-energy sites so that the contribution from low-energy sites is observable. Evaluation of different sites is based on isotherm measurements according to techniques developed for preparative liquid chromatography. For octadecylsiloxane–bonded silica phases, up to four types of sorption sites were identified, confirming the heterogeneous nature of the solvated stationary phase [6,7]. The most abundant are the low-energy sites likely located at the interface between the mobile and bonded-phase layer. The other higher-energy sites are located deeper inside the bonded-phase layer, providing a larger contact area with the analyte or more specific interactions with the silica surface. Sites at the interface between the mobile and stationary phases are associated with the classical adsorption mechanism while those sites buried deeper in the solvated bonded-phase layer are associated with partitioning. Because low values of the saturation capacity of the high-energy sites are associated with relatively large values of the equilibrium constant, the contributions to the observed retention factor from the different sites tend to be comparable. Molecular dynamic simulations support the inference that the stationary phase is heterogeneous with multiple preferred regions for solute interactions [8–10]. In addition to solute properties (size and polarity), the retention mechanism depends on the topology of the stationary phase (the chain length, surface coverage of the bonded phase, and pore shape) and mobile phase composition. For octadecylsiloxanebonded silica stationary phases, both adsorption and partition are important for small molecules of low polarity, whereas adsorption is more important for polar compounds with hydrogen-bonding capabilities. For short-chain bonded phases (< octylsiloxane), the retention mechanism is best described as adsorption for both polar and nonpolar analytes. The retention mechanism is, however, strongly dependent on the chemical nature of the solute and is subject to steric factors. Interestingly, adsorption is observed to occur at locations above or just below the mobile and stationary phase interface for different solutes, somewhat different from the classic model for the adsorption mechanism. For compounds with hydrogen-bonding functional groups, specific interactions with accessible silanol groups or water molecules bound to the silica substrate also contribute to the general retention mechanism. Large, flexible molecules with an extended form greater than the bonded alkylsiloxane chain length are retained by embedding part of the solute in the stationary phase with the remainder adsorbed at the stationary phase interface. Perhaps less obvious is that at the microscopic level the structure of water-organic solvent mixtures is also heterogeneous [11,12]. Binary water-organic solvent mixtures are composed of clusters consisting of associated water molecules, waterorganic solvent aggregates with up to several aggregation numbers, and associated

91

CHAPTER 6 Reversed-phase liquid chromatography

organic solvent molecules with a relative concentration determined by the identity and volume fraction of the organic solvent. In RPLC, it is assumed that solutes distribute themselves between the solvated stationary phase and each type of solvent cluster characterized by separate distribution constants. In this case, the observed retention factor is an average of all the individual distribution constants and cannot be adequately characterized by the bulk composition of the mobile phase.

6.2.2 Surface excess adsorption Surface excess adsorption describes the accumulation of one or more components of the mobile phase near an adsorbent surface under the influence of surface forces [7,13–17]. At equilibrium, a certain amount of the solvent with a higher affinity for the surface will be accumulated on the surface in excess of its equilibrium concentration in the bulk solution, as illustrated in Figure 6.1. This accumulation of organic solvent, typically, is assumed to occur at the tips of the bonded alkylsiloxane chains forming a continuous film at the adsorbent surface. The measurement of excess adsorption does not require adherence to any specific adsorption model, but the interpretation of adsorption isotherms is only possible by predefining a model and introducing an interface (Gibbs dividing plane). The calculation of the adsorbed solvent layer thickness requires an estimate of the surface area of the interface, which for chemically bonded phases is not amenable to measurement. The general interpretation of excess adsorption isotherms infers the formation of a multilayer of neat organic solvent perhaps 4–5 molecules deep for acetonitrile and tetrahydrofuran, while monolayer coverage is predicted for water–methanol mobile phases, in both cases, reasonably independent of the chain length of the bonded phase. This, rather classical interpretation of the surface excess adsorption of organic solvent, is the 8

Excess adsorption

92

Acetonitrile

6

4 Methanol 2

0

0

0.2

0.4 0.6 Organic solvent (%v/v)

0.8

1

FIGURE 6.1 Characteristic plot of excess adsorption (μmol/m2) of methanol and acetonitrile from a binary water-organic solvent mobile phase on an octadecylsiloxane-bonded silica stationary phase.

6.2 Parameters that affect retention

basis of the partition-displacement retention model for RPLC in which the distribution of compounds between the mobile phase and adsorbed organic solvent layer precedes their adsorption on the surface of the bonded phase [18,19]. The assumption underpinning the partition-displacement model is not supported by molecular dynamic simulations [8]. An increase in the amount of organic solvent in the interface region is observed, but the solvent composition in this region is never only organic solvent. Solvated alkylsiloxane-bonded silica stationary phases represent a heterogeneous medium with multiple preferred regions for solute interactions, which are not uniquely restricted to the interfacial region of the stationary phase.

6.2.3 Interphase model The interphase model attempts to provide a realistic picture of the active portion of the stationary phase responsible for retention in RPLC but is not associated with any specific mathematical model for the retention mechanism [2,20]. The interphase is a diffuse and possibly microscopically thick region separating two dissimilar phases in which one phase typically has limited solubility in the other. For chemically bonded phases, a major function of the bonded organic chains is to organize and assist in stabilizing an interphase region in equilibrium with a bulk-solvating mobile phase. This region is bound on one side by the silica gel substrate, which is impenetrable to solvent. On the other side, it is bounded by an imaginary dividing plane separating the interphase region from the bulk mobile phase, which has a constant composition no longer influenced by surface forces. The structure of the interphase region is heterogeneous housing several area-constrained and mobile components. For chemically bonded phases, the interphase region consists of the unreacted silica surface, the organic ligands anchored at one end to the silica surface, and components adsorbed from the mobile phase attracted into the interphase region by solvation of the bonded-phase ligands and the silica substrate. The thickness of the interphase region is changeable and depends on the mobile phase composition, the identity and surface coverage of bonded ligands, properties of the silica substrate, whether endcapped or not, and the average column temperature and pressure. The bonded ligands are expected to adopt a vertical configuration roughly perpendicular to the surface independent of the mobile phase composition. The penetration of organic solvent into the chain region increases with the volume fraction of organic solvent in the mobile phase. The concentration of water in the bonded phase region is expected to be low with water molecules largely hydrogen bonded to silanols located near the silica surface. Chemically bonded phases with polar-embedded functional groups attract and bind additional water molecules into the interphase region inhibiting solute interactions with surface silanols located below the water layer [21]. For steric reasons, it is only possible for about half of the silanol groups on a silica surface to react with typical reagents used to prepare monomeric-bonded phases [1]. The unreacted silanols are largely inaccessible to solute and solvent molecules and presumably play little part in the properties of the interphase region. However, a small number of these silanols act as high-energy adsorption sites that interact with

93

94

CHAPTER 6 Reversed-phase liquid chromatography

both polar solvent and solute molecules, particularly those capable of hydrogen bonding. Bonded phases based on type A silica generally have a higher number of these acidic silanols and exhibit strong retention of bases sometimes accompanied by peak tailing. It is speculated that these acidic silanol groups are associated with metal impurities close to the surface in type A silica. Bonded phases based on highpurity type B silica generally have a lower concentration of acidic silanols and exhibit weaker electrostatic interactions with bases, particularly for stationary phases that are exhaustively endcapped. Progress in minimizing peak tailing of bases, originally thought to be exclusively due to interactions with silanols, has been impressive but not eliminated entirely. A plausible explanation is that the tailing results from the presence of a low concentration of high-energy adsorption sites buried deep in the bonded-phase region, which are not necessarily associated with silanols [22,23]. Also, it has been speculated that silanols inaccessible to solutes could form water wires (a hydro-linked proton conduit) that facilitates the transfer of a proton from the silica surface to solutes residing in the bonded-phase region [24]. Conceptual models of solvated macroporous polymer and porous graphitic carbon stationary phases are not as well developed as those for alkylsiloxane-bonded silica stationary phases. Macroporous polymer particles are prepared by the agglomeration of extensively fused microspheres, which are themselves microporous. Retention is presumed to occur through solute interactions with the solvated surface of the polymer matrix lining the macropores as well as with solvent imbibed by the micropores of the solvated matrix [25,26]. For porous graphitic carbon, an adsorption mechanism is generally assumed with solvent effects controlling retention (a general reversed-phase mechanism) supplemented by additional dipole-type interactions at the electronically polarizable graphite surface and steric contributions on account of the planar topology of graphite [27]. Retention on porous graphitic carbon is only weakly correlated with that of octadecylsiloxane-bonded silica stationary phases.

6.2.4 Formal mechanistic retention models Major difficulties in postulating retention models on thermodynamic principles arise from the mathematical complexity of accounting for the heterogeneous properties of the interphase region, uncertainty in assigning the phase ratio, and accounting for the composite nature of retention factors, which simultaneously represent the contributions from several distribution processes. The solvophobic theory attempts to explain retention as a thermodynamic cycle in which the binding of the solute to the bonded chains of the stationary phase occurs in the gas phase prior to the transfer of the participating species into the mobile phase [1–3,28]. The free energy change for this process is the difference between the free energy required to transfer the ligandsolute complex from the gas phase into the solution and the free energy change for transferring the individual components into the solution. Stationary phase effects are attributed to solute-ligand interactions in the gas phase. Thus, the basic solvophobic theory is unable to account for the interaction of solute molecules with solvent adsorbed by the stationary phase or specific interactions with silanols. Lattice

6.2 Parameters that affect retention

theories apply statistical thermodynamic considerations to the partition or interfacial adsorption mechanisms for solute transfer between the mobile phase and solvated ligands of the stationary phase [29–31]. The driving force for retention is provided by the difference in chemical potential for the contacts of the solute with surrounding neighbors in the stationary and mobile phases and the partial ordering of the bondedphase ligands, which leads to an entropic expulsion of solute molecules relative to the properties of an isotropic liquid. Restrictions imposed by assumptions of regular solution theory (random mixing) on long-range intermolecular interactions to the nearest neighbor molecules were minimized by a quasi-chemical approach to allow for preferential solvation of solute in the solvated stationary phase. Exact thermodynamic models as described above have largely failed because of oversimplification of the retention mechanism as well as their internal mathematical complexity. The general simplicity of the variation of experimental retention factors with mobile phase composition means that even primitive retention models may describe this relationship adequately [5,32,33]. Consequently, several less-formal models that offer an overview of the retention process without being able to explain all features at a molecular level have been developed (Section 6.2.5) and widely adopted for method development purposes.

6.2.5 Semi-empirical retention models 6.2.5.1 Solvent strength Separations in RPLC can be performed by isocratic or gradient elution modes. Semiempirical retention models with mobile phase composition as a variable attempt to describe the dependence of isocratic retention factors on the composition of the mobile phase by a simple mathematical relationship are broadly based on an idealized partition [5,34–38], adsorption [5,19,39], or partition-displacement [18,19,32,40,41] model. Typically, the same models are utilized for gradient elution separations in an integral form to accommodate changes in the solvent strength with time on the apparent retention factors [5,34,35,38]. The most widely used model is the simplest, linear-solvent-strength model, Eq. (6.1), strictly applicable to intermediate binary mobile phase compositions [5,34–38]. ln k ¼ ln kw  Sφ

(6.1)

where k is the retention factor, kw the solute retention factor with water as the mobile phase, and S the slope of the fit of a linear model to the experimental data. With the assumption that the composition of the stationary phase and the phase ratio is independent of the mobile phase composition, then the S-value characterizes the free energy of transfer of the solute from water to the organic solvent. It is presumed to be independent of the identity of the stationary phase and a suitable parameter as a measure of the solvent strength of the organic solvent. S-values for common water-miscible organic solvents are summarized in Table 6.1 [1,10]. Binary solvent mixtures allow only limited possibilities for the simultaneous optimization of solvent strength and selectivity. Higher-order solvent mixtures are commonly employed for

95

96

CHAPTER 6 Reversed-phase liquid chromatography

Table 6.1 Approximate S-values for the solvent strength of water-miscible organic solvents (The S-value for water by definition ¼ 0). Solvent

S-value

Group characteristics

Methanol

3.0

Amphoteric solvents (moderately dipolar with strong hydrogen-bonding properties)

Ethanol 2-Propanol Acetonitrile

3.6 4.2 3.1

Acetone Dioxane Tetrahydrofuran

3.4 3.5 4.4

Dipolar solvent, strong hydrogen-bond base and weak hydrogen-bond acid Dipolar solvent and strong hydrogen-bond base No hydrogen-bond acidity

isocratic separations. The solvent strength for a mixed mobile phase, ST, is given by the arithmetic average of the S-values P with the volume fraction of each solvent as weighting factors according to ST ¼ i Siφi. Several mobile phase optimization strategies are based on the use of isoeluotropic solvents, that is solvent mixtures of the same solvent strength prepared with water-miscible organic solvents from different selectivity groups. Binary isoeluotropic solvents can be mixed with each other in any proportion resulting in higher-order solvent mixtures with similar overall retention (solvent strength) while allowing changes in peak position due to changes in selectivity. S-values, while generally similar for low-mass compounds, do vary with their structure, tending to increase with compound size and for compounds of low polarity [1,35,36,42]. S-values, therefore, can only be considered semi-quantitative descriptors of solvent strength. The quadratic model extends Eq. (6.1) to a wider range of binary mobile phase compositions by allowing for the curvature generally observed in plots of ln k (or log k) against φ for binary mobile phase compositions as φ ! 0 or φ ! 1, Eq. (6.2), and Figure 6.2 [43]. ln k ¼ c0 + c1 φ + c2 φ2

(6.2)

For ternary mobile phase compositions, the model reverts to. ln k ¼ c0 + c1 φ1 + c2 φ2 + c12 φ1 φ2 + c11 φ1 2 + c22 φ2 2

(6.3)

with subscripts 1 and 2 in Eq. (6.3) indicating two different organic solvents with the balance always water. The equation coefficients c1, c2, etc., are merely fitting constants and are not assigned any physical meaning. The intercept c0 is often described as the retention factor with water as the mobile phase (ln kw) but is rarely an accurate assignment. The shape of ln k against φ plots is both solute and system dependent. The shape of the plot for a particular solute is not usually the same across different systems or for varied solutes in the same system. Also, the intercept term for Eq. (6.2) is generally different than the intercept term for Eq. (6.1) for the same system; is dependent on the identity of the organic solvent; is rarely independent of the organic

6.2 Parameters that affect retention

FIGURE 6.2 Plot of the logarithm of the retention factor (log k) as a function of the volume fraction of methanol (100) in reversed-phase liquid chromatography for a silica-based octadecylsiloxane-bonded stationary phase. Solute identification: 1—naphthalene; 2 ¼ bromobenzene; 3 ¼ acetophenone; 4 ¼ 2-phenylethanol; and 5 ¼ benzamide. Reproduced with permission from Ref. [43]. Copyright Elsevier Science Publishers.

solvent composition range used for linear extrapolation, Eq. (6.1); and when obtained directly for water wettable-bonded phases rarely agrees with values obtained by linear extrapolation from binary mobile phase mixtures [42–44]. Despite the evidence, some authors continue to use ln kw (or log kw) as a convenient measure of column and solute properties in quantitative structure-retention relationships and in studies of the retention mechanism [45]. This cannot be considered good practice. On mainly polar stationary phases, U-shaped log k vs φ plots may be observed for binary mobile phases [46]. These result from a change in retention mechanism from reversed phase for φ approximately 0–70% (v/v) organic solvent to hydrophilic interaction liquid chromatography (HILIC) for 70–95% (v/v) organic solvent. These boundaries are not hard, and the turning point varies with the solute and stationary phases. Also, within the intermediate mobile phase composition range, retention may be too weak for practical separations by either mechanism. Retention in HILIC is too complex to discuss here, see Chapter 10 for specific details. However, in HILIC, major contributions to retention arise from partition into an aqueous layer accumulated close to the stationary phase surface from an organic solvent-rich mobile phase as well as adsorption interactions at polar sites on the stationary phase. HILIC separations are complementary to RPLC and increasingly used for the separation of lowmass polar compounds weakly retained in RPLC. Other models used to accommodate nonlinearity in ln k vs φ plots for binary mobile phases include [5,19,38,39] ln k ¼ ln ko + S1 φ + S2 ln φ

(6.4)

97

98

CHAPTER 6 Reversed-phase liquid chromatography

ln k ¼ ln ko + 2lnð1 + S1 φÞ  ðφS2 =½1 + S1 φÞ k ¼ ðc1 + c2 φÞ

m

(6.5) (6.6)

Eq. (6.4) can accommodate mixed RPLC and HILIC retention mechanisms but is used mainly for HILIC separations. Higher-order models generally yield a better fit to the retention data at the risk of overfitting [5,38]. Eq. (6.5) has the advantage that it can be numerically integrated in gradient elution models. Eq. (6.6) provides good results for retention factors measured at the extreme range of mobile phase compositions. All the equations in this section find applications in current practice and there is no reason to presume primacy for any model if varied compounds and multiple systems are considered. Retention model coefficients can be determined from either isocratic or gradient elution modes [5,35,38,47,48]. Scanning gradients allow faster method development but have a lower predictive capability than data obtained solely from isocratic measurements. For gradient methods, important parameters are the selection of the gradient slope values for the scanning gradients, the number of selected scanning gradients, whether the predicted data falls within the range of the scanning gradients, and whether repeat measurements are utilized to improve experimental precision.

6.2.5.2 Exothermodynamic relationships Exothermodynamic relationships are usually formulated as free energy relationships on the assumption that the total free energy of transfer of a compound from the mobile phase to the stationary phase is an additive property of each structural unit of the compound. For members of a homologous series with a structure CH3(CH2)nX, for example, the total free energy is equal to ΔG°(CH3) + nΔG° (CH2) + ΔG°(X). If it is further assumed that the difference in free energy for a methyl and a methylene group is small enough to be ignored, then a plot of log k against the number of methylene groups (n or n + 1 in this case) is expected to be linear for a fixed mobile phase composition [49,50]. A similar relationship is observed for oligomers with monomer units of low polarity, such as styrene and ethylene oxide. The first few members of a homolog series (n < 2) may be an exception as well as homologs with long alkyl chains of a similar length to the bonded ligands of the stationary phase [51]. Nonlinear plots have also been observed for stationary phases with a high bonding density and mobile phase compositions with a high water content. Methylene group selectivity (αCH2) has been proposed as a scale for column hydrophobicity [52,53]. It affords only a rough measure of column equivalency, however, and can be misleading. The precepts of exothermodynamic relationships can explain the remarkable power of chromatography to separate substances that differ only slightly in structure. In this case, the difference in free energy of transfer (separation) is proportional to the free energy change for the structural difference and not for the rest of the molecule. As an extension of this concept, functional group contributions [ΔG°(X)] for a range of parent structures have been derived to predict retention in RPLC [54].

6.2 Parameters that affect retention

This approach has met with only limited success for polyfunctional compounds because the additivity principle fails to account for intramolecular interactions between substituent groups and solute-specific steric factors that affect solutesolvent interactions.

6.2.6 Temperature and Pressure 6.2.6.1 Temperature Column thermostats were introduced originally to improve the repeatability of retention measurements by providing a constant and stable set point temperature for the column environment [55]. Column thermostats supplied with modern instruments allow a limited range of temperatures (25–75°C) to be used for method development. At one time, higher temperatures (up to 200°C) with simultaneous regulation of the column backpressure attracted interest for high-temperature liquid chromatography [56,57]. In a limited way, temperature-programmed separations with narrow-bore columns were also demonstrated. Current interest, however, in high-temperature RPLC appears limited by the availability of column ovens that minimize thermal mismatch and extra-column band broadening and the variety of durable column packing materials. The general effect of temperature on retention in isocratic RPLC is described by the Gibbs-Helmholtz relationship, Eq. (6.7) [34,58–60] ln k ¼ ðΔH°=RT Þ + ðΔS°=RÞ + ln β

(6.7)

where ΔH° is the standard molar enthalpy and ΔS° the standard molar entropy of transfer of the solute from the mobile phase to the stationary phase, R the gas constant, T temperature (K), and β the phase ratio. Plots of ln k against 1/T for a small temperature range (40°C) are known as van’t Hoff plots and are ideally linear with the standard molar enthalpy obtained from the slope and the standard molar entropy from the intercept. Nonlinear plots are common, however, with apparent changes in enthalpy as a function of temperature resulting from interactions at multiple adsorption sites, temperature-induced changes in solute or stationary phase conformations, temperature-induced variation of the phase ratio, or variation in the heat capacity for the temperature range of the measurements. For mixed retention mechanisms, the observed retention factor represents the sum of interactions at the individual sorption sites, and a fundamental interpretation is impossible [60,61]. When accurate thermodynamic information is not the goal, empirical models containing additional terms in temperature can be used to describe the change in retention factors with temperature. ln k ¼ co + c1 =T + c2 ln T

(6.8)

ln k ¼ co + c1 =T + c2 T

(6.9)

ln k ¼ co + c1 =T + c2 =T

2

(6.10)

where co, c1, and c2 are regression constants with no assigned physical meaning.

99

100

CHAPTER 6 Reversed-phase liquid chromatography

Plots of the standard molar entropy (or free energy) against the standard molar enthalpy, known as an enthalpy-entropy compensation plot, are a useful tool for establishing the similarity of the retention mechanism for varied compounds [58,61,62]. When enthalpy-entropy compensation occurs, the plots are linear and the slope is called the compensation temperature. All related processes that share the same compensation temperature are assumed to proceed by the same mechanism. Some care is needed, however. It is entirely possible to observe adventitious correlations resulting from the spurious statistical fitting of the experimental data using linear regression [63]. It is important to establish that the variance explained by the plot exceeds the random experimental variance by an appropriate statistical test. Only when a highly variegated set of solutes is chosen and enthalpy-entropy compensation is observed, then it is reasonable to assume that the retention mechanism is dominated by a common factor. Elevated temperatures result in lower operating pressures due to the lower viscosity of the mobile phase and changes in peak shape and efficiency from enhanced diffusion regulated by the compromise between faster mass transfer and greater longitudinal diffusion to the column plate height. Higher temperatures result generally in lower retention and possible changes in selectivity, and this combined with changes in peak widths can make the prediction of resolution difficult. A 1°C change in temperature is expected to change retention factors by about 1–3% [61]. Although the possibility of using temperature as an optimization variable in method development is often ignored, the simultaneous optimization of mobile phase composition and temperature can be useful [35,57]. Temperature variation and composition variation show similar trends in retention but the capability to control retention is greater by composition variation. Generally, a 1% (v/v) change in composition has about the same effect on retention as a 4–5°C change in temperature. The predominant effect of higher temperatures is to decrease retention by a reduction in the difference in cohesive energy between the mobile and stationary phases, decrease retention due to an increase in the phase ratio without affecting selectivity, and increase retention due to a decrease in the hydrogen-bond acidity of the mobile phase relative to that of the stationary phase [64]. Temperature variation, therefore, is expected to have the most effect on selectivity for compounds of different sizes and hydrogen-bond basicity.

6.2.6.2 Pressure Recent years have witnessed an expansion in the use of chromatographic instruments capable of operation at pressure up to 1500 bar for use with columns packed with sub3 μm particles. This brought attention to the effect of the average column pressure on retention and the disturbing effects of axial and radial temperature gradients caused by high-pressure operation on retention. The effect of pressure on the retention factor is primarily the result of the change in partial molar volume of the solute on transfer from the mobile to the stationary phase and the compressibility of the mobile and stationary phases [65–67]. The change in partial molar volume tends to correlate with the solute size and varies with the average column pressure [67–70]. The net result is

6.3 Linear free energy relationships

an increase in retention at higher average column pressures, more noticeable for macromolecules but not inconsequential for small molecules. For column pressure gradients 15 nm) as well as monomeric alkylsiloxane-bonded silica stationary phases with chain lengths greater than octadecylsiloxane [97–99]. It is advantageous for the separation of conformationally constrained stereoisomers, for example, polycyclic aromatic hydrocarbons, carotenoids, and steroids. A slot model was proposed to explain the elution order for polycyclic aromatic hydrocarbons. The latter are separated

2.5

2.5

Category 3

Category 1

2

2 1.5

1.5 Log k

110

Category 2

1

1

0.5

0.5

0

0 0

10

20 30 40 50 % (v/v) Methanol

60

70

0

10

20

30

40

50

60

70

% (v/v) Methanol

FIGURE 6.6 Plot of the retention factor (log k) against volume fraction of methanol for representative compounds showing (1) category 1 behavior (1-naphthol), (2) category 2 behavior (1-nitrohexane), and (3) category 3 behavior (n-butyrophenone) on the Ascentis C18 column. 1-Nitrohexane and n-butyrophenone show a loss of retention over the composition range of 10–30% (v/v) methanol due to steric resistance. Reproduced with permission from Ref. [91], Copyright Elsevier Science Publishers.

6.3 Linear free energy relationships

approximately in order of their length-to-breadth ratio and planarity. There is no term in Eq. (6.11) to accommodate contributions from shape selectivity. The accessible silica surface of chemically bonded stationary phases contains a low concentration of anionic silanols [24,52,53,100]. These are mainly associated with type-A silica-based stationary phases and to a lesser extent type-B silica based. These accessible anionic silanols lead to higher retention for basic compounds than predicted by Eq. (6.11) due to electrostatic interactions with the silica surface. These interactions are typically enhanced for acetonitrile-water mobile phase compositions [36,79,85,86,92,94]. Electrostatic interactions have been cited as the cause of poor peak shapes for basic compounds in RPLC, and although undoubtedly a contributing factor, this is unlikely the complete story [100,101].

6.3.2 Hydrophobic-subtraction model The hydrophobic-subtraction model was developed specifically to characterize column selectivity for small molecules in RPLC with three general purposes in mind: (1) to facilitate the selection of replacement columns with similar selectivity; (2) to systematically select alternative columns with different selectivity for difficult separations; and (3) to identify columns with “orthogonal” selectivity for use in multidimensional chromatography [52,53,92,102,103]. The hydrophobic-subtraction model considers five independent solute-column contributions to retention identified as (i) hydrophobic interactions, (ii) shape selectivity, (iii) interaction of solute hydrogen-bond acid groups with basic column sites, (iv) interaction of solute hydrogen-bond base groups with acidic column sites, and (v) cation-exchange interactions with ionized silanols. log k ¼ log kEB + η’H - σ’S∗ + β’A + α’B + k’C ðiÞ

ðiiÞ

ðiiiÞ

ðivÞ

ðvÞ

(6.13)

The retention factor for ethylbenzene, kEB, is used as a reference to account for differences in the phase ratios of compared columns. The Greek letters (η0 , σ 0 , β0 , α0 , and κ0 ) are solute-specific parameters with complementary properties to the columnspecific parameters (H, S*, A, B, and C). The latter characterize the system’s contributions to the retention mechanism. The column-specific parameters are defined as hydrophobicity, H; steric hindrance (the resistance of the interphase region to penetration by bulky molecules), S*; hydrogen-bond acidity, A; hydrogen-bond basicity, B; and the cation exchange capacity of ionized silanols, C. Numerical values for the column-specific parameters are determined by measuring the retention factors for 16 reference compounds with an aqueous phosphate buffer (60 mM)-acetonitrile (1:1) mobile phase at 35°C. Values for C are generally indicated at low pH (C2.8) and neutral pH (C-7.0) to mimic typical conditions for RPLC for samples containing ionizable compounds. Hydrophobic interactions dominate Eq. (6.13), and to emphasize contributions from the other terms, the column-specific parameters are normalized to an average value for a hypothetical type-B silica octadecylsiloxanebonded stationary phase. Column-specific parameters for more than 750 columns

111

112

CHAPTER 6 Reversed-phase liquid chromatography

are now available making the hydrophobic-subtraction model the most comprehensive column selectivity database for RPLC [104]. Eq. (6.13) has been modified to improve the predictive capability of the model [104]. The σ 0 S* term was deleted as it generally lacked a commonsense interpretation [52,53], and two new terms were added to better account for the influence of the solute shape (vV) and dipole-type interactions (dD) on retention, with the remaining terms adjusted to accommodate these changes. log k ¼ log kEB + hH + bA + aB + κC + vV + dD

(6.14)

The new solute-specific parameters are the Connolly solvent-excluded volume with subtraction of the volume for ethylbenzene (calculated by software), v, and the dipolarity parameter initially described by the S descriptor of the solvation parameter model with subtraction of S for ethylbenzene and subsequently adjusted to an optimum value with the remaining parameters. For the solute-specific hydrogen-bond acidity parameter, a, the original α0 parameter set to zero for all compounds without hydrogen-bond donor properties and for the hydrogen-bond basicity parameter, b, the B descriptor for the solvation parameter model with subtraction of the B value for ethylbenzene were used as initial estimates. All solute-specific parameters except for v were then optimized across the column database to give the final values for the solute-specific parameters utilized to assign the column-specific parameters for Eq. (6.14). Eq. (6.14) afforded an almost three-fold improvement in the model standard error for the fit of the calibration compounds compared with Eq. (6.13). The hydrogen-bond acidity and ion exchange column-specific parameters are highly correlated in Eqs. (6.13) and (6.14), but the contributions from hydrophobicity and hydrogen-bond basicity are not. A portion of the hydrophobicity parameter in Eq. (6.13) is likely distributed between the hydrophobicity, shape, and dipolarity parameters in Eq. (6.14). Table 6.4 summarizes representative column-specific parameters for those columns characterized by the solvation parameter model in Table 6.3 [104]. Each column-specific parameter is referenced to a hypothetical octadecylsiloxane-bonded silica column, and therefore, their numerical values and sign only indicate a change compared to the reference column. In absolute terms, or between parameters, interpretation is not obvious, although the ranking for a single column-specific parameter should be correct with respect to the definition for that parameter. A term-by-term comparison with the solvation parameter model is rendered impossible due to a lack of a common set of interaction parameters and operational differences in how contributions in common are defined [52,92]. The hydrophobic-subtraction model provides an assessment of column properties at a single mobile phase composition whereas system maps employed by the solvation parameter model allow the effect of mobile phase composition (solvent type and volume fraction) on column selectivity to be evaluated. The F metric as a single-valued parameter is proposed to facilitate the identification of selectivity-equivalent columns or to identify columns that differ significantly in selectivity from the column-specific parameter database [103–105]. The F metric is based on the Euclidian distance between compared columns in the multidimensional

6.3 Linear free energy relationships

Table 6.4 Column-specific parameters of the hydrophobic-subtraction model for siloxane-bonded silica columns with acetonitrile-aqueous buffer (1:1) as mobile phase at 35°C. Stationary phase Ascentis C18 Betasil C18 Chromolith RP18e Discovery HS C18 HyPURITY C18 SunFire C18 XTerra MS C18 XBridge C18 Kinetex EVO C18 Lunar Omega PS C18 XBridge Shield RP18 Synergi Fusion RP XBridge Phenyl Kinetex PhenylHexyl Kinetex F5 Kinetex Biphenyl

Column-specific parameters kEB

H

A

B

C

V

D

10.46 11.14 3.41

1.007 0.996 0.979

0.083 0.167 0.018

0.031 0.010 0.027

0.027 0.030 0.232

0.008 0.062 0.008

0.067 0.027 0.023

4.82

0.958

0.143

0.011

0.155

0.073

0.009

5.55 9.87 6.35 4.62 4.18

0.952 1.050 0.969 0.989 1.023

0.112 0.005 0.136 0.117 0.130

0.004 0.016 0.054 0.001 0.118

0.132 0.011 0.086 0.175 0.196

0.053 0.045 0.082 0.052 0.095

0.005 0.032 0.024 0.011 0.072

7.93

0.990

0.142

0.100

0.097

0.103

0.073

3.65

1.066

0.143

0.147

0.145

0.097

0.338

7.50

1.005

0.072

0.127

0.240

0.031

0.159

2.35 3.49

0.845 0.985

0.147 0.058

0.249 0.002

0.043 0.104

0.196 0.032

0.221 0.009

2.90 2.53

0.883 0.874

0.122 0.181

0.332 0.432

0.047 0.336

0.135 0.338

0.275 0.332

space with the column-specific parameters normalized by introducing weighting factors to represent the properties of a “typical sample”. However, these weighting factors are unlikely to be ideal for all mixtures since the F-weights should reflect the relative importance of each column-analyte interaction, and their importance may vary from one sample type to another. In other words, given that individual samples may contain quite different compounds, any average or typical sample for which the F-weights are calculated can only be approximate and possibly poorly chosen for some samples. A general criterion for columns deemed equivalent is that they have an F metric 60 if the sample contains ionic compounds or an F metric >100 if it contains only neutral compounds.

113

114

CHAPTER 6 Reversed-phase liquid chromatography

References [1] Poole CF. The essence of chromatography. Amsterdam: Elsevier; 2003. [2] Poole CF. Influence of solvent effects on retention of small molecules in reversed-phase liquid chromatography. Chromatographia 2019;82:49–64. [3] Vailaya A, Horvath C. Retention in reversed-phase liquid chromatography: partition or adsorption. J Chromatogr A 1998;829:1–37. [4] Tan LC, Carr PW. Revisionists look at solvophobic driving forces in reversed-phase liquid chromatography – 2. Partitioning vs. adsorption mechanisms in monomeric alkyl bonded phase supports. J Chromatogr A 1997;775:1–12. [5] den Uijl MJ, Schoenmakers PJ, Pirok NWJ, van Bommel MR. Recent applications of retention modeling in liquid chromatography. J Sep Sci 2021;44:88–114. [6] Gritti F, Guiochon G. Heterogeneity of the adsorption mechanism of low molecular weight compounds in reversed-phase liquid chromatography. Anal Chem 2006; 78:5823–34. [7] Horvath K, Vajda P, Felinger A. Multilayer adsorption in liquid chromatography – the surface heterogeneity below an adsorbed multilayer. J Chromatogr A 2017;1505:50–5. [8] Lindsey RK, Rafferty JL, Eggimann BL, Siepmann JI, Schure MR. Molecular simulations of reversed-phase liquid chromatography. J Chromatogr A 2013;1287:60–82. [9] Gupta PK, Meuwly M. Dynamics of water/methanol mixtures at functionalized chromatographic interfaces. J Phys Chem B 2012;116:10951–9. [10] Rybka J, Holtzel A, Trebal N, Tallarek U. Stationary-phase contributions to surface defusion in reversed-phase liquid chromatography: chain length versus ligand density. J Phys Chem C 2019;123:21617–28. [11] Guillaume YC, Guinchard C. Participation of cluster species in the solvation mechanism of a weak polar solute in a methanol-water mixture over a 0.2-0.7 water fraction range: high-performance liquid chromatography study. Anal Chem 1998;70:608–15. [12] Marcus Y. The structure of and interactions in binary acetonitrile + water mixtures. J Phys Org Chem 2012;25:1072–82. [13] Wang M, Mallette J, Parcher JF. Strategies for the determination of the volume and composition of the stationary phase in reversed-phase liquid chromatography. J Chromatogr A 2008;1190:1–7. [14] Kazakevich YV. High-performance liquid chromatography retention mechanisms and their mathematical descriptions. J Chromatogr A 2006;1126:232–43. [15] Vajda P, Felinger A, Guiochon G. Evaluation of surface access isotherms in liquid chromatography. J Chromatogr A 2013;1291:41–7. [16] Buntz S, Figus M, Liu Z, Kazakevich YV. Excess adsorption of binary aqueous organic mixtures on various reversed-phase packing materials. J Chromatogr A 2012; 1240:104–12. [17] Gritti F. Thermodynamic interpretation of the drift and noise of gradient baselines in reversed-phase liquid chromatography using mobile phase additives. J Chromatogr A 2020;1633, 461605. [18] Kazakevich YV, LoBrutto R, Chan P. Interpretation of excess adsorption isotherms on the surface of reversed-phase adsorbents – effect on analyte retention. J Chromatogr A 2001;913:75–87. [19] Tsui HW, Kuo CH, Huang YC. Elucidation of retention behaviors in reversed-phase liquid chromatography as a function of mobile phase composition. J Chromatogr A 2019;1595:127–35.

References

[20] Poole CF. An interphase model for retention in liquid chromatography. J Planar Chromatogr 2015;28:98–105. [21] Neue UD, Cheng YF, Lu Z, Alden BA, Iraneta PC, Phoebe CH, Van Tran K. Properties of reversed phase packings with an embedded polar group. Chromatographia 2001;54:169–77. [22] Gritti F, Guiochon G. Characteristics of the adsorption mechanism of acidic-basic compounds with two pK(a) in reversed-phase liquid chromatography. J Chromatogr A 2009;1216:6917–30. [23] Wang C, Guo Z, Long Z, Zhang X, Liang X. Overloading study of basic compounds with positively charged C18 column in liquid chromatography. J Chromatogr A 2013;1281:60–6. [24] McDonald PD. Improving our understanding of reversed phase separations for the 21 st century. Adv Chromatogr 2003;42:323–75. [25] Ells B, Wang Y, Cantwell FF. Influence of solvent uptake and swelling by poly(styrenedivinylbenzene) column packings on sample sorption rate and band broadening in reversed-phase liquid chromatography. J Chromatogr A 1999;835:3–18. [26] Bolliet D, Poole CF. Influence of solvent effects on retention for a porous polymer sorbent in reversed-phase liquid chromatography. Chromatographia 1997;46:381–98. [27] Lepont C, Gunatillaka AD, Poole CF. Retention characteristics of porous graphitic carbon in reversed-phase liquid chromatography with methanol-water mobile phases. Analyst 2001;126:1318–25. [28] Vailaya A. Fundamentals of reversed-phase chromatography. Thermodynamic and exothermodynamic treatment. J Liq Chromatogr Relat Technol 2005;28:965–1054. [29] Tijssen R, Shoenmakers PJ, Bohmen MF, Koopal LK, Billiet HAH. Lattice models for the description of partitioning adsorption and retention in reversed-phase liquid chromatography including surface and shape effects. J Chromatogr A 1993;656:135–96. [30] Boehm RE, Martire DE. Theory of liquid chromatographic retention and solute transfer thermodynamics using the Bethe-Guggenheim quasi-chemical approach. J Phys Chem 1994;98:1317–27. [31] Dorsey JG, Dill KA. The molecular mechanism of retention in reversed-phase liquid chromatography. Chem Rev 1989;89:331–46. [32] Nikitas P, Pappa-Louisi A, Agraflotou. Affect of the organic modifier on retention in reversed-phase liquid chromatography. II test using various simplified models. J Chromatogr A 2002;946:33–45. [33] Kaliszan R, Wiczling P, Markuszewski MJ, Al-Haj MA. Thermodynamic vs. extrathermodynamic modelling of chromatographic retention. J Chromatogr A 2011; 1218:5120–30. [34] Nikitas P, Pappa-Louisi A. Retention models for isocratic and gradient elution in reversed-phase liquid chromatography. J Chromatogr A 2009;1216:1737–55. [35] Snyder LR, Dolan JW. High-performance gradient elution. The practical application of the linear-solvent-strength model. New York: Wiley-Interscience; 2007. [36] Poole CF, Poole SK. Foundations of retention in partition chromatography. J Chromatogr A 2009;1216:1530–50. [37] Chester TL. The combination of partition, size exclusion, and hydrodynamic models in chromatography, application to bonded phases on porous supports. J Chromatogr A 2020;1620, 461011. [38] Gilar M, Hill J, McDonald TS, Gritti F. Utility of linear and nonlinear models for retention prediction in liquid chromatography. J Chromatogr A 2020;1613, 460690.

115

116

CHAPTER 6 Reversed-phase liquid chromatography

[39] Jandera P, Hajek T, Ruzickova M. Retention mechanism on core-shell columns. J AOAC Int 2017;100:1636–46. [40] Jouyban A, Soltani S, Shayanfar A, Pappa-Louisi A. Modeling the effects of type and concentration of organic modifiers, column type and chemical structure of analytes on the retention in reversed-phase liquid chromatography using a single model. J Chromatogr A 2011;1218:6454–63. [41] Kwiemiewski I, Rycyk M. The use of Oscik’s equation to describe the retention in reversed-phase liquid chromatography. J Chromatogr A 2017;1496:45–50. [42] Poole CF, Poole SK, Gunatilleka AD. In search of a chromatographic model for biopartitioning. Adv Chromatogr 2000;40:159–230. [43] Poole CF, Poole SK. Column selectivity from the perspective of the solvation parameter model. J Chromatogr A 2002;965:263–99. [44] Tate PA, Dorsey JG. Column selection for liquid chromatographic estimation of the kw’ hydrophobicity. J Chromatogr A 2004;1042:37–48. [45] Poole CF, Atapattu SN. Determination of physicochemical properties of small molecules by reversed-phase liquid chromatography. J Chromatogr A 2020;1626, 461427. [46] Jandera P, Janas P. Recent advances in stationary phases and understanding of retention in hydrophilic interaction chromatography. A review. Anal Chim Acta 2017;967:12–32. [47] den Uijl MJ, Schoenmakers PJ, Schulte GK, Stoll DR, van Bommel MR, Pirok BWJ. Measuring and using scanning-gradient data for use in method optimization for liquid chromatography. J Chromatogr A 2021;1636, 461780. [48] Baeza-Baeza JJ, Garcia-Alvarez-Coque MC. Extension of the linear solvent strength retention model including a parameter that describes the elution strength changes in liquid chromatography. J Chromatogr A 2020;1615, 460757. [49] Jandera P. Reversed-phase liquid chromatography of homologous series. A general method for the prediction of retention. J Chromatogr 1984;314:13–36. [50] Moldoveanu SC, Caiali E, David V. Results from solvophobic theory applied to methylene selectivity in reversed-phase HPLC. J Liq Chromatogr Rel Technol 2018;41:24–32. [51] Tchapla A, Colin H, Guiochon G. Linearity of homologous series retention plots in reversed-phase liquid chromatography. Anal Chem 1984;56:621–5. [52] Poole CF. Chromatographic test methods for characterizing alkylsiloxane-bonded silica columns for reversed-phase liquid chromatography. J Chromatogr B 2018;1092:207–19. [53] Zuvela P, Skoczylas M, Liu JJ, Baczek T, Kaliszan R, Wong MW, Buszewski B. Column characterization and selection systems in reversed-phase high-performance liquid chromatography. Chem Rev 2019;119:3674–729. [54] Smith RM. Retention and selectivity in liquid chromatography. Amsterdam: Elsevier; 1995. [55] Wolcott RG, Dolan JW, Snyder LR, Bakalyar, Arnold MA, Nichols JA. Control of column temperature in reversed-phase liquid chromatography. J Chromatogr A 2000;869:211–30. [56] McNeff CV, Yan B, Stoll DR, Henry RA. Practice and theory of high temperature liquid chromatography. J Sep Sci 2007;30:1672–85. [57] Heinisch S, Rocca J-L. Sense and nonsense of high-temperature liquid chromatography. J Chromatogr A 2009;1216:642–58. [58] Ogden PB, Coym JW. Retention mechanism of a cholesterol-coated C18 stationary phase: Van’t Hoff and linear solvation energy relationships (LSER) approaches. J Chromatogr A 2011;1218:2936–43.

References

[59] Horner AR, Wilson RE, Groskreutz SR, Murray BE, Weber SG. Evaluation of three temperature – and mobile phase dependent retention models for reversed-phase liquid chromatographic retention and apparent retention enthalpy. J Chromatogr A 2019; 1589:73–82. [60] Sepsey A, Horvath E, Catani M, Felinger A. The correctness of van’t Hoff plots in chiral and achiral chromatography. J Chromatogr A 2020;1611, 460594. [61] Gritti F, Guiochon G. Adsorption mechanism and effect of temperature in reversedphase liquid chromatography of the classic Van’t Hoff plot in chromatography. Anal Chem 2006;78:4642–53. [62] Ranatunga R, Vitha MF, Carr PW. Mechanistic implications of the equality of compensation temperatures in chromatography. J Chromatogr A 2002;946:47–9. [63] Li J. Effect of temperature on selectivity in reversed-phase liquid chromatography, a thermodynamic analysis. Anal Chim Acta 1998;369:21–37. [64] Bolliet D, Poole CF. Influence of temperature on retention and selectivity in reversedphase liquid chromatography. Analyst 1998;123:295–9. [65] McCalley DV. The impact of pressure and frictional heating on retention, selectivity and efficiency in ultra-high-pressure liquid chromatography. Trends Anal Chem 2014;63:31–43. [66] Martin M, Guiochon G. Effects of high pressure in liquid chromatography. J Chromatogr A 2005;1090:16–38. [67] Gritti F, Guiochon G. Ultra high pressure liquid chromatography. Column permeability and changes in eluent properties. J Chromatogr A 2008;1187:165–79. [68] Paruda I, Barto E, Felinger A. Influence of pressure on the retention of resorcinarenebased cavitands. J Chromatogr A 2018;1535:123–8. [69] Mann BF, Makarov AA, Wang H, Welch CJ. Effects of pressure and frictional heating on protein separation using monolithic columns in reversed-phase chromatography. J Chromatogr A 2017;1489:58–64. [70] Asberg D, Samuelsson J, Fornstedt T. A fundamental study of the impact of pressure on the adsorption mechanism in reversed-phase liquid chromatography. J Chromatogr A 2016;1457:97–106. [71] De Villiers A, Lauer H, Szucs R, Goodall S, Sandra P. Influence of frictional heating on temperature gradients in UPLC on 2.1 mm I.D. columns. J Chromatogr A 2006;1113:84–91. [72] Makarov AA, Mann BF, Regaldo EL, Pirrone GF, Sun C, Sun SW, Nowak T, Wang H, Mangion I. Visualizing and studying frictional heating effects in reversed-phase liquid chromatography using infrared thermal imaging. Anal Chim Acta 2018;1018:1–6. [73] Fallas MM, Hadley MR, McCalley DV. Practical assessment of frictional heating effects and thermostat design on the performance of conventional (3 μm and 5 μm) columns in reversed-phase liquid chromatography. J Chromatogr A 2009;1216:3961–9. [74] Novakova L, Veuthey J-L, Guillarme D. Practical method transfer from high performance liquid chromatography to ultra-high performance liquid chromatography: the importance of frictional heating. J Chromatogr A 2011;1218:7971–81. [75] Haddad PR, Taraji M, Szucs R. Prediction of analyte retention time in liquid chromatography. Anal Chem 2021;93:228–56. [76] Poole CF. Evaluation of the solvation parameter model as a quantitative structureretention relationship model for gas and liquid chromatography. J Chromatogr A 2020;1626, 461308. [77] Vitha MH, Carr PW. The chemical interpretation and practice of linear solvation energy relationships in chromatography. J Chromatogr A 2006;1126:143–94.

117

118

CHAPTER 6 Reversed-phase liquid chromatography

[78] Studzinska S, Buszewski B. Linear solvation energy relations in the determination of specificity and selectivity of stationary phases. Chromatographia 2012;75:1235–46. [79] Poole CF, Nicole L. Applications of the solvation parameter model in reversed-phase liquid chromatography. J Chromatogr A 2017;1486:2–19. [80] Abraham MH, Ibrahim A, Zissmos AM. Determination of sets of solute descriptors from chromatographic measurements. J Chromatogr A 2004;1037:29–47. [81] Poole CF, Atapattu SN, Poole SK, Bell AN. Determination of solute descriptors by chromatographic methods. Anal Chim Acta 2009;652:32–53. [82] Poole CF, Ariyasena TC, Lenca N. Estimation of the environmental properties of compounds from chromatographic measurements and the solvation parameter model. J Chromatogr A 2013;1317:85–104. [83] Poole CF. Wayne State University experimental descriptor database for use with the solvation parameter model. J Chromatogr A 2020;1617, 460841. [84] Poole CF. Solvation parameter model: a tutorial on its application to separation systems for neutral compounds. J Chromatogr A 2021;1645, 462108. [85] Poole CF. Selection of calibration compounds for selectivity evaluation of siloxanebonded silica columns for reversed-phase liquid chromatography by the solvation parameter model. J Chromatogr A 2020;1633, 461652. [86] Poole CF, Atapattu SN. Selectivity evaluation of core-shell silica columns for reversedphase liquid chromatography using the solvation parameter model. J Chromatogr A 2020;1634, 461692. [87] Abraham MH, Acree WE. Descriptors for ions and ion-pairs for use in linear free energy relationships. J Chromatogr A 2016;1430:2–14. [88] Soriano-Meseguer S, Fuguet E, Abraham NH, Port A, Roses M. Linear free energy relationship models for the retention of partially ionized acid-base compounds in reversedphase liquid chromatography. J Chromatogr A 2021;1635, 461720. [89] Kiridena W, Poole CF, Koziol WW. Effect of solvent strength and temperature for a polar-endcapped, octadecylsiloxane-bonded silica stationary phase with methanolwater mobile phases. J Chromatogr A 2004;1060:177–85. [90] Bolliet D, Poole CF. Mixture-design approach to retention prediction using the solvation parameter model and ternary solvent system in reversed-phase liquid chromatography. Anal Commun 1998;35:253–6. [91] Atapattu SN, Poole CF, Praseuth MB. System maps for the retention of small neutral compounds on a superficially porous particle column in reversed-phase liquid chromatography. J Chromatogr A 2016;1468:250–6. [92] Poole CF. Reversed-phase liquid chromatography system constant database over an extended mobile phase composition range for 25 siloxane-bonded silica-based columns. J Chromatogr A 2019;1600:112–26. [93] Kiridena W, Atapattu SN, Poole CF, Koziol WW. Comparison of the separation characteristics of the organic-inorganic hybrid stationary phases XBridge C8 and phenyl and XTerra phenyl in RPLC. Chromatographia 2008;68:491–500. [94] Kiridena W, Atapattu SN, Poole CF, Koziol WW. System maps for reversed-phase liquid chromatography on an octadecylsiloxane-bonded silica stationary phase (SunFire C18). Chromatographia 2008;68:11–7. [95] Gritti F, Brousmiche D, Gilar M, Walter TH, Wyndham K. Kinetic mechanism of water dewetting from hydrophobic stationary phases utilized in liquid chromatography. J Chromatogr A 2019;1596:41–53.

References

[96] Poole CF, Ahmed H, Kiridena W, DeKay C, Koziol WW. Contribution of steric repulsion to retention on an octadecylsiloxane-bonded silica stationary phase in reversedphase liquid chromatography. Chromatographia 2005;62:553–61. [97] Carr PW, Dolan JW, Neue UD, Snyder LR. Contributions to reversed-phase column selectivity. 1. Steric interaction. J Chromatogr A 2011;1218:1724–42. [98] Limsavarn L, Dorsey JG. Influence of stationary phase solvation on shape selectivity and retention in reversed-phase liquid chromatography. J Chromatogr A 2006; 1102:143–53. [99] Mallik AK, Qiu HD, Takafuji M, Ihara H. High molecular-shape-selective stationary phases for reversed-phase liquid chromatography: a review. Trends Anal Chem 2018;108:381–404. [100] Bocian S, Buszewski B. Residual silanols at reversed-phase silica in HPLC. A contribution for a better understanding. J Sep Sci 2012;35:1191–200. [101] McCalley DV. The challenges of the analysis of basic compounds by high performance liquid chromatography. Some possible approaches for improved separations. J Chromatogr A 2010;1217:858–80. [102] Snyder LR, Dolan JW, Carr PW. The hydrophobic-subtraction model of reversed-phase column selectivity. J Chromatogr A 2004;1060:77–116. [103] Snyder LR, Dolan JW, Marchand DA, Carr PW. The hydrophobic-substraction model of reversed-phase column selectivity. Adv Chromatogr 2012;50:297–376. [104] Stoll DR, Dahliseid R, Rutan SC, Taylor T, Serret JM. Improvements in the prediction accuracy of the hydrophobic subtraction model for reversed-phase selectivity. J Chromatogr A 1636;2021:461482. [105] Graesboll R, Nielsen NK, Christensen JH. Using the hydrophobic-subtraction model to choose orthogonal columns for online comprehensive two-dimensional liquid chromatography. J Chromatogr A 2014;1326:39–46.

119

CHAPTER

Secondary chemical equilibria in reversed-phase liquid chromatography

7

Marı´a Celia Garcı´a-Alvarez-Coque, Jos e Ramo´n Torres-Lapasio´, M.J. Ruiz-Angel, and Jose Antonio Navarro-Huerta University of Valencia, Burjassot, Spain

7.1 Introduction Theoretically, the retention behavior in reversed-phase liquid chromatography (RPLC), with hydro-organic mixtures as mobile phases, is explained by the adsorption of the eluted compounds on the alkyl-bonded phase and their solubility in the mobile phase. Consequently, the behavior should be exclusively related to compound hydrophobicity: the more hydrophobic the compound, the longer its retention [1]. RPLC allows the separation of analytes in a wide range of polarities. However, ionized organic compounds and inorganic anions or metal ions, which are polar, show little or no retention. This has been a challenge in environmental, clinical, and food chemistry throughout the development of RPLC. Another challenge is related to the fact that there is still no ideal support for preparing RPLC stationary phases. Most supports are still prepared with silica, due to its attractive properties, including incompressibility, mechanical stability, easy derivatization, and the possibility to control particle size and porosity. However, due to steric problems in their derivatization, silanol groups remain on the stationary phase in a non-negligible amount and, when ionized, interact with ionic analytes by ion exchange. This results in the attraction or repulsion of cationic and anionic analytes to the anionic free silanols, respectively, which increases and decreases their retention, in some cases in a high extent. Furthermore, the sorption-desorption kinetics of analytes on free silanols is a slow process that produces broad, tailed peaks [2]. In the late 1970s, Horva´th and other authors published a series of seminal reports that tried to provide a solution to the separation of ionizable compounds and inorganic ions, using conventional RPLC instrumentation, silica-based materials, and hydro-organic mixtures [3]. In these reports, the idea of modulating the experimental conditions by introducing secondary reactions on the support or within the mobile phase, through the addition of several reagents, was suggested. The field has since expanded to include a variety of reactions: dissociation-protonation of ionizable compounds by tuning the pH, ion-exchange processes by adsorption of an ionic Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00012-6 Copyright # 2023 Elsevier Inc. All rights reserved.

121

122

CHAPTER 7 Secondary chemical equilibria

lipophilic reagent on the stationary phase to attract analytes with an opposite charge or suppress the silanol activity, formation of analyte-reagent ion pairs in the mobile phase, metal complexation, and redox reactions, among others [4]. In hydro-organic RPLC without additives, analytes are generally eluted in one chemical form, which is separated according to its distribution between the mobile and stationary phases. In the presence of secondary equilibria, analytes appear as at least two forms, which will separate according to the differences in their secondary equilibrium constants [5]. In general, these secondary equilibria can be expressed as: A + X⇆AX

(7.1) +

where A is the analyte or the silanol group on the support, and X is H , a lipophilic ion, a ligand, or other added species. The observed retention factor (k) is a weighted average of the retention factors of both chemical forms (A and AX): k ¼ kA δA + kAX δAX ¼

kA + kAX K ½X 1 + K ½X

(7.2)

where δA and δAX are their molar fractions, [X] is the molar concentration of X in the mobile phase, and K is the formation constant (for an acid-base reaction, log K ¼ pKa, where Ka is the dissociation constant). The situation can be more complex, since two or more secondary equilibria may exist simultaneously in both mobile and stationary phases. Secondary equilibria can provide improved selectivity for the separation of analytes under intermediate conditions, where comparable amounts of both chemical forms exist. Therefore, they represent a very powerful tool to enhance the chromatographic performance in RPLC (in terms of absolute and relative retention, and even peak shape). Secondary equilibria have given rise to new chromatographic modes, with an impressive increase in the number of compounds that can be analyzed by RPLC. The main features of several modes are described below.

7.2 Use of acid–base secondary equilibria 7.2.1 Changes in retention with pH Eq. (7.2) (with [X] ¼ [H+]) describes a sigmoidal change in RPLC retention of weak acids and bases as a function of the pH in the mobile phase, with a pronounced drop around pH ¼ pKa (referred to the hydro-organic mixture) [6–8]. The height of the transition depends on the hydrophobicity of the neutral species. Acids ionize by losing a proton when the pH increases (Figure 7.1A) and bases by accepting a proton when the pH decreases (Figure 7.1D). For polyprotic systems, the k-pH curve will depend on the charge of the different acid-base species. Small variations in the mobile-phase pH at values close to pKa result in significant changes in retention and selectivity. When this region is used, a strict control of the pH is needed. However, the implementation of robust methods requires a region hardly affected by changes in pH.

7.2 Use of acid–base secondary equilibria

HA Retention factor

B

A–

(D)

Retention factor

(A)

BH+

A–

B

BH+ HA

(B)

(E) BH +

Retention factor

HA

(C)

A–

pH

B

(F)

pH

FIGURE 7.1 RPLC retention versus pH trends of acidic (A–C) and basic (D–F) compounds, without additive (A and D), and in the presence of cationic (B and E) and anionic (C and F) additives. The arrows indicate the shifts in pKa.

The chromatographic mode used to separate weak acids at acidic pH, with a dominant-neutral species, has been called ion-suppression chromatography. To protonate (and thus deactivate) the silanols on the stationary phase, acidic pH is also used in the analysis of basic compounds. However, separations at low pH are not always feasible, due to the long analysis times and column instability. The retention

123

124

CHAPTER 7 Secondary chemical equilibria

behavior of ionizable analytes under gradient elution is especially cumbersome. Even using buffered mobile phases, the changes in organic solvent along the programmed gradient can lead to strong changes in the pH and pKa values for both the analyte and the buffer system [9,10].

7.2.2 Buffers and measurement of pH The working pH range for conventional silica-based columns in RPLC is 2.5–7.5. Outside this range, the silica packing can be severely damaged (below pH ¼ 2 siloxane bonds hydrolyze, and above pH ¼ 8 silica dissolves). Innovative supports that contain short carbon chains between the silicon atoms, as well as protecting polymer layers, have extended the pH range to 2–12. A buffer system is required to achieve reproducible retention for ionizable compounds at adequate pH values. Common buffers correspond to the acid-base systems of phosphoric, citric, tris(hydroxymethyl)aminomethane (Tris), phthalic, acetic, formic, and ammonium. The most popular are phosphoric and citric buffers, as they provide control over wide pH ranges, with the disadvantage of leading to precipitation of their inorganic salts within the column if the proportion of organic solvent is too high (particularly with acetonitrile) or at low column temperature. Only volatile buffers such as acetic, trifluoroacetic (TFA), and formic acids and their ammonium salts are compatible with evaporative light scattering (ELS) and mass spectrometry (MS) detection. However, TFA reduces the sensitivity in MS, particularly when working in the negative ion mode. Buffering capacity shows a maximum in the range pH ¼ pKa,buffer  1. The pH must be measured in the hydro-organic mixture, rather than in the aqueous buffer solution. Moreover, the electrode system should preferably be calibrated with standard buffers prepared with the same solvent composition as the mobile phase (the socalled sspH scale). As these standards are not commercially available and require careful maintenance, one solution is to measure the pH in the hydro-organic mixture and calibrate the electrode system with aqueous buffers (swpH scale, which can be easily converted to the sspH scale) [11,12]. Column temperature must be also controlled, as it affects the degree of ionization of analytes and buffers [13,14]. Hydrophilic interaction chromatography (HILIC) is accepted today as a complimentary separation mechanism to RPLC for the separation of polar and ionized solutes that are poorly retained in RPLC. In HILIC, retention has been attributed to partitioning of the solute between a water layer held on the surface of a polar stationary phase and the bulk mobile phase typically containing a high concentration of acetonitrile. The pH of solutions of formic, phosphoric, trifluoroacetic, and heptafluorobutyric acids cover a relatively narrow range when used in water (w wpH 1.9–2.8), but this range is much wider in 90% acetonitrile when the true thermodynamic pH scale is considered (sspH 2.4–5.2). These differences can partially explain the considerable selectivity changes observed between HILIC and conventional RPLC for ionizable compounds, with a relatively low amount of acetonitrile [15].

7.3 Ion-interaction chromatography

7.3 Ion-interaction chromatography 7.3.1 Retention mechanism The scope of RPLC is significantly broadened by adding amphiphilic anions or cations to the hydro-organic mixture [16–18]. The reagent added typically contains a hydrophobic tail that strongly interacts with the bonded chains in the stationary phase, and a charged head that projects into the mobile phase to interact with the analytes. The modification of the stationary phase facilitates the separation of mixtures of ionic and neutral species. Retention is regulated by the nature and concentration of the counterion in the reagent, the organic solvent, and competing ions with the same charge as the analyte added to the mobile phase. The retention mechanism when amphiphilic ions are added is not yet fully understood [18–20]. Due to the complexity of the mobile phases, which contain the ionizable or ionic analyte(s), and at least the additive and buffer ions (and their co-ions), it is not easy to determine their mutual influence on the adsorption behavior. In the origin of RPLC, bonded phases were considered equivalent to a “mechanically held liquid phase.” Therefore, the theory of the combination of the analyte and lipophilic ions of opposite charge to form an ion pair in the mobile phase, capable of partitioning into the non-polar stationary phase, is not surprising. Hence, the name ion-pair chromatography (IPC) is taken from liquid-liquid separations. Instead, experimental facts suggest a dynamic ion-exchange mechanism, which considers that the lipophilic ion is dynamically distributed between the mobile phase and stationary phase, where it is adsorbed (immobilized), behaving as an ion exchanger for oppositely charged analytes. This model involves an essentially Coulombic interaction and pioneered the stoichiometric approach that has been followed for decades. Broader perspectives (non-stoichiometric approaches) describe the ionic analyte as under the influence of all ions in the chromatographic system. In addition, the role of the double electrical layer formed by the lipophilic ion (a region of primary charged ions) and the counterion (a diffuse external region) is envisaged. Here, the analyte is not specifically associated with any charged moiety, and its retention requires its transfer through the double layer. This creates a surface potential, which depends primarily on three parameters: the surface concentration of the lipophilic ions and the relative permittivity and ionic strength of the mobile phase. The higher the surface concentration of the lipophilic ions, the greater the effective ion-exchange capacity, and hence the retention of solutes with a charge opposite to that of the lipophilic ion. This is expected to be spaced over the stationary phase due to repulsion, which leaves much of the surface unaltered and available for the separation of neutral species. The same framework does not apply to small hydrophilic organic and inorganic anions, which probably interact primarily through Coulombic forces. However, in general, other interactions within the mobile phase should not be neglected: the actual mechanism is therefore much more complex. Deep insight into the composite mechanistic processes is obscure since the accurate determination of equilibrium constants is difficult.

125

126

CHAPTER 7 Secondary chemical equilibria

IPC is by far the most widely used term for this RPLC mode, but from the above comments, it usually does not accurately describe the actual mechanism. The term “IPC” is often associated with the addition of a small amount of the lipophilic ion, avoiding any excess in the mobile phase. Meanwhile, the terms ion-interaction chromatography (IIC) or ion-modified chromatography have been suggested to describe the use of several types of ionic additives in RPLC at varying concentration. Other names are also found in the literature, such as paired-ion chromatography, hydrophobic chromatography with dynamically coated stationary phase, surfactant (or soap) chromatography (referring to the use of ionic detergents as additives), and hetaeric chromatography (referring to the use of hetaerons, “counterions”). The adsorbed amphiphilic reagent essentially changes the stationary phase from a non-polar (hydrophobic) to a polar (hydrophilic) surface, generating charge sites that serve as ion exchangers for the analytes, positive or negative depending on their nature (Figure 7.2). The main advantage of dynamic coating is the possibility of controlling the ion-exchange capacity of the column by varying the mobile-phase composition. A quite different alternative is the stationary-phase equilibration with a highly lipophilic ion. This coating is tightly bound and persists for long periods of subsequent use. The method is known as permanent coating IIC and is close to ion-exchange chromatography, where charged groups are covalently attached to the stationary phase.

7.3.2 Common reagents and operational modes In principle, any salt containing a lipophilic ion can be used as IIC reagent. To separate anions, the stationary phase must contain immobilized cations. Conversely, to separate cations, it must contain immobilized anions (Figure 7.2A and B). Alkylammonium or tetraalkylammonium salts for anions, and alkyl sulfates or alkylsulfonates for cations (with different alkyl chain lengths) cover the most common applications. The longer the alkyl chain, the more hydrophobic the reagent and stronger its adsorption on the stationary phase. The anion in alkylammonium salts can be inorganic (chloride, hydroxide, or phosphate) or organic (salicylate or tartrate). The cation for alkyl sulfate and alkylsulfonate salts is usually sodium or potassium. Newer reagents are perfluorinated carboxylic acids, chaotropic ions, and ionic liquids (ILs). New methods may be developed by tailoring the mobile-phase composition to suit the retention of a particular analyte, or the separation of a particular mixture. Popular choices tend to favor relatively less lipophilic IIC reagents to avoid long separation times. These reagents must be replaced by a more lipophilic ion when retention is too short. The same column can be converted into an anion exchanger or a cation exchanger. When needed, the adsorbed layer of lipophilic ion can be removed by washing the column with an organic solvent, such as methanol. As the concentration of IIC reagent increases, retention increases, if the surface of the stationary phase remains unsaturated. Meanwhile, as the organic solvent concentration increases, retention decreases due to desorption of the reagent and

NH3+

SO3

7.3 Ion-interaction chromatography

Si

Si

Si

Si

Si

Si

Si

Si

Si

SO3

SO3

SO3

SO3

SO3

NH3+ O OH O OH O O O O O O O O O O O O O O O O O Si

(A) O OH O OH O O O O O O O O O O O O O O O O O Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

N+

(D)

O OH O OH O O O O O O O O O O O O O O O O O Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

(B)

N+

N+

N+

c

N+

O OH O OH O O O O O O O O O O O O O O O O O Si

(C)

Si

Si

Si

Si

Si

Si

Si

Si

O OH O OH O O O O O O O O O O O O O O O O O

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

(E)

FIGURE 7.2 Simplified solute environments in a C18 chromatographic system with mobile phases containing (A) hexylamine, (B) 1-octanesulfonate, (C) 1-hexyl-3-methyl-imidazolium tetrafluoroborate, (D) sodium dodecyl sulfate, and (E) cetyltrimethylammonium chloride.

127

128

CHAPTER 7 Secondary chemical equilibria

competition equilibria in the mobile phase. Therefore, both the IIC reagent and organic solvent must be controlled in the mobile phase at specific concentrations, to maintain reproducible ion-exchange capacity. It is not essential that the IIC counterion functions as the ion-exchange competitor ion. Often a separate component, such as phosphate, citrate, oxalate or phthalate, is added to aid in the elution of strongly retained anions. The analytes must be ionized to interact with the IIC counterion. Therefore, the retention of ionizable compounds depends on the pH and their pKa (which changes due to the interaction of the ionic species with the IIC counterion, Figure 7.1B, C, E, and F). The adsorption of counterions on the stationary phase, the interaction between ionic solutes and counterions, and especially the ionization of solutes and buffer components are temperature dependent; therefore, the reproducibility of the separation requires precise temperature control. Other considerations are the requirement for a longer equilibration time to obtain a constant counterion coating (especially in gradient elution); the fact that some counterions tend to associate very strongly with the stationary phase, changing the initial properties of the column; the need to saturate the mobile phase with silica for some IIC reagents by inserting a guard column between the pump and the injection system; and the presence of system peaks in the chromatograms. Also, traditional lipophilic reagents are not usually compatible with ELS and MS detection.

7.3.3 Separation of inorganic anions Surfactant coatings are an easy and inexpensive way to convert silica-based RPLC packings into ion exchangers [21]. Its appeal stems from the different ion-exchange capabilities and selectivities, simply by altering the coating conditions. However, some problems with the stability of these coatings have been described, which make retention times drift, needing periodic regeneration of the column. This has belittled their use for routine separations. Reproducible behavior is only possible with careful equilibration to the plateau capacity of the column. Cationic surfactants with quaternary ammonium groups are frequently used in the analysis of inorganic anions. However, coating first with a layer of non-ionic surfactant and then with cationic surfactant creates a more efficient column yielding shorter retention times. On the other hand, when using a surfactant with only one functionality (anionic or cationic), release of analyte from the Stern layer to the bulk solution requires a mobile phase with a competing ion to exchange the analyte. If, instead, the stationary phase is coated with a zwitterionic surfactant (with positive quaternary ammonium and negative sulfonate groups close to each other), the analyte experiences simultaneous attractive and repulsive forces. This means that it can be retained by the stationary phase but also be released without the need for a competing ionexchange ion. This chromatographic mode, termed electrostatic ion chromatography, constitutes a kind of green chromatography, as the mobile phase can just be pure water or an electrolyte solution, such as NaHCO3 or Na2B4O7. The addition of a

7.3 Ion-interaction chromatography

cationic surfactant to the coating solution containing a zwitterionic surfactant reverses the elution order of monovalent and divalent anions. Aliphatic amines are also used as cationic ion pair reagents for the analysis of inorganic anions, either metallic (such as CrO42, VO3, MoO42, and WO42) and non-metallic (such as Cl, Br, I, NO2, NO3, and SO42), whose retention increases. Other applications concern the analysis of a variety of organic anions.

7.3.4 The silanol effect and its suppression with amine compounds Basic nitrogen compounds make up a significant fraction of the drugs used in modern therapy. Regrettably, RPLC analysis of such compounds with silica-based columns suffers from several problems, including long retention, poor efficiency, peak tailing, and strong dependence of retention on sample size. These effects are due to the ionic exchange of the cationic analyte on the active (dissociated) silanols on the support, the acidity of which raises due to the presence of metallic impurities [2]. Silanol ionization is not entirely suppressed by using mobile phases in the pH range 2.5–7.5. Consequently, much effort has been invested in bonded-phase chemistry to remove residual metal impurities and silanols. There is extreme diversity in the behavior of packing materials of the same type toward basic compounds, such as bonded C18-silica, which is due to differences in carrier silica, bonded silane type, amount of free silanols, and coating level, which all result in a variable concentration of surface silanols. However, instead of being inconvenient, brand-to-brand variation in the selectivity of bonded-phase materials is attractive. RPLC would never have reached such broad applicability if only hydrocarbon-like stationary phases were available. With the newer generation of RPLC columns, based on “ultrapure” silica and improved bonding technologies, the influence of surface silanols on the retention of basic analytes is less pronounced. However, some tailing remains. At least three solutions have been suggested to avoid the silanol effect [2,22]: reducing the pH to less than 3 to protonate residual silanols (however, using an extreme pH can damage the packing), increasing the pH to obtain neutral analytes (but simultaneously more silanols dissociate), and masking the electrostatic interaction of silanols with IIC reagents (but additional background appears for MS detection, and column properties may be permanently altered if the reagent cannot be removed from the stationary phase). Peak shapes can be improved by using acidic mobile phases containing hydrophobic anions, such as alkyl sulfates or alkylsulfonates, but this is not always successful, and the retention of basic compounds may be excessively increased by attraction to the mobile phase covered with the anionic additives. The use of amines as silanol blockers (suppressors or anti-tailing agents) is also widespread [23,24]. Better silanol suppression is achieved with bulky substituents, such as salts of quaternary amines (with alkyl chain lengths usually between 1 and 4) or amines with long alkyl chains (between 4 and 10), which yield stronger interactions. The most usual anion for amines is Cl. Other options are Br, OH, or PO43, and organic

129

130

CHAPTER 7 Secondary chemical equilibria

ions such as acetate, salicylate, or tartrate. Concomitant with the improvement in peak shape, the adsorbed amine decreases retention. The presence of an anion with adsorptive properties in the stationary phase can also affect the separation capability. Another option is to use a suitable combination of two oppositely charged counterions in the mobile phase, such as an alkylsulfonate and an amine. Whereas the alkylsulfonate acts as an IIC reagent, the organic amine masks the residual silanols, producing an efficient separation in a reasonable time.

7.3.5 Use of perfluorinated carboxylate anions and chaotropic ions The ionization of silanols and carboxylic groups in amino acids, peptides, proteins, and other biochemically relevant zwitterionic compounds can be suppressed at low pH. This can lead to early elution (and poor resolution) unless anionic reagents such as alkylsulfonates or perfluorinated carboxylates are added. However, alkylsulfonates can be strongly adsorbed on the stationary phase, making regeneration of the column difficult. Therefore, perfluorinated carboxylates are preferable, which are volatile and therefore compatible with ELS and MS detection and suitable for preparative chromatography. Among these, TFA is used most frequently due to its high purity, solubility in water, and transparency at 220 nm. Other less common volatile perfluorinated acids are pentafluoropropionic acid and heptafluorobutyric acid. In addition to perfluorinated carboxylates, other anions (mainly inorganic) are suitable for separating zwitterions and basic compounds in the low pH region. Longer retention and improved peak symmetry are obtained with anions with less localized charge, greater polarizability, and lower degree of hydration, with the following trend (called the Hofmeister or lyotropic series of anions):         PF 6 > ClO4 > BF4 > CF3COO > NO3 > Cl > CH3SO3 > HCOO > H2PO4 . More lipophilic anions can exhibit similar performance to traditional amphiphilic anions, but with fewer drawbacks. The retention mechanism of the most hydrophilic anions is not clear, since their adsorption capability is small. This has been explained considering that basic cationic analytes are usually well solvated by the aqueous mobile phase and have little affinity for the lipophilic phase. However, they can interact in the mobile phase with the anionic additives to form an ion pair, causing disruption of the solvation shell. As the ion pair is more lipophilic than the unpaired analyte, the stationary phase retains it more strongly. The ability to increase the disorder of water is called chaotropicity (or chaotropic effect) and depends on the position of the anion in the Hofmeister series. This effect also explains the influence of the nature of buffers on retention.

7.3.6 Use of ionic liquids Only the anion or the cation is adsorbed on the stationary phase for IIC reagents such as sodium hexanesulfonate and tetrabutylammonium hydroxide. In contrast, reagents such as hexylamine salicylate, butylammonium phosphate, or ILs have a dual

7.3 Ion-interaction chromatography

character (both the cation and anion are adsorbed) (Figure 7.2C). This creates a bilayer, positively or negatively charged, depending on the relative strength of the cation and anion adsorption, respectively. Although known primarily as green solvents, ILs behave in RPLC just like dissociated salts [25,26]. ILs are stable in water, soluble in typical RPLC solvents, and at low concentration, the viscosity of the mobile phase is not drastically altered. Meanwhile, various types of intermolecular interactions of ILs are maintained with the stationary phase and analytes: hydrophobic, electrostatic, and other specific ones. Most of the reported applications have focused on ILs with a large imidazolium    (or pyridinium) cation and the anions BF 4 , PF6 , Cl , or Br . The IL cation can interact through specific electrostatic interactions with silanols on the surface of alkylbonded silica and compete with the polar cationic group of basic analytes. At the same time, the alkyl group of the heterocyclic ring or quaternary cation in ILs can interact with the apolar alkyl groups of the stationary phase through hydrophobic and other non-specific interactions. The observed retention behavior and peak shape (peak broadening and tailing), with improvements in resolution, are a combination of the silanol-masking effect of the IL cation with the chaotropic character of its anion. The relative adsorption of the anion and cation of an IL is useful in adjusting the selectivity. If the compounds elute too rapidly, an IL with a short alkyl chain imidazolium cation, such as 1-ethyl- or 1-butyl-3-methylimidazolium, can be combined with a lyotropic anion, such as PF6, BF4, or ClO4. If the compounds are highly retained, a long alkyl chain cation such as 1-hexyl-3-methylimidazolium (the solubility of 1-octyl-3-methylimidazolium is too low to be practical) with a low lyotropic anion, such as Cl, should be used. If there is no problem with retention, ILs containing 1-hexyl-3-methylimidazolium and BF4 or Cl are recommended as the first choice. ILs with a stronger chaotropic anion such as PF6 cause excessive retention. A series of IL-based stationary phases with interesting properties have also been prepared for the separation of various compounds [27].

7.3.7 Measurement of the enhancement of column performance using additives Extremely narrow signals would yield maximal information quality in RPLC, but due to solute dispersion, the signals are peaks with diverse widths and asymmetries (non-Gaussian peaks are quite common in practice). Peak variance results from several factors with two origins: extra-column contributions (dispersion in the tubing, unions, and detector cell) and column (diffusion and interaction with the support and stationary phase). The magnitude of the latter contributions depends on the column geometry; substrate properties; and the type of interactions among solutes, stationary phase, and mobile phase (i.e., secondary equilibria). System performance can be conveniently visualized through the correlations between the peak half-widths and the retention time of analytes [28]. For isocratic elution, the plots are nearly linear. These can be obtained using the half-widths/retention time data for a set of analytes experiencing similar kinetics, eluted with a mobile

131

132

CHAPTER 7 Secondary chemical equilibria

phase of fixed or variable composition (if the kinetics of interaction with the stationary phase does not change). The half-width plot approach is a simple tool that facilitates the characterization of chromatographic columns. Different studies have shown that, likely, bulky additives do not interact with free silanols by direct association, and the peak shape enhancement with basic analytes occurs by interaction with the coating of additive on the stationary phase. Only small additives can block silanol groups by direct electrostatic interaction, but this masking mechanism appears to be less effective. The larger the size of the cation, and therefore its adsorption capability, the more intense the masking of the silanol effect (i.e., the better the peak shape). Meanwhile, the specific nature of the additive does not appear to influence the peak shape. Thus, for example, the benefits obtained in the presence of amines can be similar or even greater than those obtained in the presence of some ILs used as additives in RPLC [24]. To illustrate this behavior, several half-widths plots are depicted in Figure 7.3 for mobile phases containing amines and ILs. In the absence of additive, the slope of the right half-width is significantly larger than that of the left half-width, indicating tailing peaks. The three additives (cycloheptylamine, N,N-dimethyloctylamine, and 1-hexyl-3-methylimidazolium), especially the latter two, enhanced the peak shape (Figure 7.3C and D). This suggests that these additives efficiently hindered the access of basic drugs to the silanols in the column.

7.4 Micellar liquid chromatography 7.4.1 An additional secondary equilibrium in the mobile phase Above a certain concentration of an IIC reagent in the mobile phase, the stationary phase becomes saturated, and more reagent remains in the mobile phase. Beyond this threshold, retention, instead of continuing to increase, progressively decreases due to a series of side effects, such as the displacement of the adsorbed analyte by the IIC counterion, the formation of ion pairs between the analyte and IIC counterion in the mobile phase, or in the case of surfactants, the interaction with dynamic aggregates called micelles, which are formed above the so-called critical micelle concentration (CMC) [29–31]. Micelles behave as a new phase (a pseudophase) within the mobile phase, which leads into the field of another RPLC mode, named micellar liquid chromatography (MLC) (Figure 7.2D and E). MLC is classified among the pseudophase liquid chromatography modes, where the mobile phase contains entities that interact with analytes, such as micelles, cyclodextrins, vesicles, or nanometer-sized oil droplets in oil-in-water microemulsions. MLC has had more impact than other pseudophase modes. Its unique selectivity is attributed to the ability of micelles to organize solutes at the molecular level (Figure 7.4). However, the association between the surfactant monomers and the bonded phase (forming a structure like the micelle surface) has profound implications with respect to retention and selectivity. The amount of surfactant adsorbed

7.4 Micellar liquid chromatography

2.5 0.6

A, B (min)

2.0 1.5

0.4

1.0 0.2 0.5 0

0 0

10

30

20

(A)

0

2

0

2

4

6

8

10

12

(C) 1.6 0.6

A, B (min)

1.2

0.8

0.4

0.4

0.2

0

(B)

0 0

4

8

12

Time (min)

16

20

(D)

4

6

8

10

Time (min)

FIGURE 7.3 Half-widths plots (A, left half-width (●) and B, right half-width ()), including the data obtained with nine basic drugs (β-adrenoceptor antagonists). The RPLC mobile phases contained 15% acetonitrile but no additives (A), and different amounts of cycloheptylamine (B), N,N-dimethyloctylamine (C), and 1-hexyl-3-methylimidazolium (D).

remains constant or is near saturation above the CMC, which is an important characteristic with regard to robustness. The analytes are separated based on their differential partitioning between the bulk aqueous phase and the micellar aggregates or the surfactant-coated stationary phase. Therefore, a secondary equilibrium is added to the mobile phase, which can be altered for ionizable compounds by tuning the pH (Figure 7.1B, C, E, and F). Insoluble species partition via direct transfer of the micelles to the surfactant-modified stationary phase. Surfactants with ionic, zwitterionic, and non-ionic head groups can be used to separate ionic or neutral analytes that can interact with the surfactant. The steric

133

CHAPTER 7 Secondary chemical equilibria

60 d Low submicellar

50 Acetonitrile, % (v/v)

134

40

c b Hydro-organic

30 20

i

High submicellar

h

Transition region

g f

a

Micellar

e

10 0

0

0.04

0.08

0.12

0.16

SDS, M

1 2 3 4 5

(A)

2

6

0

10

7 8 9 20

2

40

50

53

4

6

6

1

60

0

7

9

8

2

1 2

(C) 3 5

10 9

3

4

5

0

4

68

7

9

2

80

120 0

8 7 64

40

(H) 76 4

0

20 Time (min)

200

7 6 4

40

10 9 60

76

4

9 10

9 10 40

0

80

(I)

3 5 8

10 9 60

5

20 1 2

3 5 8

150

10 9

1 2

35

3

8

4

40

100

(F)

6 7 8

10 0

50

10

1

(E)

10

(G)

20

1

2 3 5

2

0

6 87

1

(D)

8 4

2

(B)

10

30

1

0

5 4 3

1

10

20

FIGURE 7.4 Chromatographic performance for mobile phases containing acetonitrile in the absence and presence of SDS, using a C18 Kromasil column. Top: Mobile-phase compositions. Bottom: Chromatograms for (A) 15% acetonitrile, (B) 30% acetonitrile, (C) 30% acetonitrile/0.001 M SDS, (D) 50% acetonitrile/0.005 M SDS, (E) 10% acetonitrile/0.1125 M SDS, (F) 17.5% acetonitrile/0.1125 M SDS, (G) 25% acetonitrile/0.1125 M SDS, (H) 35% acetonitrile/ 0.1125 M SDS, and (I) 45% acetonitrile/0.1125 M SDS. Compounds: (1) atenolol, (2) carteolol, (3) pindolol, (4) timolol, (5) acebutolol, (6) metoprolol, (7) esmolol, (8) celiprolol, (9) oxprenolol, and (10) labetalol.

7.4 Micellar liquid chromatography

factor can also be important. The anionic sodium dodecyl sulfate (SDS) is by far the most common surfactant in MLC, used in two-thirds of the reports, followed by the cationic cetyltrimethylammonium bromide (CTAB) and the non-ionic polyoxyethylene-(23)-dodecyl ether (Brij-35) [29,30]. Brij-35 is also applied to emulate in vitro the partitioning process in biomembranes in a mode called biopartitioning MLC [32]. The polar hydrophilic head of the Brij-35 molecule (the polyoxyethylene chain with a hydroxyl end group, which is oriented away from the surface of the stationary phase) increases the polarity of the stationary phase, which remains neutral. The hydroxyl end group of Brij-35 can also interact with polar or moderately polar solutes by the formation of hydrogen bonds with hydroxyl and amino groups, increasing their retention [33].

7.4.2 Hybrid micellar liquid chromatography The idea of using aqueous micellar solutions as mobile phases (i.e., only water and surfactant) is attractive, but suffers two main drawbacks compared to conventional RPLC: excessive retention of apolar compounds and poor efficiency owing to the increased volume of stationary phase because of the adsorption of surfactant. This reduces the rate of analyte mass transfer within the stationary phase. To reduce retention to practical values, the alcohols propanol, butanol, and pentanol (especially 1-propanol) are usually added to the mobile phase, giving rise to the so-called hybrid MLC. Acetonitrile, a common solvent in RPLC, has been used sparingly. Butanol and pentanol are useful for increasing the strength of the mobile phase and eluting strongly retained compounds. Equally important is that organic solvents reduce the amount of surfactant adsorbed on the stationary phase, improving the peak shape, with similar or better performance with respect to conventional RPLC (Figure 7.4). The highly symmetrical peaks obtained with SDS for cationic basic drugs indicate that the ion-exchange mechanism with the sulfate group in the surfactant is a fast process and prevents the penetration of the analyte into the bonded alkyl chains to interact with the buried silanols. A drawback is that the attraction of cationic compounds to the negatively charged stationary phase can significantly increase retention. In contrast, the addition of Brij-35 to an organic mobile phase produces poor peak shape for basic drugs. However, the efficiency with Brij-35 increases significantly with temperature, approaching that obtained with an acetonitrile-water eluent at 80°C [33]. As discussed, the anionic SDS requires the addition of an organic solvent to decrease retention times and increase efficiency, especially for basic drugs. Meanwhile, the non-ionic Brij-35 has the interesting characteristic of reducing the polarity of the stationary phase. This significantly reduces retention times. However, the retention of polar compounds may be too short in the absence of specific interactions with Brij-35. An interesting solution is the preparation of mixed mobile phases of SDS and Brij-35 without organic solvent [34]. This results in successful “green” RPLC procedures, which produce good resolution and adequate separation times for basic drugs of intermediate polarity.

135

136

CHAPTER 7 Secondary chemical equilibria

Although hybrid MLC is still predominantly micellar in nature, micelles are affected by the organic solvent, leading to changes in CMC and surfactant aggregation number. A high percentage of organic solvent is in principle undesirable, because of micelle disruption. The concentration of organic solvent that still preserves the integrity of the micelles is approximately 15% for propanol and acetonitrile, 10% for butanol, and 6% for pentanol (the solubility of the two latter alcohols increases significantly in the surfactant medium). However, a submicellar RPLC mode (with surfactant monomers in the mobile phase but without micelles), obtained at high concentration of surfactant and organic solvent, can produce good resolution and short separation times (Figure 7.4H and I) [35]. In this mode, peak shape is improved over MLC, due to the significant reduction in the amount of surfactant that covers the stationary phase. The most interesting characteristics of MLC are the richness of interactions among solutes, stationary phase, aqueous phase, and micelles; the possibility of separating neutral compounds and ions in a single run, or analytes of different hydrophobicity in narrower retention time windows than for classical RPLC (which makes gradient elution less necessary); the high solubilization capacity of the micelles, which facilitates the dissolution of most matrices (saving time in sample preparation and allowing direct injection of physiological fluids into the column); the low concentration of organic solvents (translated into lower cost, toxicity, and environmental impact of the wastes compared to conventional RPLC); less evaporation of organic solvents (which makes the micellar phases stable for a longer time and easily recyclable); and improved luminescence detection, among others. The only real limitation is related to the use of ELS and MS detection, as direct online coupling is hampered by the presence of high concentrations of surfactant in the mobile phase. Despite the benefits of implementing MLC procedures using isocratic elution, the use of gradients may be desirable to reduce analysis times when separating mixtures of analytes within a wide range of polarities [36]. Linear gradients of organic solvent with acetonitrile, propanol, or butanol are preferable, keeping the surfactant concentration constant (mainly SDS or Brij-35). Gradient elution in MLC is especially attractive for determining moderate-to-low polar analytes in physiological samples by direct injection [37]. However, a high content of organic solvent is incompatible with this type of sample, due to micelle disruption with the subsequent precipitation of proteins. To avoid this, the proteins must be eluted under pure micellar conditions or in low organic solvent at the beginning of the separation. After the protein is removed from the column, the elution strength can be increased to gradually separate the analytes and obtain adequate retention even for those most retained, reaching submicellar conditions if necessary. However, when a column is coated with surfactant, and the gradient involves a rapid change in organic solvent, significant alterations in the baseline (produced by surfactant desorption) can distort the signals and ruin the separation [38]. The use of linear gradients can be an acceptable alternative, even when using extended ranges of organic solvent, while complex multilinear gradients should be avoided. Only if the different consecutive segments of the

7.4 Micellar liquid chromatography

gradient have moderate or similar slopes, acceptable baselines are obtained. The only disadvantage of using linear organic solvent gradients is the need for column reequilibration before a new gradient begins. Baseline and re-equilibration problems are reduced by using a C8 column, where the amount of surfactant adsorbed is smaller.

7.4.3 Microemulsion liquid chromatography Oil and water are immiscible, due to the high surface tension between the two liquids. However, in the presence of micellized surfactant (and often of co-surfactant), an organized, macroscopically homogeneous, and thermodynamically stable liquid system (a microemulsion) is formed. Microemulsion liquid chromatography (MELC) is a relatively new chromatographic mode, in which the mobile phase is an oil-in-water (O/W) microemulsion [39–41], consisting of nanometer-sized droplets of a waterimmiscible liquid (e.g., typically heptane, octane, cyclohexane, diisopropyl ether, or ethyl acetate) stabilized in a small amount at the core of a microstructure (the micelle). The addition of surfactant (usually SDS or Brij-35) and co-surfactant (especially 1-propanol, 1-butanol, or 1-pentanol) reduces the surface tension at the O/W interface to almost zero, due to micelle formation, which offers a defined boundary between the two immiscible liquids. The co-surfactant is needed since it meets the geometric requirements to obtain the proper curvature in the interfacial region of the micelle. The co-surfactant carbon tail is in the oil, while the hydrophilic group remains in the water, bridging the O/W interface. The high water content of O/W microemulsions makes them compatible with RPLC columns, while the droplets of hydrophobic oil offer the ability to dissolve apolar analytes and sample matrices. MELC has been developed based on the principles of MLC. An O/W microemulsion is, in fact, a modification of a micellar system. Like MLC, in addition to the incorporation of a pseudophase into the mobile phase, the stationary phase is modified by the adsorption of surfactant monomers. All this provides secondary mechanisms for the separation of analytes, which partition from both the mobile and stationary phases into the microemulsion droplets. As both co-surfactant and oil molecules are also adsorbed on the stationary phase, solute interactions in MELC are more complex relative to MLC. As in the case of hybrid MLC and the submicellar mode (Section 7.4.2), using SDS as surfactant, satisfactory peak profiles are obtained for basic compounds (which in hydro-organic RPLC produce broad and asymmetric peaks). This means that the suppression of the silanol effect by the surfactant coating of the stationary phase is preserved. Changes in polarity and concentration of the MELC reagents affect not only the formation of stable oil droplets (within the swollen micelles), but also the nature of the stationary phase. All this leads to variations in the equilibrium distributions between the solutes and the two phases, and consequently, to changes in the absolute and relative retention of analytes. The advantages of reduced retention times and improved peak profile in MELC must be added to the ability of microemulsions to dissolve compounds in a wide range of polarities, analyze water-insoluble samples (such as creams, ointments,

137

138

CHAPTER 7 Secondary chemical equilibria

and suppositories) containing compounds with low polarity, and the possibility of direct injection of physiological samples, which is useful for the determination of apolar compounds. MELC mobile phases have been considered as environmentally friendly alternatives to traditional solvents in RPLC, because of the low amount of organic solvent needed and low volume of mobile phase due to the short separation times. However, the oil phase of typical microemulsions is still volatile, toxic, and flammable. Therefore, it would be desirable to avoid it from an environmental perspective. Certainly, the best solvent would be no solvent (pure water), considering health hazards, waste generation and treatment, and the economy. However, water has the drawback of the low solubility of most organic compounds. This brings us back to the use of ILs. Recently, the non-polar organic solvent (octane) used in the MELC analysis of phenolic acids was replaced with a low-polarity IL (1-hexyl-3-methylimidazolium hexafluorophosphate), which contains a long alkyl chain capable of forming micelles in aqueous medium [42]. A microemulsion made up of SDS, 1-butanol, and this IL resulted in a more environmentally friendly and successful MELC procedure.

7.5 Metal complexation 7.5.1 Determination of metal ions RPLC is a good alternative to direct spectroscopic methods and ion-exchange chromatography, as it can determine several metals simultaneously, removing matrix interferences, coupling with different detectors, and allowing high sensitivity. However, direct IIC separation of transition metal ions is difficult due to their similar hydration energies. The required selectivity is achieved through a series of secondary equilibria: complex formation, dynamic ion exchange, and eventually, ion pair or association with a micelle in the mobile phase in addition to acid-base equilibria [43–45]. Neutral complexes elute with hydro-organic mixtures, but most often complexes are anionic, and therefore, alkylammonium or tetraalkylammonium salts of a wide range of lipophilicity are used to retain them in the IIC mode, with or without a competing anion in the mobile phase. Cationic surfactants such as CTAB or cetylpyridinium chloride can also be used below or above the CMC. There are two main approaches: precolumn (offline) formation of the complexes with subsequent separation, and injection of the metal ions combined with online formation with a ligand added to the mobile phase. The feasibility of these approaches depends on the stability of the complexes. Usually, binary complexes are formed with some examples of ternary complexes to improve both selectivity and sensitivity. The separation of chelates with metallochromic ligands containing highly absorbent chromophores dispenses with the need for postcolumn derivatization, with sub-μg/mL-level detection limits. Higher selectivity and sensitivity can be achieved using fluorimetric reagents, which can reach ng/mL levels.

7.5 Metal complexation

Many chelating reagents are used, several adopted from spectrophotometric methods, such as 8-hydroxyquinoline, 4-(2-pyridylazo)resorcinol, 1,10phenanthroline, and various dithiocarbamates and azo dyes, which form stable neutral or ionic chelates with a number of metal ions, easily detected by spectrophotometry. In some cases, selectivity is improved by adding a second ligand to mask interferences and eliminate the corresponding peak. The integrity of metal chelates is susceptible to pH, as side reactions are expected at low pH with the ligand (protonation) and at high pH with the metal ions (formation of complexes with hydroxyl ion). Because of these reactions, the narrow pH range of conventional columns may be unsuitable for complex formation. Complete separation of the chelate from the excess reagent added at the offline chelation step allows detection of the chelate signals in the absence of background contributions. When the complexation reaction is slow at room temperature, it may be necessary to heat prior to injection. Selective and sensitive analysis is possible by combining offline complexation with solvent extraction, which also allows the analysis of neutral complexes. The poor solubility in water of some chelates requires a mobile phase with a high proportion of organic solvent or surfactant. With offline complexation, only thermodynamically or kinetically stable chelates survive during elution and reach the detector, as each chelate migrates separately from the ligand, resulting in a steep decrease in ligand concentration close to the chelate peaks. Under these conditions, weak complexes tend to dissociate on column, typically through solvolysis or ligand-exchange reactions. This means that the column can function not only as a conventional separation device, but also as a kinetic discriminator to selectively detect chelates. The approach has been named kinetic differentiation chromatography. Here, the synergic interactions of four unique selectivity origins are combined: precolumn chelation, chromatographic separation, dissociation kinetics, and spectral selectivity. Many chelates used to determine metal ions by spectrophotometry after solvent extraction are not strong enough and dissociate in the RPLC column. This can be prevented by a combination of offline and online approaches (i.e., injection of the complexes and inclusion of the ligand in the mobile phase). Furthermore, a strong chelating reagent may be useful for metal ion extraction, but not at all for RPLC separation, due to the lack of selectivity or instability of the complexes at the separation conditions, or for detection. The ligand-exchange approach can solve this problem by replacing the first ligand with another added to the mobile phase. A simpler approach is the direct injection of the metal ion (without prior extraction), which is complexed inside the column (online), known as dynamic chelating (or complexation) chromatography. The separation is based on a combination of ion exchange and complexation selectivity, which is provided by the strengths and rates of reaction of the metal with the ligand and the IIC counterion in the mobile phase. Kinetic problems can be alleviated by controlling the temperature of the column or by using more suitable ligands. Hydrophobic metallochromic ligands, such as xylenol orange and methyl thymol blue, can be used to coat an RPLC stationary phase, producing a chelating capacity to

139

140

CHAPTER 7 Secondary chemical equilibria

separate metal ions. Two approaches are possible: precoating the stationary phase with the ligand and eluting with an inorganic salt or including the ligand within the mobile phase to dynamically coat the stationary phase. The second approach allows for higher column capacity and stability, better separation efficiency and selectivity, and the ability to exploit the ligand in the mobile phase for metal detection.

7.5.2 Determination of organic compounds Metal cations can also be used to modulate the selectivity for the separation of organic compounds by complex formation. There are two basic approaches: the introduction of metal ions into the stationary phase or into the mobile phase. When metal ions are added as salts of non-complexing anions, such as nitrate or perchlorate, the mobile phase must be acidic to avoid hydrolysis of the metal. Column performance is often poor with this approach in terms of selectivity and peak shape. The addition of charged metal chelates (anionic or cationic depending on the analyte charge) to the mobile phase is a more versatile and simpler approach, which has shown improved performance compared to conventional IIC reagents, especially for the separation of free or derivatized amino acids and peptides, and aromatic compounds. A ligand-exchange process can occur between the analyte and the ligands in the complexes. In some cases, the formation of ternary complexes has also been suggested. This process involves hydrophobic selectivity, but steric selectivity can be quite high, connected with the conformationally rigid structures of the chelates, which act as templates. The choice of metal is a compromise among several factors, such as the ability to form complexes, the solubility in the hydro-organic solvent, and detection. The general classification of transition metals according to their tendency to form complexes is as follows: Pt4+ > Pd2+ > Hg2+ > Cu2+ > Ni2+ > Co2+ > Zn2+ > Cd2+ > Fe2+ > Mn2+ > Ag+ (inversions can occur depending on the ligands). Metal salts of Cu2+, Ni2+, Zn2+, and Ag+ are most common for the analysis of organic compounds. Silver ion (or argentation) chromatography is particularly useful for lipid analysis. In this approach, the incorporation of Ag+ into the solid support is preferred since a mobile phase containing Ag+ is difficult to handle.

7.6 Redox reactions When analytes exhibit redox behavior, redox reactions can also be useful to improve separation selectivity. The redox reaction can occur in the column (on-column) or online [46]. On-column derivatization can be assisted by the redox activity of the packing material, such as porous graphitic carbon or carbon manipulated using an electrochemically modulated liquid chromatographic technique. The analyte migrates along the column as a mixture of oxidized and reduced forms, so retention is determined by the relative concentration of both within the column (like the case

References

for acid-base species, Eq. (7.2)). The online system consists of two separation columns with one redox derivatization unit between them. The redox reaction proceeds rapidly in the derivatization unit, so that the analyte migrates as its original form in the first column, while as its oxidized or reduced form in the second column. The retention of the analytes is controlled by the length of the two separation columns.

Acknowledgment The authors thank the Ministry of Science and Innovation of Spain for the financial support of Project PID2019-106708GB-I00 (State Research Agency: Ref. PROYECTO/AEI/10.13039/ 501100011033).

References [1] Garcı´a-Alvarez-Coque MC, Baeza-Baeza JJ, Ramis-Ramos G. Reversed phase liquid chromatography. In: Anderson JL, Stalcup A, Berthod A, Pino V, editors. Analytical separation science series, Vol. 1. New York: Wiley; 2015. p. 159–97. [2] Nawrocki J. The silanol group and its role in liquid chromatography (review). J Chromatogr A 1997;779:29–71. [3] Horva´th C. Enhancement of retention by ion-pair formation in liquid chromatography with nonpolar stationary phases. Anal Chem 1977;49:2295–305. [4] Lochm€uller CH, Hangac HH. Mobile phase additives vs. bonded phases for HPLC (review). Crit Rev Anal Chem 1997;27:27–48. [5] Foley JP, May WE. Optimization of secondary chemical equilibria in liquid chromatography: theory and verification. Anal Chem 1987;59:102–9. [6] Schoenmakers PJ, Tijssen R. Modelling retention of ionogenic solutes in liquid chromatography as a function of pH for optimization purposes (review). J Chromatogr A 1993;656:577–90. [7] Neue UD, Phoebe CH, Tran K, Cheng YF, Lu Z. Dependence of reversed-phase retention of ionisable analytes on pH, concentration of organic solvent and silanol activity. J Chromatogr A 2001;925:49–67. [8] Roses M, Bosch E. Influence of mobile phase acid-base equilibria on the chromatographic behaviour of protolytic compounds (review). J Chromatogr A 2002;982:1–30. [9] Zisi C, Fasoula S, Pappa-Louisi A, Nikitas P. Properties of the retention time of ionizable analytes in reversed-phase liquid chromatography under organic modifier gradients in different eluent pHs. J Chromatogr A 2013;1314:138–41. [10] Andres A, Roses M, Bosch E. Prediction of the chromatographic retention of acid-base compounds in pH buffered methanol-water mobile phases in gradient mode by a simplified model. J Chromatogr A 2015;1385:42–8. [11] Roses M. Determination of the pH of binary mobile phases for reversed-phase liquid chromatography (review). J Chromatogr A 2004;1037:283–98. [12] Suu A, Jalukse L, Liigand J, Kruve A, Himmel D, Krossing I, Roses M, Leito I. Unified pH values of liquid chromatography mobile phases. Anal Chem 2015;87:2623–30.

141

142

CHAPTER 7 Secondary chemical equilibria

[13] Pous-Torres S, Torres-Lapasio´ JR, Baeza-Baeza JJ, Garcı´a-Alvarez-Coque MC. Combined effect of solvent content, temperature and pH on the chromatographic behaviour of ionisable compounds. J Chromatogr A 2007;1163:49–62. [14] Gagliardi LG, Tascon M, Castells CB. Effect of temperature on acid-base equilibria in separation techniques (review). Anal Chim Acta 2015;889:35–57. [15] McCalley DV. Study of retention and peak shape in hydrophilic interaction chromatography over a wide pH range. J Chromatogr A 2015;1411:41–9. [16] Bartha A, Vigh G, Varga-Puchony Z. Basis of the rational selection of the hydrophobicity and concentration of the ion-pairing reagent in reversed-phase ion-pair high-performance liquid chromatography. J Chromatogr 1990;499:423–34. [17] Cecchi T. Ion pairing chromatography (review). Crit Rev Anal Chem 2008;38:161–213. [18] Garcı´a-Alvarez-Coque MC, Ramis-Ramos G, Ruiz-Angel MJ. Liquid chromatography, ion pair. In: Reedijk J, editor. Reference module in chemistry, molecular sciences and chemical engineering series. Waltham, MA: Elsevier; 2015. [19] Chen JG, Weber SG, Glavina LL, Cantwell FF. Electrical double-layer models of ionmodified (ion-pair) reversed-phase liquid chromatography (review). J Chromatogr A 1993;656:549–76. [20] Cecchi T. Theoretical models of ion pair chromatography: a close up of recent literature production. J Liq Chromatogr Relat Technol 2015;38:404–14. [21] Gennaro MC, Angelino S. Separation and determination of inorganic anions by reversedphase high-performance liquid chromatography (review). J Chromatogr A 1997;789:181–94. [22] McCalley DV. The challenges of the analysis of basic compounds by high performance liquid chromatography: some possible approaches for improved separations (review). J Chromatogr A 2010;1217:858–80. [23] Kiel JS, Morgan SL, Abramson RK. Effects of amine modifiers on retention and peak shape in reversed-phase high-performance liquid chromatography. J Chromatogr A 1985;320:313–23. [24] Calabuig-Herna´ndez S, Garcı´a-Alvarez-Coque MC, Ruiz-Angel MJ. Performance of amines as silanol suppressors in reversed-phase liquid chromatography. J Chromatogr A 2016;1465:98–106. [25] Ferna´ndez-Navarro JJ, Garcı´a-Alvarez-Coque MC, Ruiz-Angel MJ. The role of the dual nature of ionic liquids in the reversed-phase liquid chromatographic separation of basic drugs. J Chromatogr A 2011;1218:398–407. [26] Garcı´a-Alvarez-Coque MC, Ruiz-Angel MJ, Berthod A, Carda-Broch S. On the use of ionic liquids as mobile phase additives in high-performance liquid chromatography (review). Anal Chim Acta 2015;883:1–21. [27] Pino V, Afonso AM. Surface-bonded ionic liquid stationary phases in high-performance liquid chromatography (review). Anal Chim Acta 2012;714:20–37. [28] Baeza-Baeza JJ, Ruiz-Angel MJ, Carda-Broch S, Garcı´a-Alvarez-Coque MC. Half-width plots, a simple tool to predict peak shape, reveal column kinetics and characterise chromatographic columns in liquid chromatography: state of the art and new results. J Chromatogr A 2013;1314:142–53. [29] Berthod A, Garcı´a-Alvarez-Coque MC. Micellar liquid chromatography. New York: Marcel Dekker; 2000. [30] Ruiz-Angel MJ, Garcı´a-Alvarez-Coque MC, Berthod A. New insights and recent developments in micellar liquid chromatography (review). Sep Purif Rev 2009;38:45–96.

References

[31] Garcı´a-Alvarez-Coque MC, Ruiz-Angel MJ, Carda-Broch S. Micellar liquid chromatography: fundamentals. In: Anderson JL, Stalcup A, Berthod A, Pino V, editors. Analytical separation science series, Vol. 2. New York: Wiley-VCH; 2015. p. 371–406. [32] Escuder-Gilabert L, Sagrado S, Villanueva-Caman˜as RM, Medina-Herna´ndez MJ. Quantitative retention-structure and retention-activity relationship studies of local anesthetics by micellar liquid chromatography. Anal Chem 1998;70:28–34. [33] Baeza-Baeza JJ, Da´vila Y, Ferna´ndez-Navarro JJ, Garcı´a-Alvarez-Coque MC. Measurement of the elution strength and peak shape enhancement at increasing modifier concentration and temperature in RPLC. Anal Bioanal Chem 2012;404:2973–84. [34] Ruiz-Angel MJ, Peris-Garcı´a E, Garcı´a-Alvarez-Coque MC. Reversed-phase liquid chromatography with mixed micellar mobile phases of Brij-35 and sodium dodecyl sulphate: a method for the analysis of basic compounds. Green Chem 2015;17:3561–70. [35] Ruiz-Angel MJ, Carda-Broch S, Garcı´a-Alvarez-Coque MC. High submicellar liquid chromatography (review). Sep Purif Rev 2014;43:124–54. [36] Madamba LS, Strasters JK, Khaledi MG. Gradient elution in micellar liquid chromatography. II. Organic modifier gradients. J Chromatogr A 1994;683:335–45. [37] Nakao R, Schou M, Halldin C. Rapid metabolite analysis of positron emission tomography radioligands by direct plasma injection combining micellar cleanup with high submicellar liquid chromatography with radiometric detection. J Chromatogr A 2012;1266:76–83. [38] Navarro-Huerta JA, Vargas-Garcı´a AG, Torres-Lapasio´ JR, Garcı´a-Alvarez-Coque MC. Interpretive search of optimal isocratic and gradient separations in micellar liquid chromatography in extended organic solvent domains. J Chromatogr A 2020;1616, 460784. [39] Marsh A, Clark BJ, Altria KD. A review of the background, operating parameters and applications of microemulsion liquid chromatography (review). J Sep Sc 2005;28:2023–32. [40] Peris-Garcı´a E, Pankajkumar-Patel N, Ruiz-Angel MJ, Carda-Broch S, Garcı´a-AlvarezCoque MC. Oil-in-water microemulsion liquid chromatography (review). Sep Purif Rev 2020;49:89–111. [41] Pankajkumar-Patel N, Peris-Garcı´a E, Ruiz-Angel MJ, Carda-Broch S, Garcı´a-AlvarezCoque MC. Modulation of retention and selectivity in oil-in-water microemulsion liquid chromatography (review). J Chromatogr A 2019;1592:91–100. [42] Peng L-Q, Cao J, Du L-J, Zhang Q-D, Shi Y-T, Xu J-J. Analysis of phenolic acids by ionic liquid-in-water microemulsion liquid chromatography coupled with ultraviolet and electrochemical detector. J Chromatogr A 2017;1499:132–9. [43] Wang P, Lee HK. Recent applications of high-performance liquid chromatography to the analysis of metal complexes (review). J Chromatogr A 1997;789:437–51. [44] Sarzanini C. Liquid chromatography: a tool for the analysis of metal species (review). J Chromatogr A 1999;850:213–28. [45] Paull B, Haddad PR. Chelation ion chromatography of trace metal ions using metallochromic ligands (review). Trends Anal Chem 1999;18:107–14. [46] Shibukawa M, Saitoh K, Ozaki S, Nakajima H. Redox derivatization chromatography. Bunseki Kagaku 2013;62:985–1000.

143

CHAPTER

Ultrafast high-performance liquid chromatography

8

Tivadar Farkas1 and Bezhan Chankvetadze2 1

Phenomenex Inc., Torrance, CA, United States, 2Institute of Physical and Analytical Chemistry, School of Exact and Natural Sciences, Tbilisi State University, Tbilisi, Georgia

8.1 Introduction An early realization led to the belief that speed and performance are interdependent in chromatographic techniques in the sense that both speed and high performance cannot be achieved at the same time. Randau and Schnell expressed this belief in the following manner back in 1971: “Man hat sich also von Fall zu Fall zu entscheiden, ob man eine langsame Trennung mit hoher Aufl€osung oder eine schnelle Trennung mit geringerer Aufl€ osung haben will”; in English translation, this reads as follows: One has to decide from case to case whether to pursue a slow separation with high resolution or a fast separation with low resolution [1]. This view was based on the very basics of the chromatographic theory which translated into the following practical consequences: High-efficiency separations can only be achieved with long chromatographic columns, operated at their optimal mobile-phase linear velocity. Both using such long columns and operating them at their optimal velocity produce lengthy analyses, hardly conducive to both fast and efficient chromatographic analyses. Analytical scientists working in drug development, in clinical laboratories, and in other fields face challenging numbers of samples that need to be analyzed in a timely manner. This leads to a dire need for high-speed analyses that provide adequate performance. Preserving the resolving power of a separation system requires adequate sample preparation, the selection of highly efficient columns that preserve their resolving power when operated at high flow rates, and the use of HPLC instruments specifically designed (and properly maintained) for such level of performance. Ultrafast chromatography is also very useful for in-process analytics for monitoring and optimization of industrial-scale chemical processes. Various omics fields of research (such as metabolomics, proteomics, foodomics) also require fast analytical methods. Although sensors are widely used for the continuous monitoring of chemical processes, they have the disadvantage of lacking specificity due to cross-talk between various species present in the sample (including matrix components). Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00031-X Copyright # 2023 Elsevier Inc. All rights reserved.

145

146

CHAPTER 8 Ultrafast high-performance liquid chromatography

Chromatography, as a separation-based technique, is free of some of the disadvantages of using sensors (however, it cannot yet match sensors in speed). More than 20 years after Randau and Schnell mused over the apparent incompatibility between high-efficiency and high-speed separations, Kirkland announced a breakthrough in his paper entitled “HPLC Method Development: Practical Aspects of Increasing Analysis Speed While Maintaining Separation Resolution” [2]. He demonstrated that carefully configured columns made with 3.5-μm particles having a narrow particle-size distribution resist the plugging usually encountered with columns packed with traditional 3-μm particles and show considerable promise for developing fast, rugged, and reproducible methods. His results suggested that such columns permit typical separations in one-half the time without loss in resolution compared to commonly used columns [2]. While performing HPLC analyses in half the time was significant in 1993, little we knew how modest this improvement would turn out to be in light of advancements achieved 10–15 years later, but also how much faster analyses had to be to meet the needs of practitioners. Nevertheless, halving analysis time demonstrated that there was hope for achieving both highly efficient and fast HPLC analyses.

8.2 High-efficiency high-speed separations 8.2.1 Sub-2-μm fully porous packing materials The basics of chromatography teach that achieving high-performance separations requires good selectivity and high efficiency (i.e., high plate numbers associated with the column used). While good selectivity is a luxury unavailable in challenging separations, high plate numbers require the use of long columns commonly leading to long analysis times. A striking example of good performance achieved despite marginal selectivity is the use of a simulated exceptionally long column in the multicycle experiment performed for the separation of isotopomers, i.e., of enantiomers having only minimal structural differences, in that some hydrogen atoms are replaced by deuterium atom [3]. A racemic mixture of the R- and S-isomers of phenyl(phenyl-d5)-methanol was separated by HPLC using cellulose trisbenzoatecoated silica as stationary phase and a 2-propanol/hexane (5/95) mixture as mobile phase. The cellulose derivative showed limited preferential retention for (R)-()phenyl(phenyl-d5)-methanol over the (S)-(+)-isomer, with a minimal separation factor reported as only 1.0080. Still, baseline separation could be achieved after 65 cycles, with a virtual column length of 9.75 m which generated 350,000 plates in over 40 h. Fortunately, many separations are favored by adequate selectivity (much better than the marginal value of 1.0080 of the above example) and require about 25,000 plates in reasonable analysis time ( 15.5 are adequately eluted using an ACN gradient from 0 to 100%. Less polar solutes, going down to δX  8.5, are eluted by substituting ACN for THF. The polarity range of solutes properly eluted from a silica column with alkaneisopropanol mixtures in NPLC is shown in Figure 11.1B. As can be seen, it is approximately 11.5 < δX < 13.5, which falls within the range covered by RPLC. Therefore, all analytes eluted by NPLC can also be eluted with optimal retention factors using RPLC. However, this does not mean that NPLC and RPLC have the same or similar chromatographic value. Thus, hydrophobic samples as mineral and vegetable oils that can be directly injected into an NPLC system are not compatible with most mobile phases in RPLC. Furthermore, NPLC and RPLC can provide quite different values of selectivity and efficiency depending on the nature of the solutes. Finally, in HILIC, where solutes are retained in a layer of water (δS  23.5, Figure 11.1C), highly polar solutes in the 18 < δX < 21 range (mainly ions, polyions, or zwitterions) are eluted with water–ACN mixtures increasing the water content from 5% to 50%. However, a problem with HILIC is that samples and polar analytes must be soluble in the required organic-rich mobile phases, particularly at the beginning of the gradient.

11.4 Isoeluotropic mixtures Fine-tuning of polarity through discrete or continuous changes in mobile phase composition in the isocratic and gradient elution modes, respectively, is primarily achieved by adjusting the modifier concentration in the solvent mixture. Meanwhile, selectivity is controlled by changing the solvent nature, and for some solutes, also by modifying the mobile phase pH [14], or column temperature [15,16]. For ionic analytes, the concentration of an ion-pairing salt is also an important factor (see Chapters 7 and 12). Selectivity depends mainly on the specific interactions of solutes with the stationary and mobile phases [17,18], that is, on the profile of the contributions to the global polarity of solutes and phases.

257

258

CHAPTER 11 Mobile phase selection in liquid chromatography

(A)

5

C18

10

15

20

25

25

δS

RPLC 5

10

15

20

5

10

15

20

25

Water

δX

δM

THF ACN MeOH

(B) 5

10

15 Silica

20

25

δS

NPLC 5

10

15

20

25

5

10

15

20

25

Alkane CHCl3 isoPrOH

(C)

20 Water 25

5

10

15

5

10

15

20

25

5

10

15

20

25

δX

δM

δS

HILIC

ACN

δX

δM

5% water 50% water FIGURE 11.1 Graphic expression of the Schoenmakers’ rule. Within the limits of predictions based on the Hildebrand solubility parameter, global polarity range of solutes that are adequately eluted when a wide elution gradient is applied for: (A) RPLC with C18 and ACN-water; (B) NPLC with underivatized silica and isopropanol-heptane; and (C) HILIC with a layer of water and waterACN as mobile phase. The global Hildebrand polarities of the stationary phase, solute, and mobile phase are represented on the δS, δX, and δM scales, respectively.

A basic question in optimizing selectivity is how the character of a solvent mixture should be modified, without altering the selected elution strength. Mixtures with the same elution strength but prepared with different modifiers are called isoeluotropic mixtures. For binary mixtures of MeOH, ACN, or THF with water, using Eq. (11.3) and the Hildebrandt parameter as a measure of the global polarity, assuming linear behavior:

11.5 Solvent-selectivity triangles

ACN-water 0

10

20

30

40

50

60

70

80

90

100 MeOH-water

0

20

0

10

40 20

60 30

40

80 50

100 60

70

THF-water 80

90

100

FIGURE 11.2 Nomogram showing isoeluotropic binary mixtures in RPLC. The compositions are obtained by connecting the solvent scales with a vertical line. The example indicates that aqueous binary mixtures having 60% ACN, 70% MeOH, or 46% THF are isoeluotropic. Adapted from Sigma-Aldrich.com/Supelco 2009–2010 chromatography products catalog, p. 38.

δMeOH φMeOH + δH2O ð1  φMeOH Þ ¼ δACN φACN + δH2O ð1  φACN Þ ¼ δTHF φTHF + δH2O ð1  φTHF Þ

(11.7)

By substituting the polarity values given in Table 11.1, the following results are obtained: φMeOH ¼ 1:27 φACN ¼ 1:60 φTHF

(11.8)

Hence, the elution strength of an aqueous mobile phase with 20% MeOH is approximately the same as for 15.7% ACN or 12.5% THF. Since THF is the most hydrophobic solvent, the same elution strength is achieved with a lower percentage of organic solvent. As noted above, elution strength predictions deviate from linearity at high modifier concentrations. To address this problem, nonlinear relationships and nomograms, such as Figure 11.2, can be used. In this nomogram, all possible isoeluotropic binary mixtures consisting of water and ACN, MeOH, or THF can be estimated. Note that the ACN scale is linear, so it was necessary to draw nonlinear scales for MeOH and THF. ACN is generally stronger than MeOH, and THF is appreciably stronger than ACN. However, due to the inherent limitations of global polarity parameters, predictions are approximate and highly dependent on solute properties.

11.5 Solvent-selectivity triangles 11.5.1 The Snyder’s solvent-selectivity triangle Mobile phase selectivity can be understood by considering the profile of the contributions of solvent-solvent intermolecular interactions to the global polarity. Six types of interactions are included in the calculation of the Hildebrand solubility parameter [10]: between permanent dipoles, between induced dipoles, between permanent and induced dipoles, donation of hydrogen atoms (hydrogen-bond acidity), acceptance of hydrogen atoms (hydrogen-bond basicity), and electrostatic interactions. However, as discussed below, these are not the only possible interactions. Owing to the different contributions, if solutes with the same global polarity but

259

260

CHAPTER 11 Mobile phase selection in liquid chromatography

structural differences are separated by chromatography, retention times will be close but still different. We could add “fortunately different,” because otherwise the selectivity optimization would not be possible. To deal with more than three parameters, multivariate statistics is required, where the solvents in the multivariate space are projected onto the reduced space of the first principal components [2]. However, in the strategy proposed by Snyder in 1974 [6,19], electrostatic interactions were neglected, and some of the more akin interactions (between permanent and induced dipoles) were summarized in a single property called dipolarity (i.e., polarity and polarizability). Accordingly, the selectivity of the mobile phase was characterized by only three parameters: donation of hydrogen atoms, acceptance of hydrogen atoms, and dipolarity. This made it possible to plot the properties of the solvent on a triangular diagram, known as the Snyder’s solventselectivity triangle (SST), where each corner represents one of the properties (Figure 11.3) [20]. Solvent properties were estimated using three probes: ethanol (e), 1,4-dioxane (d), and nitromethane (n), which is a simplification of the six-probe system previously proposed by Rohrschneider to represent solvent properties. Using these three probes, the intended properties are: “hydrogen atom donor” (ethanol), “hydrogen atom acceptor” (1,4-dioxane), and “polar or polarizable” (nitromethane). In fact, none of the three probes uniquely represents these characteristics: Ethanol is predominantly a hydrogen atom donor but also a weak acceptor and is moderately dipolar; 1,4-dioxane is a good hydrogen atom acceptor, weakly dipolar, and a nonhydrogen atom donor; nitromethane is strongly dipolar but also both a weak hydrogen atom donor and acceptor. Although far from ideal, the selected probes led to a useful solvent classification. Solvents were classified according to their ability to interact with the three probes, which was estimated from gas–liquid partition constants. Snyder described a global polarity, P0 (Table 11.1), as the sum of the three contributions: P0 ¼ log k0e + log k0d + log k0n

ke0 ,

kd0 ,

(11.9)

kn0

where and are the gas–liquid partition constants of the probes, which were determined from their equilibrium concentrations in a sealed vial, containing a fixed volume of the solvent to be characterized. Each partition coefficient was defined as the relationship between the solute concentration in the solvent and in the vial void volume, after making two corrections to eliminate the effect of the solvent volume and the nonspecific contributions (CdH weak permanent dipole or induced interactions, obtained with n-octane). Finally, to eliminate the differences among the global polarities of the solvents, a normalization was carried out: 1¼

log k0e log k0m log k0n + + ¼ xe + xd + xn 0 0 P P P0

(11.10)

where xe represents the acceptor character, xd the donor character, and xn the dipolar character of the solvent (Table 11.2). Using this approach, the character of a solvent is defined by the balance or profile of the three normalized parameters, regardless of

11.5 Solvent-selectivity triangles

Proton acceptor 0.1

0.7 Triethylamine

0.2

0.6

isoPrOH

xe

MeOH

Diisopropyl ether

II

0.3

0.5 I

H2O HAcO

0.4

THF

DMF

xd

IV CHCl3

VIII

0.5

Proton donor 0.6 0.2

III

0.4

0.3

Ethyl acetate

VI ACN

0.3 V

CH2Cl2 VII

CS2

0.4

0.5

0.2

xn

0.6

Dipolar

FIGURE 11.3 Snyder’s solvent selectivity triangle, based on donation of hydrogen ions, acceptance of hydrogen ions, and dipolarity, indicating the eight solvent families (large circles). The location of several solvents is indicated, including those most commonly used in RPLC and NPLC (DMF, dimethylformamide; HAcO, acetic acid; isoPrOH, isopropanol). The arrows starting with chloroform illustrate how to read the scales.

Table 11.2 Normalized selectivity factors. Derived from gas–liquid partition data of Rohrschneider’s probesb

Derived from Kamlet–Taft solvatochromic parametersc

Solventa

xd

xe

xn

α

β

π*

Diisopropyl ether Hexane Carbon disulfide Triethylamine Carbon tetrachloride Ethyl acetate Toluene Tetrahydrofuran

0.10 –d 0.39 0.07 0.38 0.23 0.28 0.20

0.51 –d 0.22 0.61 0.30 0.34 0.25 0.38

0.39 –d 0.39 0.32 0.32 0.43 0.47 0.42

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.64 0.00 0.10 0.84 –d 0.45 0.17 0.49

0.36 0.00 0.90 0.16 0.59 0.55 0.83 0.51 Continued

261

262

CHAPTER 11 Mobile phase selection in liquid chromatography

Table 11.2 Normalized selectivity factors. Continued Derived from gas–liquid partition data of Rohrschneider’s probesb

Derived from Kamlet–Taft solvatochromic parametersc

Solventa

xd

xe

xn

α

β

π*

Chloroform Dichloromethane Methyl ethyl ketone Acetone Carbon disulfide 1,4-Dioxane Pyridine Isopropanol 1-Butanol 2-Methoxyethanol Dimethylformamide Ethanol Dimethylsulfoxide Acetonitrile 1-Propanol Acetic acid Methanol Formamide Water

0.35 0.33 0.22 0.23 0.39 0.24 0.22 0.19 0.19 0.24 0.21 0.19 0.27 0.27 0.19 0.31 0.22 0.33 0.37

0.31 0.27 0.35 0.35 0.22 0.36 0.41 0.55 0.59 0.38 0.39 0.52 0.35 0.31 0.54 0.39 0.48 0.38 0.37

0.34 0.40 0.43 0.42 0.39 0.40 0.36 0.27 0.25 0.38 0.40 0.29 0.38 0.42 0.27 0.30 0.31 0.30 0.25

0.43 0.20 –d 0.06 0.00 0.00 0.00 0.35 0.37 –d 0.00 0.39 0.00 0.15 0.36 0.54 0.43 0.33 0.43

0.00 0.09 –d 0.38 0.10 0.40 0.42 0.43 0.41 –d 0.44 0.36 0.43 0.25 0.40 0.15 0.29 0.21 0.18

0.57 0.78 –d 0.56 0.90 0.60 0.58 0.22 0.22 –d 0.56 0.25 0.57 0.60 0.24 0.31 0.28 0.46 0.45

a

Solvents ordered according to Table 11.1. Large values of xd, xe, and xn denote good hydrogen ion donor, good hydrogen ion acceptor, and large permanent or induced dipole moments, respectively [5–7]. c α, β, and π* represent solvent ability to interact as hydrogen ion donor, hydrogen ion acceptor, and by polar and polarization effects, respectively [21,22]. d Not available. b

the global polarity. Therefore, it is assumed that a solvent that preferably retains ethanol or 1,4-dioxane rather than nitromethane should have a predominant acceptor and donor character, respectively, and a solvent that preferably retains nitromethane rather than the other two probes has a polar character or is easily polarizable rather than being a hydrogen-bond donor or acceptor. The xe, xd, and xn data for several solvents are plotted on the SST (Figure 11.3). Solvents are grouped according to their properties into eight families: (I) aliphatic ethers and amines; (II) aliphatic alcohols; (III) pyridine and THF; (IV) glycols and acetic acid; (V) dichloromethane and dichloroethane; (VI) aliphatic ketones, esters, 1,4-dioxane, and nitriles; (VII) aromatic hydrocarbons and nitrocompounds; and (VIII) phenols and water. The scales on the triangle sides should be read counterclockwise: xe is represented on the right side (the higher on the scale, the stronger

11.5 Solvent-selectivity triangles

the acceptor character of the solvent), xd is on the left side (the lower on the scale, the stronger the donor character), and xn is on its base (with the solvent dipolarity increasing to the right). The representation shows that the most common solvents in RPLC provide different selectivity, as they have quite different profiles for the three properties defined in the SST. Thus, water is a strong hydrogen-bond donor and acceptor (it is located at half-height in the SST), but a weak dipole (it is on the left). ACN is a weaker hydrogen-bond donor than water, but significantly more dipolar. MeOH is an appreciably stronger hydrogen-bond donor (higher in the diagram), it is more dipolar than water, and less dipolar than ACN. Finally, THF has both hydrogen-bond donor and acceptor character, but it is more dipolar than water. The SST scales should not be interpreted as “percentages” of the intended properties, since these were obtained from solutes with mixed character, and therefore, the vertices do not represent “pure” properties. For example, a strong acceptor solvent such as triethylamine is not near the top vertex due to its acceptor capacity, but because it strongly retains ethanol and weakly retains 1,4-dioxane and nitromethane. Ideally, if the SST scales corresponded to pure properties (each vertex representing a 100% hydrogen-bond donor, a 100% hydrogen-bond acceptor, and a 100% dipolar solvent), mixtures of three hypothetical solvents, each located at each vertex, would provide a whole universe of possibilities. However, these solvents do not exist. Furthermore, the actual solvents located near the vertices of the SST are not miscible with each other or are not compatible with common stationary phases. ACN, MeOH, and THF are in intermediate locations in the SST, being excellent options to achieve a wide range of properties in RPLC. Not surprisingly, these solvents were already popular when the SST was developed.

11.5.2 Prediction of the character of solvent mixtures The SST makes it possible to predict whether the elution strength will increase or decrease for certain solutes when one modifier is replaced by another. For example, replacing a MeOH-water mixture with an isoeluotropic ACN-water mixture will reduce the ability of the mobile phase to accept hydrogen atoms; so, the elution strength will be reduced for solutes which are hydrogen-bond donors. Simultaneously, the dipolar character of the mobile phase will increase so that dipolar and polarizable compounds will elute earlier. This reasoning can help in the identification of solutes. As shown in the SST of Figure 11.4, the character of all possible mixtures of water, ACN, MeOH, and THF is delimited by straight lines connecting the four solvents. This illustrates how wide the selectivity range in RPLC is. The character of isoeluotropic mixtures of the four solvents, at increasing elution strength, is indicated by the three small triangles a, b, and c. Their location on the SST was established according to the compositions obtained from the nomogram of Figure 11.2. A linear variation of the properties with the concentration of the modifier was also assumed. The small triangles a, b, and c in Figure 11.4 illustrate how the character of a mixture

263

264

CHAPTER 11 Mobile phase selection in liquid chromatography

0.2

0.6

Proton acceptor 100% MeOH 0.3

xd 0.4

0.5

xe

39% H2O

21% 46% 72%

a

30%

Proton donor

0.5

70%

b

100% THF

0.4

c

60% 100% ACN 0.3

0.4

xn

0.5

Dipolar

0.3

FIGURE 11.4 Snyder’s solvent-selectivity triangle indicating the character of water, ACN, MeOH, and THF mixtures. The small triangles a, b, and c describe isoeluotropic mixtures with increasing elution strength. In a, the lowest vertex corresponds to 30:70 ACN-water, the upper vertex to 39:61 MeOH-water, and the left vertex to 21:79 THF-water. Other points on the sides of the small triangle a correspond to ternary mixtures, and the points inscribed in triangle a correspond to quaternary mixtures. Similarly, the small triangles b and c correspond to isoeluotropic mixtures with respect to 60:40 ACN-water and 100% ACN, respectively.

of solvents changes as its composition varies, while maintaining a constant elution strength, as estimated by the Hildebrand solubility parameter, δ.

11.5.3 A solvatochromic triangle of solvent selectivity The essential conclusion of Snyder’s SST and alternative diagrams based on solvatochromic properties is that to explore the full range of possibilities during optimization of mobile phase selectivity, solvents having both mutual miscibility and, at the same time, maximal differences in their properties should be selected. Another application of the diagrams is the visualization of the possibility of substituting a solvent by an equivalent one with improved non-chromatographic characteristics, such as price, availability, or better compliance with the principles of green chemistry (see Section 11.7). Lastly, the diagrams are also useful for predicting solvent miscibility and solute solubility in solvents with similar properties. Consistent with the “mixed” character of the probes used to construct the SST, xe in fact reflects a composite of hydrogen-bonding acceptor and donor, and dipolarity; xd reflects a composite of solvent donor and dipolarity; and xn reflects predominantly the dipolarity of the solvent with small contributions from hydrogen-bonding

11.5 Solvent-selectivity triangles

acceptor and donor. In 1989, Rutan and Carr [7,20,23] replaced the gas–liquid partition constants obtained for the Rohrschneider’s probes with the Kamlet–Taft “solvatochromic parameters” (Table 11.2). These parameters, derived mainly from spectroscopic measurements, separately estimate the strength of the hydrogen-bond donor (α), hydrogen-bond acceptor (β), and dipolarity/polarizability (π*) properties of solvents as contributors to the global polarity of the solvent. The solvatochromic parameters are averages over the results obtained for several probes. Therefore, it is normally assumed that they provide more exact measurements of the addressed properties than gas–liquid partition constants derived from just three probes (as in Snyder’s approach). However, the reconstruction of the SST using normalized solvatochromic parameters was rather disappointing, as many solvents were placed on a line joining the hydrogen-bonding acceptor and dipolar tops of the triangle, and consequently, solvent discrimination was rather poor [20].

11.5.4 Other solvent descriptors and alternative diagrams for classification and comparison of solvents An alternative to the Snyder descriptors and Kamlet–Taft solvatochromic parameters is the Hansen parameters [24,25]. These are derived from the Hildebrand solubility parameter, which is divided into three contributions: δ2 ¼ δd 2 + δp 2 + δh 2

(11.11)

representing dispersive forces (δd), polarity (δp), and hydrogen bonding (δh) (both donor and acceptor). Using the Hansen parameters, an alternative SST to Snyder’s, which also shows a good dispersion of solvents, was constructed. A somewhat more complex but widely accepted solvent classification system is that based on the five linear solvation energy relationships (LSERs) or Abraham descriptors [26–32]. The solvation parameter model describes five interactions through five descriptors related to compound properties: E (excess molar refraction, related to the presence of n- and π-electrons that result in charge transfer, π–π interactions and dipole-induced dipole interactions); S (representing the presence of dipoles and polarizability); A and B (describing donor and acceptor hydrogen bonding, respectively); and V (McGowan’s characteristic volume, related to dispersive interaction and energy for cavity formation). Representation procedures other than triangles should be used to deal with five descriptors. One possibility is to use projections after rotation of a principal component. However, when using principal components, the chemical significance of the axes is lost. An alternative is the use of spider diagrams [33], like the one shown in Figure 11.5. With this representation, several parameters more than three can be projected onto a plane with little loss of information. Careful selection of the order of the axes is essential to minimize the loss of information due to the reduction of the number of dimensions. Therefore, the descriptors that are more positively correlated (e.g., E and S for the LSER descriptors) should be juxtaposed, as opposed to those that are negatively correlated, while the least correlated should be orthogonal. However, as in any other projection

265

266

CHAPTER 11 Mobile phase selection in liquid chromatography

FIGURE 11.5 Spider diagram based on the Abraham descriptors E, S, A, B, V. The size of the points is proportional to the V/U ratio. Reproduced with permission from Lessellier [33].

technique, a compensation of the descriptors will cause quite different solvents be found in close positions on the spider diagram. In Figure 11.5, water is at the lower right, showing its high hydrogen-bond donor strength (A is large) and its weak hydrophobicity (V is low). Alcohols, acetic acid, and formamide are found near water. Nitriles (like ACN) show higher dipole-type interactions and are located on the right side of the diagram, above alcohols. Alkanes, with high hydrophobicity, are on the opposite side of the figure, on the left, near the V axis. Aromatic solvents are at the top of the diagram, around the E axis. THF, 1,4-dioxane, acetone, and ethyl acetate belong to the same group, in the center of the diagram.

11.6 Practical guidelines for the optimization

The Abraham descriptors are useful to explain the differences in selectivity between the three most used solvents in RPLC. Thus, MeOH is the best hydrogen-bond donor and acceptor, ACN has the highest value for dipolar interactions, and THF with the highest McGowan’s volume favors the solubility of most organic compounds through dispersive interactions, which explains its high elution strength in RPLC. Finally, Abraham’s descriptors also provide a useful global polarity scale defined as  1=2 U X ¼ E2X + S2X + A2X + B2X + V 2X

(11.12)

where the equation is written for a given solute, X. This global parameter can be used to estimate the elution strength of solvent mixtures, as done previously by Eq. (11.3) using the Hildebrand parameter. In Figure 11.5, the size of the symbol representing each solvent was made proportional to V/U.

11.6 Practical guidelines for the optimization of mobile phase composition 11.6.1 Selecting the chromatographic mode The optimization of the modifier type and its volume fraction in the mobile phase are often done on a trial-and-error basis. In this section, some guidelines to streamline and accelerate this process are given. After selecting the chromatographic mode (e.g., RPLC, NPLC, or HILIC), and deciding between isocratic or gradient elution, the elution strength should be adjusted, and finally, the selectivity optimized until all peak pairs for the target analytes are resolved. To select the chromatographic mode, two characteristics are attended as follows: 1. Solute nature. If solute molecules contain extensive hydrophobic regions in “external” structural parts, they will be retained in the hydrophobic RPLC stationary phases. In contrast, if the influence of ionic or polar groups predominates (e.g., dCOOH, dOH, or dNH2), the solute will experience poor retention and require polar stationary phases typical of NPLC. A good solution to increase the retention of ionic compounds is ion pairing [34]. In this technique, a salt is added to the mobile phase. Retention is enhanced by mixed mechanisms involving association of ions of opposite charge in the hydro-organic mobile phase, and by ion exchange on the surface of the stationary phase, where the added salt is adsorbed. Since ions and other highly polar solutes are not compatible with NPLC mobile phases, HILIC provides an alternative. However, a frequent limitation in HILIC is the poor solubility of ionic compounds in mobile phases rich in organic solvent that are required. 2. Sample compatibility with the mobile phase. Direct injection of samples soluble in water or in hydro-organic mixtures (e.g., serum, urine, and other aqueous samples or aqueous extracts) is possible by RPLC. Meanwhile, for these samples,

267

268

CHAPTER 11 Mobile phase selection in liquid chromatography

if HILIC is selected, the elution strength should be decreased by evaporation of water in the sample, followed by redissolution in a mixture rich in ACN, or by dilution with ACN at the expense of a poorer limit of detection. For hydrophobic samples (oils, fats, hydrocarbons, or extracts in heptane, dichloromethane, or other hydrophobic solvents), NPLC is needed. Extracts in solvents that provide high elution strength, such as ethyl acetate in NPLC, or isopropanol in both RPLC and NPLC, should be avoided. It is often possible to change the solvent initially used to extract the sample. For example, an aqueous sample can be extracted with heptane or dichloromethane, a vegetable oil can be extracted with an aqueous buffer or MeOH, and the compounds of interest in an environmental aqueous sample can be concentrated on a solid phase, followed by elution with an appropriate solvent. Within the limits of the analytes’ solubility or stability, it is possible to change the solvent character by evaporation and dilution to make the medium compatible with a given chromatographic mode.

11.6.2 Description of retention using the modifier content as a factor Solute retention is mostly controlled by the modifier concentration in the mobile phase. In this case, to predict optimal chromatographic conditions, it is useful to know the retention behavior as the amount of modifier varies [35]. In RPLC, the retention for a solute X can be expressed in terms of the solubility parameters according to [36]: ln kX ¼

i νX h nS ðδM  δX Þ2  ðδS  δX Þ2 + ln RT nM

(11.13)

where R is the gas constant, T the absolute temperature, νX the solute molar volume, and nM and nS the moles of mobile phase and stationary phase in the column, respectively. For a binary mixture of water (w) and organic solvent (o), the polarity of the mobile phase can be calculated as a function of the volume fraction of the modifier. Substituting Eq. (11.13) in Eq. (11.3), for binary mixtures, a general-purpose equation is obtained, which is commonly used to characterize retention in RPLC [37]: log k ¼ c0 + c1 φ + c2 φ2

(11.14)

The quadratic relationship can be simplified to a linear model (Eq. 11.15), called Linear Solvent Strength (LSS) model. log k ¼ c0 + c1 φ ¼ log kw + Sφ

(11.15)

where kw is the retention factor in the absence of modifier (i.e., in water), and S, the elution strength parameter, which is constant. However, the linear model only describes retention when it is governed by a partition mechanism. In practice, it usually works only for small ranges of modifier.

11.6 Practical guidelines for the optimization

Surface adsorption in NPLC is best described by empirical nonlogarithmic and logarithmic models [38]: 1 ¼ ðc0 + c1 φÞn k

(11.16)

log k ¼ c0 + c1 logφ

(11.17)

where φ is again the concentration of the stronger solvent (here the most polar) in a binary mobile phase. Eq. (11.16) is also suitable for RPLC (where φ would be the less polar solvent). The retention behavior in HILIC is more complex and is described by equations combining partition and adsorption phenomena [35,39,40] as follows: log k ¼ c0 + c1 φ + c2 log φ

(11.18)

where φ here is the fraction of water in the mobile phase. The correlation of the parameters in the fitted models is used to reveal the similarities in the retention mechanisms for solutes analyzed with the same column [5,41]. A good correlation means that all compounds follow a similar mechanism (e.g., partition with Eq. (11.15)). When the correlation shows significant scattering, a combined retention mechanism should exist. Recently, a global retention model has been derived based on Eq. (11.15): dk ¼ Sk dφ

(11.19)

where the dependence of the elution strength with the modifier concentration is assumed to be linear [42]. For a non-linear dependence, Eq. (11.19) should be adapted as follows: dk ¼ Sg kg dφ

(11.20)

where Sg is a constant parameter. The integration of Eq. (11.20) yields:  1=ð1gÞ kw k ¼ kw 1 + ðg  1Þkg1 ¼ 1=ðg1Þ w Sg φ g1 1 + ðg  1Þkw Sg φ

(11.21)

Eqs. (11.20) and (11.21) include the parameter g that describes the variation of the elution strength with φ. This parameter is the elution degree and characterizes the profile of the elution strength as the modifier concentration changes. When g tends to 1, Eq. (11.21) will reproduce the linear behavior between log k and φ in the LSS model. However, mixed retention mechanisms are usual in both HILIC and RPLC, which gives rise to curvilinear changes of the elution strength with the modifier content. For RPLC, where the partition mechanism is dominant, g will be closer to 1, while for HILIC with a combination of retention mechanisms, g is usually closer to 2. In both cases, the elution strength decreases with an increase in the modifier concentration.

269

270

CHAPTER 11 Mobile phase selection in liquid chromatography

11.6.3 Systematic trial-and-error optimization of mobile phase composition in isocratic elution Isocratic elution can be selected if the polarity of the compounds in the sample is similar. If the polarity spans a wide range, then gradient elution is needed. For an unknown problem, it is preferable to start any optimization in the gradient elution mode. However, in this section, we focus first on the simpler development of an isocratic method. Typically, in RPLC, a C18 stationary phase is tested first. If no prior information on solute polarity is available, it is advisable to start with a mobile phase with a high elution strength, such as 95% (v/v) ACN. This ensures elution of most compounds in the sample, although many can elute close to the column hold-up time. If the retention of one or more solutes is still too high (k > 20), NPLC is probably preferable. Other options are to change the C18 column to C8 or C4 or use a higher column temperature. The retention of solutes eluting close to the column hold-up time should be increased using progressively smaller concentrations of the modifier (e.g., 60%, 40%, and 20%). At this stage, gradient elution is likely necessary if the solutes of interest cannot be migrated into the target range of k values, with the modifier concentrations tested. An analogous strategy can be followed for NPLC: Initially, a polar column (e.g., bare silica or cyanopropyl-silica) and a mobile phase of high elution strength are selected. However, be aware that a few percent change of a polar modifier in the mobile phase can cause dramatic changes in retention. For instance, a smaller increase in retention can be produced by decreasing the ethyl acetate concentration from 40% to 2% than from 2% to 0%. This is because in NPLC, the “strong” solvent which mainly determines the solvating properties of the mobile phase is the modifier. Therefore, in NPLC with moderate concentrations of modifier, most solutes probably elute close to the column hold-up time. In the absence of excessively retained solutes, the elution strength should be progressively reduced by decreasing the amount of modifier until appropriate retention times are reached. Similarly, for HILIC, aqueous mixtures containing up to 50% water can be tested initially, followed by gradually reducing the water concentration. Selectivity can be further optimized to improve resolution for all peak pairs. For this, solvent mixtures of similar elution strength, another pH or column temperature, or if necessary, a different stationary phase, can be tested. Here, we will discuss the selection of an isoeluotropic mixture. This can be based on the properties of the solute guided by the polarity scales described earlier, with the help of any of the classification diagrams derived from them. For example, in the RPLC elution of two solutes with the same retention but different hydrogen-bond donor character, the stronger donor elutes earlier if ACN is replaced by MeOH. However, the properties of solutes are often not known or the interpretation of possible solute-solvent interactions in multi-functional solutes is not straightforward. Therefore, selectivity is most often optimized in an empirical fashion.

11.6 Practical guidelines for the optimization

Following an empirical experimental scheme, in RPLC, the first modifier to be tested is usually ACN, due to its low viscosity and short ultraviolet (UV) cutoff wavelength (190 nm) (Table 11.1), which allow low backpressure and a UV detection window capable of detecting most compounds. If the separation is not satisfactory, the second option is MeOH. The viscosity of the MeOH–water mixtures is higher compared to ACN–water mixtures, with a maximum at 40% (v/v) MeOH, which owing to high backpressure is unsuitable for working at high flow rates with long packed columns, or with small particle sizes. The cutoff wavelength of MeOH is also longer (205 nm). The third option, THF, has even higher viscosity, a cutoff wavelength of 212 nm, and requires long equilibration times. Therefore, it is not surprising these solvents are always tested in the same order: ACN, MeOH, and THF. This is indicated by vertices A–B–C in the method development triangle (Figure 11.6). If one of the three isoeluotropic mixtures is successful, the problem is over. If some peaks remain unresolved, ternary or even quaternary isoeluotropic mixtures may be tested. For this purpose, the order of the D–G mixtures in Figure 11.6 is generally followed. After selecting the optimal isoeluotropic mixture, its composition can be slightly changed until all peaks of interest are satisfactorily resolved. Consider a 70:30 ACN–water mixture, for which all peaks in each sample are in the target range of k values. If the resolution between some peak pairs is unsatisfactory, following the scheme in Figure 11.6 and the nomogram in Figure 11.2, the mobile phase to be tested next is 78:22 MeOH-water (point B in Figure 11.6). If necessary, we continue with 52:48 THF-water (point C), 35:39:26 ACN-MeOH-water (point D),

A

D

F G

B

E

C

FIGURE 11.6 Method development triangle. A, B, and C represent isoeluotropic binary mixtures of water with ACN, MeOH, and THF, respectively; D–F are isoeluotropic ternary mixtures (e.g., point D is an ACN–MeOH–water mixture, where half of the first modifier has been replaced by an isoeluotropic amount of the second modifier). The central point G is the isoeluotropic quaternary mixture ACN-MeOH-THF-water, where two-thirds of the first modifier have been replaced by isoeluotropic amounts of the other two modifiers.

271

272

CHAPTER 11 Mobile phase selection in liquid chromatography

39:26:35 MeOH-THF-water (point E), and so on. Mixtures D and E were calculated by substituting half the ACN content of mixture A with its equivalent amount of MeOH or THF, respectively. This trial-and-error method is more common in practice than the use of polarity descriptors, owing to its simplicity, and because it does not require knowledge of the properties of the solute. However, when the problem remains unresolved, polarity descriptors or computer-assisted interpretive optimization is helpful (see Section 11.6.5). Similarly, optimization of selectivity in NPLC and HILIC can be conveniently carried out by systematically substituting the modifier for other miscible solvents exhibiting a different profile of its descriptors, thus placing itself in a different location on any SST or selectivity spider diagram.

11.6.4 Systematic trial-and-error optimization of mobile phase composition in gradient elution For samples with solutes that cover a wide range of polarity, a gradient of elution strength is required to obtain adequate retention of the first peaks in the chromatogram and progressively accelerate the elution of the more retained solutes. For this, at least two solvent mixtures with different elution strength must be combined (mixtures A and B, with B stronger). The gradient normally starts at the time of sample injection, although total control over the actual conditions of the gradient is lost if the delay time (dwell time) or time required for the mobile phase to reach the column from the gradient mixer is not considered. During the gradient time, tG (the time the gradient runs), the flow of B and A rise and fall, respectively, keeping the sum of the two flows constant, until only B is pumped. To reduce the noise from the baseline due to fluctuations in the mixture composition, which can be large for quaternary pumps, mixtures A and B containing at least 5% of the minor solvent should be used. In gradient elution starting with mobile phases with low elution strength, strongly retained analytes migrate very slowly, so that this range of mobile phase compositions does not contribute significantly to their elution. As the elution strength increases during the gradient, the analytes “accelerate” through the column. A graphical image of the effect is described by: “a solute sits at the head of a column until a solvent strong enough comes along to migrate it through the column, leaving the other solutes behind, then travels to the column outlet fairly quickly” [43]. The point at which this occurs depends on the strength of the interactions of the solute with the mobile and stationary phases. Therefore, solutes analyzed in RPLC using gradient elution rarely experience the full range of mobile phase composition. The fraction of the solvent composition range that affects solute migration has been termed “significant solvent concentration range” [44]. Thus, in addition to the chromatographic separation mechanisms, gradient elution also functions as a fractional extraction, causing the analytes to progress through the column as they are extracted from the stationary phase. For the first run of an unknown sample, a wide gradient with a small slope is recommended to ensure elution of all solutes (e.g., in RPLC, 5% to 100% ACN).

11.6 Practical guidelines for the optimization

The Δt/tG ratio, Δt being the difference between the retention times of the first and last peaks of interest in the chromatogram, provides a criterion for deciding whether the sample can be separated isocratically or whether gradient elution is required. If Δt/tG < 0.25, the sample can be eluted isocratically within the target retention region using a mobile phase composition close to the midpoint of Δt. In contrast, Δt/ tG > 0.25 means that solutes elute over a wide retention range, and isocratic elution is not practical. In this case, the new gradient must be focused between the mobile phase composition at the time of the first eluted peak (start of Δt; new mixture A) and the time of the last peak (end of Δt, new phase B). If the sample contains other components that are more retained than the analytes, then a final gradient step with a high elution strength is required to clean the column. This will avoid cross-contamination between successive injections. If some peak pairs remain unresolved, the composition of mixtures A and B should be modified without significantly altering their respective elution strengths. In RPLC, this can be achieved by substituting ACN for MeOH or THF, or using isoeluotropic ternary or quaternary mixtures, as discussed for isocratic elution. When all solutes are resolved satisfactorily, the gradient time can be further reduced without losing resolution. The simplest way is to increase the slope of the gradient to as much as the resolution of the least resolved peak pair will tolerate. Another option is to use a segmented or multi-linear gradient, that is, a gradient whose slope changes according to the distribution of peaks: the slope is shallow in regions with poorly resolved peaks and steeper in regions without peaks. Nonlinear gradients with concave or convex profiles are also occasionally applied when dealing with multi-component samples that require additional resolution. Gradients often include isocratic periods at the beginning and/or end of runs, or inserted between linear or nonlinear gradient segments. Inverse gradients (with decreasing modifier concentration) can be useful in some cases (e.g., to elute amphiphilic analytes whose solubility increases by increasing both the polar and less polar component of the mobile phase). In addition to the elution strength gradients, it is possible to establish selectivity gradients by increasing the hydrogen-bond donor/acceptor or dipolarity of the mobile phase, or any other polarity descriptor, either at constant or increasing global elution strength. Therefore, in principle, there are four possibilities: 1. Isocratic isoselective elution, where the composition of the mobile phase is constant. 2. Isocratic elution with a selectivity gradient, obtained by modifying the solvent mixture in such a way that the polarity descriptors, for example hydrogen-bond donor/acceptor or dipolarity, are varied while a global polarity descriptor remains unchanged. This implies the continuous modification of the coordinates of the mixtures used in an SST or a selectivity spider diagram, with the restriction of not modifying δX (Hildebrand solubility) or UX (Abraham global polarity, see Eq. (11.12)). For example, in Snyder’s SST, a selectivity gradient is obtained by following any line along the sides of the small triangles a,

273

274

CHAPTER 11 Mobile phase selection in liquid chromatography

b, or c in Figure 11.4 that correspond to isoeluotropic mixtures. Obviously, any translation along the surface of the triangle implies a change in selectivity. 3. Elution in an isoselective gradient, where the elution strength increases but the selectivity is not modified. Isoselective gradients are implemented using mixtures A and B with the same profile of normalized polarity descriptors (e.g., corresponding to the same point on a given selectivity diagram), but where solvent mixture B has a higher global polarity than solvent mixture A. Then, as the B/A ratio increases, the global polarity of the mixture increases but without a substantial modification of the selectivity. 4. Double-gradient elution, where both elution strength and selectivity are modified. These are the most common gradients: When the content of ACN or MeOH is increased in a mixture with water throughout a gradient, not only does the global elution strength increase, but also the polarity descriptors are varied, making the coordinates at any SST or selectivity spider diagram also to change. Dual RPLC gradients can be programmed by progressively decreasing the water flow, while simultaneously increasing the flow for one or more modifiers at different rates. In this way, the elution strength is increased, and, simultaneously, the selectivity is continuously modified in the desired direction (higher hydrogenbond donation and acceptance, dipolarity, etc.).

11.6.5 Computer-assisted interpretive optimization Despite being particularly slow and inefficient, the trial-and-error strategies discussed above still prevail. However, many mixtures are so complex that the protocol can be too long, and often the best (or at least acceptable) conditions are not found. Fortunately, the best conditions can be selected using computer-assisted interpretive strategies, which accelerate method development and allow more reliable results [45–49]. They also make a comprehensive optimization possible, which is essential when dealing with complex samples. An interpretive optimization includes two steps: modeling of the system using experimental data and predicting separations by computer simulation. In the first step, to fit equations (as those in Section 11.6.2) or train algorithms to predict analyte retention, a series of experiments are carried out as small and informative as possible [50]. In addition to retention times, other properties that summarize a chromatogram are also inferred from the experiments, such as peak width and skewness. The objective is to develop models capable of predicting the separation in any new arbitrary condition [51]. Then, based on the fitted models, the separation quality is predicted for many separation conditions, to find the one that gives the maximum (or at least appropriate) resolution of all peak pairs. In practice, this is done by simulating the separation of the sample within a preset factorial space and calculating a numerical value that scores the chromatograms, ideally in accordance with the analyst’s resolution assessment. In addition to resolution, properties such as separation time, minimal solvent consumption, or desirable peak profiles (i.e., high efficiency and low skewness) can be optimized.

11.6 Practical guidelines for the optimization

The reliability of the entire process depends on the quality of the experimental design used to build (train) the retention models, which is desired to give access to the best conditions with minimal effort and maximum accuracy. Ideally, the training design should be common for a group of analytes with a reasonably small number of elution conditions. The experiments should be carefully planned so that they are as informative as possible and cover the entire factorial space, to fit the models that describe the retention behavior for each analyte. The search of the best design is more complex using gradients compared to isocratic elution, due to the number of variables and profiles involved (gradients with one or more isocratic steps, linear or multi-linear, among others). Experimental designs based on isocratic experiments provide richer information on analyte retention than those for gradient runs. Therefore, they give the best prediction of performance. A small isocratic design (resulting in at least one degree of freedom) are satisfactory, provided the domain is well covered by the design. Isocratic designs are, however, limited by excessive retention times for the most hydrophobic analytes in a mixture, at the lowest modifier contents. In contrast, gradient designs have the advantage of sampling the region of lowest elution strength in a sufficiently short time. Therefore, many analysts prefer gradient elution, not only for routine work, but also for modeling purposes. A drawback is that each gradient only examines narrow windows of the solvent domain, then good predictions for isocratic conditions are only possible if the analyte elution is well sampled by the set of gradients, which is not always the case. Meanwhile, in training designs consisting solely of isocratic experiments, all mobile phases fully participate in solute migration. Finally, multi-linear gradients improve the quality of predictions with respect to designs composed of simple linear gradients. However, multi-linear gradient designs are more likely to depend on the level of adaptation to the information requirements for particular solutes. Some recommendations for constructing experimental gradient designs are as follows [50]: (i) Each composition along the gradient must participate significantly in solute retention, and the composition range must be as wide as possible. (ii) The design should not leave solvent concentrations unsampled. This will compromise the prediction accuracy, as predictions will require extrapolation. (iii) Isocratic segments before the gradient begins, or gradients with multiple initial levels, are alternatives to isocratic designs. To aid interpretive optimization, several software packages have been commercialized, such as DryLab [52], ChromSword [53], Osiris [54], PREOPT-W [55], and MICHROM [56]. Users can also develop their own software with the help of a spreadsheet or a highly efficient programming environment, such as MATLAB or R. More information on computer-assisted method development can be found in Chapters 13, 15, and 30.

275

276

CHAPTER 11 Mobile phase selection in liquid chromatography

11.6.6 Use of combined mobile phases or gradients to achieve complete resolution Conventional LC presents significant challenges for the analysis of complex samples. When a separation fails, the usual choice is to introduce a drastic change in the chromatographic system (column, solvent, pH, temperature, and/or use of additives). However, the possibilities of LC can also be extended by other strategies that combine mobile phases or gradients. Thus, the use of one or more pulses of a weak eluent (e.g., 200 μL of water or 500 μL of buffer solution in an RPLC system), inserted strategically to alter abruptly the local mobile phase composition, can improve resolution between poorly resolved peaks but with little or no effect on the neighboring peaks already resolved [57]. This is practical when full resolution has been achieved for most analytes. Another approach, called solvent modulation, introduces individual solvent zones of constant composition (usually two, A and B, such as 90% and 100% MeOH, or 75% MeOH and 60% ACN), in variable or repeated sequence [58]. The applied sequence is established by the ratio of the length of the solvent zones A and B within a cycle, and the number of cycles carried out throughout the elution. Because the solvent zones are spatially and temporally separated from each other, the nonideal solvent-solvent interactions are effectively eliminated, and the overall retention is just a linear combination of the retention times in the individual solvent zones. The advantage is that the effect on the chromatogram of changing the length of the zones is easy and accurately predicted. This approach has also been applied in gradient elution, in the socalled relay gradients, which is a special type of segmented gradients where the nature of the modifiers changes abruptly between segments. On the other hand, it is not uncommon to analyze a sample using two different columns or the same column, using two different isocratic or gradient conditions, to separate different target analytes. The possibilities of this approach can be fully exploited if the two solvent systems are complementary [59]: A separation condition focuses on the resolution of some compounds in the sample, while the other analytes remain unresolved, but they are optimally resolved in a second (or subsequent) condition(s). When the results of the optimal complementary separation conditions are considered together, all analytes are adequately resolved. The approach using parallel columns can involve different separation modes, such as RPLC and HILIC, to deal with samples containing compounds in a wide polarity range. However, for high-throughput analysis, performing separate chromatographic runs with different columns is not practical, hence the interest in coupling RPLC and HILIC columns. However, although both modes use the same solvents, diametrically opposite concentrations are needed: HILIC needs a high organic solvent content, while RPLC needs a large amount of water. The solvent strength incompatibility between RPLC and HILIC is solved by increasing the ACN content in the RPLC column eluate (intended to separate low-polarity solutes) by online mixing with ACN to meet the solvent requirements of the HILIC column (intended to separate high-polarity solutes) [60]. Another option is the direct

11.7 Additional considerations for the selection of solvents

connection of RPLC and HILIC, using a unique gradient program that starts with a high organic solvent content compatible with both RPLC and HILIC [61]. The more sophisticated configurations connect the two columns through valves that allow operation with different solvent systems in a two-dimensional (2D) fashion [62]. The principle of operation is to perform the offline or online transfer of specific fractions of the eluent from the outlet of the first column (representing the first dimension) to the inlet of the second column (the second dimension). In comprehensive 2DLC separation (LC  LC), the whole first-dimension eluate is chopped into small segments that are continuously separated in the second dimension. Instead, in heart-cutting 2DLC (LC–LC), only selected segments of the eluate from the first dimension, presumably those containing unresolved target analytes, are transferred to the second dimension for further separation. This is technically simpler than LC  LC, since the segments can be parked for a time at the head of the column or different columns, until the system is ready to proceed with the elution in the second dimension. Optimization of the elution conditions and data handling is also much simpler in LC–LC than in LC  LC. For both approaches, the advantage of exploiting different retention mechanisms, and the freedom to independently manipulate the mobile phase gradient in each column, produces a considerable increase in resolution. Chromatographic optimization of 2DLC is not trivial, but it has the advantage of greatly expanding the available peak capacity.

11.7 Additional considerations for the selection of solvents There may be several reasons for choosing a solvent other than its appropriate elution strength and selectivity, or the limits established by the solvent viscosity and cutoff wavelength (Table 11.1) [63,64]. Thus, below 220 nm, the baseline drift caused by the differential solvent absorbance may be sufficient to prevent the use of certain solvents, such as MeOH or THF. In turn, MeOH is less expensive and less toxic than ACN, and its higher polarity reduces the risk of buffer precipitation. In general, solvents that produce a high detector background or baseline drift cannot be used. In this regard, the continuous modification of the concentration of a minor component in the mobile phase could be much more significant in gradient methods than in isocratic approaches. This occurs, for example, when a solvent with some absorption is used with UV detection or one of the components of the mixture contains a conductive buffer if conductimetric detection is used, and in all instances with refractometric detection. Additionally, solvent variability from batch to batch can affect UV detection, especially when working near the cutoff wavelength. A broader range of solvents is compatible with evaporative light-scattering detectors, corona-charged aerosols, mass spectrometry, and ion-mobility spectrometry; however, low-volatility buffers and solvents cannot be used with these detectors. Other desired characteristics are solvent stability, low reactivity, and the ability to dissolve a wide range of compounds. THF has the drawback of its relative instability compared to other solvents. However, the use of other ethers instead of THF can be

277

278

CHAPTER 11 Mobile phase selection in liquid chromatography

problematic due to their limited miscibility with water. Analytes can also be affected by reactivity with the solvent. For example, higher alcohols (e.g., isopropanol) may be preferred, as they tend to denature biomolecules less than MeOH. Also, one of the reasons ACN is a popular choice for LC is its ability to dissolve a wide range of compounds with minimal chemical change. Care must also be taken with bacterial growth, which is a source of unexpected and unexplained chromatographic peaks, promoted by certain reagents added to aqueous mobile phases. Lack of availability or legal restrictions must also be addressed. For example, from late 2008 to early 2009, the production of ACN decreased leading to a significant price increase. There is also concern that many volatile organic solvents are toxic or hazardous to health or the environment (e.g., chlorinated solvents deplete the ozone layer). Therefore, legislation that restricts the use of these solvents may affect their choice or promote the search for alternatives. To reduce solvent consumption and its environmental impact, columns with a narrower internal diameter and/or smaller particle size should be used. Additionally, solvent recycling technologies can be a solution. All these reduced consumption patterns are supported by commitments to “greener” strategies in an effort to minimize pollution and wastes and increase sustainability. As discussed above, various “green” solvents of plant origin, mainly terpenes, have been recommended to replace alkanes. Ethanol and solketal are green alternatives to ACN and MeOH, but with the disadvantage of their higher viscosity. Additionally, ethanol is subject to restrictions in some countries to prevent illegal diversion for human consumption. Acetone is a good eco-friendly alternative, but the limit for UV–vis detection is large, approximately 330 nm. The organic solvent required in RPLC for a given separation can be reduced using high column temperatures. Currently, commercial equipment is available for the control and programming of the column temperature up to 200°C, with mobile phase preheating and postcolumn cooling, as well as bonded-silica columns capable of routinely withstanding high temperatures [65]. Preheating is necessary to avoid loss of efficiency caused by radial gradients within the column. Postcolumn cooling is also required to avoid boiling of the mobile phase when the pressure drops. Water becomes less polar at high temperatures. This increases its elution strength. From room temperature to 200°C, a 5°C increase is roughly equivalent to a 1% and 1.3% increase in ACN and MeOH, respectively. This allows the development of more environmentally friendly water-based RPLC methods, albeit at the cost of the additional energy required to maintain the oven temperature and preheating and cooling systems [66,67]. The selectivity changes achieved by increasing the temperature are complementary with respect to those produced by modifying the solvent composition of the mobile phase. These changes are mainly due to a different polarity of the solvent mixture, which is also highly dependent on solute molecules (derived from entropic, steric, conformational, and ionization effects). Unfortunately, the elution strength of water remains relatively low below 200°C, which in most cases makes it difficult to fully replace organic solvents with water.

References

Acknowledgments The authors thank the Ministry of Science and Innovation of Spain for the financial support of Project PID2019-106708GB-I00 (State Research Agency: Ref. PROYECTO/AEI/10.13039/ 501100011033).

References [1] Dolan JW. Selectivity in reversed-phase LC separations, part I: solvent-type selectivity. LCGC N Am 2010;28:1022–7. [2] Chastrette M, Rajzmann M, Chanon M, Purcell KF. Approach to a general classification of solvents using a multivariate statistical treatment of quantitative solvent parameters. J Am Chem Soc 1985;107:1–11. [3] Bruno TJ, Svoronos PDN. Handbook of basic tables for chemical analysis. 2nd ed. Boca Raton, FL: CRC Press; 2003. [4] Karger BL, Snyder LR, Eon C. An expanded solubility parameter treatment for classification and use of chromatographic solvents and adsorbents: parameters for dispersion, dipole and hydrogen bonding interactions. J Chromatogr 1976;125:71–88. [5] Poole CK. The essence of chromatography. Amsterdam: Elsevier; 2003. p. 369. [6] Snyder LR. Classification of the solvent properties of common liquids. J Chromatogr 1974;92:223–30. [7] Rutan SC, Carr PW, Cheong WJ, Park JH, Snyder LR. Re-evaluation of the solvent triangle and comparison to solvatochromic based scales of solvent strength and selectivity. J Chromatogr A 1989;463:21–37. [8] Garcı´a-Alvarez-Coque MC, Ramis-Ramos G, Baeza-Baeza JJ. Reversed phase liquid chromatography. In: Anderson JL, Stalcup A, Berthod A, Pino V, editors. Analytical separation science series, vol. 1. New York: Wiley; 2015. p. 159–98. [9] Younes AA, Galea C, Mangelings D, Vander Heyden Y. Normal-phase and polar organic solvents chromatography. In: Anderson JL, Stalcup A, Berthod A, Pino V, editors. Analytical separation science series, vol. 1. New York: Wiley; 2015. p. 227–44. [10] Hildebrand JH, Scott RL. The solubility of non-electrolytes. 3rd ed. New York: Dover Publication; 1964. [11] Schoenmakers PJ, Billiet HAH, de Galan L. The solubility parameter as a tool in understanding liquid chromatography. Chromatographia 1982;15:205–14. [12] Lo´pez-Grı´o S, Garcı´a-Alvarez-Coque MC, Hinze WL, Quina FH, Berthod A. Effect of a variety of organic additives on retention and efficiency in micellar liquid chromatography. Anal Chem 2000;72:4826–35. [13] Abbott TP, Kleiman R. Solvent selection guide for counter-current chromatography. J Chromatogr 1991;538:109–18. [14] Roses M. Determination of the pH of binary mobile phases for reversed phase liquid chromatography. J Chromatogr A 2004;1037:283–98. [15] Wolcott RG, Dolan JW, Snyder LR, Bakalyar SR, Arnold MA, Nichols JA. Control of column temperature in reversed-phase liquid chromatography. J Chromatogr A 2000;892:211–30. [16] Pous-Torres S, Torres-Lapasio´ JR, Baeza-Baeza JJ, Garcı´a-Alvarez-Coque MC. Combined effect of solvent content, temperature and pH on the chromatographic behaviour of ionisable compounds. J Chromatogr A 2007;1163:49–62.

279

280

CHAPTER 11 Mobile phase selection in liquid chromatography

[17] Rafferty JL, Siepmann JI, Schure MR. Understanding the retention mechanism in reversed-phase liquid chromatography: insights from molecular simulation. Adv Chromatogr 2010;48:1–55. [18] Schure MR, Rafferty JL, Zhang L, Siepmann JI. How reversed-phase liquid chromatography works. LCGC N Am 2013;31:630–7. [19] Johnson AR, Vitha MF. Chromatographic selectivity triangles. J Chromatogr A 2011;1218:556–86. [20] Snyder LR, Carr PW, Rutan SC. Solvatochromically based solvent-selectivity triangle. J Chromatogr 1993;656:537–47. [21] De Juan A, Fonrodona G, Casassas E. Solvent classification based on solvatochromic parameters: a comparison with the Snyder approach. Trends Anal Chem 1997;16:52–62. [22] Snyder LR. Changing reversed-phase high performance liquid chromatography selectivity. Which variables should be tried first? J Chromatogr B 1997;689:105–15. [23] Li J, Zhang Y, Carr PW. Novel triangle scheme for classification of gas chromatographic phases based on solvatochromic linear solvation energy relationships. Anal Chem 1992;61:210–8. [24] Hansen CM. The three dimensional solubility parameters. Key to paint component affinities. I. Solvents, plasticizers, polymers and resins. J Paint Technol 1967;39:104–17. [25] Hansen CM, Smith AL. Using Hansen solubility parameters to correlate solubility. Carbon 2004;42:1591–7. [26] Kamlet MJ, Taft RW, Carr PW, Abraham MH. Linear solvation energy relationships. Part 9. Correlation of gas/liquid partition coefficients with the solvatochromic parameters, π*, α and β. J Chem Soc Faraday Trans I 1982;78:1689–704. [27] Kamlet MJ, Abboud JLM, Abraham MH, Taft RW. Linear solvation energy relationships. 23. A comprehensive collection of the solvatochromic parameters, π*, α, and β, and some methods for simplifying the generalised solvatochromic equation. J Org Chem 1983;48:2877–87. [28] Abraham MH. Scales of solutes hydrogen-bonding: their construction and application to physicochemical and biochemical processes. Chem Soc Rev 1993;22:73–83. [29] Abraham MH, Ibrahim A, Zissimos AM. Determination of solute descriptors from chromatographic measurement. J Chromatogr A 2004;1037:29–47. [30] Jover J, Bosque R, Sales J. Determination of Abraham solute parameters from molecular structure. J Chem Inf Comput Sci 2004;44:1098–106. [31] Ruiz-Angel MJ, Carda-Broch S, Garcı´a-Alvarez-Coque MC, Berthod A. Effect of ionization and the nature of the mobile phase in quantitative structure-retention relationship studies. J Chromatogr A 2005;1063:25–34. [32] Vitha M, Carr PW. The chemical interpretation and practice of linear solvation energy relationships in Pchromatography. J Chromatogr A 2006;1112:143–94. [33] Lessellier E. pider diagram: a universal and versatile approach for system comparison and classification. Application to solvent properties. J Chromatogr A 2015;1389:49–64. [34] Garcı´a-Alvarez-Coque MC, Ramis-Ramos G, Ruiz-Angel MJ. Liquid chromatography, ion pair. In: Reedijk J, editor. Reference module in chemistry, molecular sciences and chemical engineering series. Waltham, MA: Elsevier; 2015. [35] Peris-Garcı´a E, Ruiz-Angel MJ, Baeza-Baeza JJ, Garcı´a-Alvarez-Coque MC. Comparison of the fitting performance of retention models and elution strength behaviour in

References

[36]

[37] [38]

[39] [40] [41] [42]

[43] [44]

[45]

[46] [47] [48]

[49] [50]

[51]

[52] [53]

hydrophilic-interaction and reversed-phase liquid chromatography. Separations 2021;8:54. Schoenmakers PJ, Billiet HAH, de Galan L. Description of solute retention over the full range of mobile phase compositions in reversed-phase liquid chromatography. J Chromatogr 1983;282:107–21. Nikitas P, Pappa-Louisi A. Retention models for isocratic and gradient elution in reversed-phase liquid chromatography. J Chromatogr A 2009;1216:1737–55. Jandera P, Kucerova´ M. Prediction of retention in gradient-elution normal-phase high performance liquid chromatography with binary solvent gradients. J Chromatogr A 1997;759:13–25. Euerby MR, Hulse J, Petersson P, Vazhentsev A, Kassam K. Retention modelling in hydrophilic interaction chromatography. Anal Bioanal Chem 2015;407:9135–52. Jandera P, Jana´s P. Recent advances in stationary phases and understanding of retention in hydrophilic interaction chromatography. J Chromatogr A 2017;967:12–32. Xiao HB, Liang XM, Lu RC. Classification of structurally related compounds from Astragalus extract by correlation of the log kw and S. Chromatographia 2000;51:212–20. Baeza-Baeza JJ, Garcı´a-Alvarez-Coque MC. Extension of the linear solvent strength retention model including a parameter that describes the elution strength changes in liquid chromatography. J Chromatogr A 2020;1615, 460757. Dolan JW. The hazards of adjusting gradients. LCGC N Am 2002;20:940–6. Vivo´-Truyols G, Torres-Lapasio´ JR, Garcı´a-Alvarez-Coque MC. Estimation of significant solvent concentration ranges and its application to the enhancement of the accuracy of gradient predictions. J Chromatogr A 2004;1057:31–9. Snyder LR, Glajch JL, Kirkland JJ. Theoretical basis for systematic optimisation of mobile phase selectivity in liquid-solid chromatography. J Chromatogr 1981;218:299–326. Nyiredy S, Meier B, Erdelmeier CAJ, Sticher O. Prisma: a geometrical design for solvent optimization in HPLC. J High Resolut Chromatogr Chromatogr Comm 1985;8:186–8. Snyder LR, Quarry MA, Glajch JL. Solvent-strength selectivity in reversed-phase HPLC. Chromatographia 1987;24:33–44. Torres-Lapasio´ JR, Garcı´a-Alvarez-Coque MC. Levels in the interpretive optimisation of selectivity in high-performance liquid chromatography: a magical mystery tour: a review. J Chromatogr A 2006;1120:308–21. Cela R, Ordon˜ez EY, Quintana JB, Rodil R. Chemometric-assisted method development in RPLC. J Chromatogr A 2013;1287:2–22. Gisbert-Alonso A, Navarro-Huerta JA, Torres-Lapasio´ JR, Garcı´a-Alvarez-Coque MC. Testing experimental designs in liquid chromatography (II): influence of the design geometry on the prediction performance of retention models. J Chromatogr A 2021;1654, 462458. Garcı´a-Alvarez-Coque MC, Torres-Lapasio´ JR, Baeza-Baeza JJ. Models and objective functions for the optimisation of selectivity in reversed-phase liquid chromatography: a review. Anal Chim Acta 2006;579:125–45. Molnar I. Computerized design of separation strategies by reversed-phase liquid chromatography: development of DryLab software. J Chromatogr A 2002;965:175–94. Galushko SV, Kamenchuk AA, Pit GL. Calculation of retention in reversed-phase liquid chromatography. IV. ChromDream software for the selection of initial conditions and for simulating chromatographic behaviour. J Chromatogr A 1994;660:47–59.

281

282

CHAPTER 11 Mobile phase selection in liquid chromatography

[54] Heinisch S, Lesellier E, Podevin C, Rocca JL, Tchapla A. Computerized optimization of RP-HPLC separation with nonaqueous or partially aqueous mobile phases. Chromatographia 1997;44:529–37. [55] Cela R, Leira E, Cabaleiro O, Lores M. PREOPT-W: off-line optimization of binary gradient separations in HPLC by simulation—IV. Phase 3. Comput Chem 1996;20:315–30. [56] Torres-Lapasio´ JR, Garcı´a-Alvarez-Coque MC, Baeza-Baeza JJ. Global treatment of chromatographic data with MICHROM. Anal Chim Acta 1997;348:187–96. [57] Berry VV. Universal liquid chromatography methods: V. “Pulsing” with weak eluent to resolve peaks. J Chromatogr 1985;321:33–43. [58] Wahl JH, Enke CG, McGuffin VL. Solvent modulation in liquid chromatography: experimental verification and comparison with conventional premixed mobile phases. Anal Chem 1991;63:1118–26. [59] Ortiz-Bolsico C, Torres-Lapasio´ JR, Garcı´a-Alvarez-Coque MC. Approaches to find complementary separation conditions for resolving complex mixtures by highperformance liquid chromatography. J Chromatogr A 2012;1229:180–9. [60] Louw S, Pereira AS, Lynen F, Hanna-Brown M, Sandra P. Serial coupling of reversedphase and hydrophilic interaction liquid chromatography to broaden the elution window for the analysis of pharmaceutical compounds. J Chromatogr A 2008;1208:90–4. [61] Chalcraft KR, McCarry BE. Tandem LC columns for the simultaneous retention of polar and nonpolar molecules in comprehensive metabolomics analysis. J Sep Sci 2013;36:3478–85. [62] Jandera P. Comprehensive two-dimensional liquid chromatography: practical impacts of theoretical considerations. Cent Eur J Chem 2012;10:844–75. [63] Barwick VJ. Strategies for solvent selection: a literature review. Trends Anal Chem 1997;16:293–309. [64] Berthod A, Faure K. Solvents in chromatography and electrophoresis. In: Anderson JL, Stalcup A, Berthod A, Pino V, editors. Analytical separation science series, vol. 1. New York: Wiley; 2015. p. 135–58. [65] Heinisch S, Rocca JL. Sense and nonsense of high-temperature liquid chromatography. J Chromatogr A 2009;1216:642–58. [66] Pawlowski TM, Poole CF. Solvation characteristics of pressurized hot water and its use in chromatography. Anal Commun 1999;36:71–5. [67] Smith RM. Superheated water chromatography—a green technology for the future. J Chromatogr A 2007;1184:441–55.

CHAPTER

Co-solvents and mobile phase additives in HPLC

12

Michele Protti1, Andrea Carotti2, Laura Mercolini1, and Roccaldo Sardella2 1

Department of Pharmacy and Biotechnology (FaBiT), Alma Mater Studiorum, University of Bologna, Bologna, Italy, 2Department of Pharmaceutical Sciences, University of Perugia, Perugia, Italy

12.1 Introduction In reversed-phase (RP) mode, the optimization of the thermodynamic and kinetic features of the chromatographic process can be accomplished through the careful modification of the mobile phase composition by varying the type and concentration of the organic modifier (usually acetonitrile or methanol), as well as of specific compounds deliberately added to the eluent. For ionizable analytes, the variation of the eluent pH also represents one of the main strategies to tune analyte retention in the analysis of single analytes as well as to enhance selectivity in that of more complex mixtures. Improving the retention profile of ionic and ionizable compounds still poses a challenge due to their inherent polarity and consequent early elution with aqueous eluents. Tuning the amount of the organic component in a RP eluent often has only a marginal effect on the retention and selectivity of fully ionized species. On the other hand, eluent pH could dramatically improve their retention contributing to the amelioration of selectivity in the multicompound analysis. Another approach (which is simultaneous in many applications to the pH optimization) to modulate retention and selectivity deals with the addition of specific compounds to the eluent which can have a beneficial effect also on the kinetic features of the chromatographic process as a result of the abolition or strong mitigation of unwanted secondary interactions with some functional groups of the stationary phase. The main aim of this chapter is to shortly describe the beneficial effect deriving from the use of a series of mobile phase additives in the RP-HPLC analysis of ionic and neutral compounds. Some selected examples regarding the use of fluorinated ion-pairing agents (carboxylic acids [1–7], amines [8], and alcohols [9–11]), ionic liquids [12–15], deep eutectic solvents [16–19], chaotropic [20–22] and kosmotropic [23,24] agents, and surfactants [25–27] are discussed. Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00025-4 Copyright # 2023 Elsevier Inc. All rights reserved.

283

284

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

12.2 Fluorinated ion-pairing agents: Carboxylic acids, amines, and alcohols Perfluorinated carboxylic acids can be efficiently used as mobile phase additives for the direct separation of underivatized amino acids. The direct methods based on the in-column ion-pair (IP) formation eliminate the need for pre- or post-column derivatization procedures bringing the advantages of enhanced simplicity, flexibility, and separation speed. Perfluorinated carboxylic acids have the additional benefit of low boiling points which makes them compatible with evaporative light-scattering [1,28,29] and mass spectrometry [2,30,31] detectors (commonly referred to as ELSD and MS, respectively). The IP formation at the basis of the separation process is particularly convenient for the analysis of polar amino acids otherwise co-eluting near the void volume when reversed-phase packing materials are used. The IP reagent added to an aqueous mobile phase (usually at a concentration < 1 mM) induces a peculiar modification of the surface of the reversed-phase packing material. Under these conditions, the interactions between hydrophilic amino acids and the stationary phase occur through a concerted mechanism involving the stationary phase, the ion-pairing agent, and the solute [1,2,28,30,31]. Therefore, the nature (type and chain length of the hydrophobic group) of the IP agent is of prior importance for the retention of charged solutes in these chromatographic settings. Elfakir and co-workers performed in-depth investigations concerning the use of perfluorinated carboxylic acids as mobile phase additives for the separation of amino acids in IP-reversed-phase liquid chromatography (RPLC) systems [1]. In this framework, the authors found that semi-long (5–9 carbons) n-alkyl chain perfluorinated carboxylic acids, such as tridecafluoroheptanoic acid (TDFHA) and pentadecafluorooctanoic acid (PDFOA), are among the best IP agents for the analysis of the most difficult-to-separate polar amino acids. Improved selectivity usually derived from the use of surfactant with a long side chain, even though these produced long analysis times. As a rule, retention increases with the additive concentration, whereas no simple rules exist in terms of selectivity, especially when complex mixtures are analyzed: In such cases, the best condition is commonly selected to provide a convenient compromise. A method for the separation of all the 20 proteinogenic amino acids by IP chromatography was successfully developed by the same group of authors with TDFHA under gradient RP conditions with a water-acetonitrile (ACN)-based mobile phase [2]. In Figure 12.1, the efficient direct LC–MS analysis of a standard solution of the 20 proteinogenic underivatized amino acids with a C18 column is shown. The elution order of the basic and more hydrophobic amino acids was easily explained by their basicity and lipophilicity, respectively. On the contrary, the elution order of the less retained compounds was of more difficult rationalization, being correlated with charge and polarity characteristics of the analytes. Two main points were highlighted by the authors: (i) Changing the type of IP additive (that is, for example,

12.2 Fluorinated ion-pairing agents

FIGURE 12.1 (A) LC–MS analysis of 20 proteinic underivatized amino acids under gradient elution. Column: Supelcosil ABZ+ Plus (150  4.6 mm I.D.); gradient elution: solvent A, 1.0 mM TDFHA in water, solvent B, ACN; gradient profile: 0% B for 6 min, from 0 to 20% B in 2 min, then 20% B is maintained for 8 min, from 20 to 25% B in 2 min, then 25% B is maintained to the end of the analysis; room temperature; flow rate: 1.0 mL/min. (B) LC-ELSD analysis of the 20 underivatized amino acids in gradient elution. Column: Hypercarb (100  2.1 mm I.D.). Gradient elution: solvent A, 20 mM NFPA in water, solvent B, ACN; gradient profile: from 0 to 15% B in 10 min, then 26% B in 10 min and finally 50% B in a further 10 min. 50% B is maintained until the end; column temperature: 10°C; flow rate: 200 μL/min. Imp: impurity. Adapted with permission from Chaimbault et al. [2] and Chaimbault et al. [3].

285

286

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

PDFOA in place of TDFHA) can result in a modification of the elution order of some species; and (ii) a careful regeneration of the column based on the consecutive flowing of methanol (MeOH), ACN, tetrahydrofuran (THF), and again MeOH through the column was strongly recommended to ensure that no irreversible surface modification can occur between each experiment. In this study, MS was necessary for the identification of all the proteinogenic amino acids. Indeed, the complete separation of the 20 compounds by means of an ELSD system was prevented by co-elution of some species either between themselves or with system peaks occurring during the gradient elution. The complete chromatographic selectivity was obtained by Elfakir and co-workers in a following study [3], in which a porous graphitic carbon (PCG) stationary phase was used under RP conditions by applying a gradient elution program whose details are reported in the caption to Figure 12.1. For this successful application, nonafluoropentanoic acid (NFPA) was used as IP agent. In analogy to the previous IP-RPLC systems, also with the PCG phase, analyte retention increased as the IP concentration in the mobile phase was increased. Accordingly, the best selectivity was obtained with 20 mM to 25 mM of NFPA. Among the main advantages of the PGC-based system in comparison with the conventional RP18 support, the authors underlined the following points: (i) absence of system peaks (SPs) occurring as a result of the gradient program; (ii) full compatibility with MS, due to the high volatility of NPFA; and (iii) faster equilibration time than the conventional systems based on octadecylsilica (ODS)bonded materials. The authors also underlined that the elution order of the amino acids on PGC support is quite different from that produced by ODS packings. Indeed, in addition to its strong RP behavior, delocalized π-electrons from the graphite can participate in the retention mechanism and even dominate it [32,33]. Perfluorinated carboxylic acids have been successfully used as IP agent also for the analysis of other classes of compounds including, inter alia, peptides [4] and pharmaceutically active ingredients [5]. Perfluorinated carboxylic acids have been mostly used to improve the retention and separation efficiency of oppositely charged (i.e., positively charged) analytes through their IP capacity. Instead, only very limited attention has been paid to their effect on the retention and selectivity of similarly charged (i.e., negatively charged) compounds. Yamamoto and co-workers described the use of partially fluorinated carboxylic acids as IP reagents for the analysis of basic compounds of pharmaceutical relevance under RP conditions with electrospray ionization (ESI)-MS [6]. The authors found that some of the partially fluorinated carboxylic acids tested, such as CF3–(CF2)n–CH2–COOH (n≧1), provide profitably long retention coupled to high sensitivity in ESI-MS. In a recent paper [7], Lajin and Goessler systematically evaluated the effect of the incorporation in the mobile phase of three perfluorinated carboxylic acids (that is, trifluoroacetic acid—TFA, pentafluoropropionic acid—PFPA, and heptafluorobutyric acid—HFBA) for the separation of halogenated carboxylic acids and sulfonic

12.2 Fluorinated ion-pairing agents

acids under RP condition with either a C18 or a C5 stationary phase. The authors demonstrated that the use of these additives as “ion repelling agents” can be of practical utility to fine-tuning the chromatographic selectivity of organic acids. The same authors have recently investigated the use of four commercially available fluoroalkylamines (trifluoroethylamine— TFEAm, pentafluoropropylamine— PFPAm, heptafluorobutylamine—HFBAm, and nonafluoropentylamine—NFPAm) as cationic IP reagents to use in combination with a conventional C18 stationary phase [8]. The four selected fluoroalkylamines shared the same characteristics: (i) water miscibility at concentrations levels spanning in the 1–10 mM range, under acidic pH values ensuring their protonated state; (ii) high volatility to allow full compatibility with MS detectors; and (iii) a carbon chain length of ClO4 > BrO3 > Cl > CH3COO > IO3 ,    I > SCN > NO3

 IO 4 > Br ,

For cations, Na+ > K+ > Li+ > Ba3+ > Rb+ > Ca+ > Ni2+ > Co2+ > Mg2+ > Fe2+ > Zn2+ > + Cs > Mn2+ > Al3+ > Fe3+, Cr3+ > NH+4 > H+

12.6 Kosmotropic ions

This order is not always verified, but it is subject to variations due to the nature of the system and the type of effect studied. A further representation of the Hofmeister series for anions and cations adapted from [62] appears in Figure 12.16, showing anions and cations in decreasing order of kosmotropic effect from top to bottom. The effect has been assessed by several studies [56] also studying the effect on protein stability under specific conditions [63]. Figure 12.17 shows a possible mechanism according to which interactions of this kind arise between ions and a protein, with no direct interaction affecting the water

FIGURE 12.16 Hofmeister series for anions and cations in aqueous solution. Adapted with permission from Mazzini and Craig [62].

305

306

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

FIGURE 12.17 Possible mechanism of ion effects over the primary structure of a protein. (A) The cation can “solvate” carbonyl directly or indirectly by increasing the availability of water hydrogen, (B) the strongly hydrated anion competes with the carbonyl for hydration, and (C) the cation and anion have opposite effects on the hydrogen donor/acceptor equilibrium. Adapted with permission from Xie and Gao [54].

structure [54]. For example, kosmotropic anions can polarize water molecules that are hydrogen-bonded to proteins on the positive side of the dipole. In addition to the ways in which salts can interact with proteins influencing protein–protein interactions, they also show effects on stability and solubility. These phenomena are usually caused by an increased interfacial tension between proteins and water, thus causing stronger hydrophobic interactions between the protein and the stationary phase [64]. Such hydrophobic effects induced by salts are at the basis of HIC [64]. The addition of kosmotropic ions has a small resolution enhancement effect, while also improving selectivity and peak shape. Therefore, in HIC settings, analyte retention is modulated by variations in the salt concentration within the mobile phase, while in protein, it retains its native structure as much as possible. There is still no wide agreement between the theory and experimental data regarding HIC to prove this concept, due to the complex interactions in the framework of chromatographic systems [65]. However, there is consensus that kosmotropic ions of the Hofmeister series affect protein solubility and stability in two possible ways: the first one, by changing the polarizability of the water-mediated ion-protein interaction, and the second one by direct binding of the ions to the protein [55]. HIC chromatographic methods for protein analyses have been proposed and developed in several studies [66–68]. The use of mobile phases containing kosmotropic salts, such as ammonium sulfate, has been exploited in HIC systems. Separations are usually obtained with an inverse salt concentration gradient, from high to

12.6 Kosmotropic ions

low concentrations of ammonium sulfate. The addition of the kosmotropic salt to mobile phase promotes hydrophobic interactions between the protein species and the stationary phase, based on the Hofmeister series to predict the effects [43]. The mobile phase conditions in HIC are usually obtained through an empirical process that can be simplified by applying generic RPLC gradient elution relationships. It has been shown that retention and separation of proteins in HIC regimens can be explained by the linear solvent strength (LSS) gradient relationship model for RPLC [69]. Growing interest in HIC-based analysis of proteins led to generic guidelines to be applied for method development. These include the use of other possible kosmotropic agents besides ammonium sulfate, as well as the optimization of salt concentration, gradient, and gradient profile. For example, the salt type and concentration has been studied in order to optimize HIC mobile phase for the determination of monoclonal antibodies [67,70]. Within these studies, the influence of 2 M ammonium sulfate was tested with regard to selectivity for antibody separation, and it has been compared with other salts (ammonium formate, sodium chloride, and ammonium acetate). It has been observed that the use of other salt can allow similar selectivity by properly adjusting kosmotropic strength. This can be done by assaying the concentration of kosmotropic salt leading to similar selectivity. For example, 2 M ammonium sulfate in the mobile phase led to the same selectivity as 5 M sodium chloride. Moreover, selectivity can be adjusted by changing the concentration of Cl [70]. Similar selectivity to that obtained with 2 M ammonium sulfate can be observed with either sodium formate or sodium acetate added to the mobile phase in concentrations ranging from 5.0 to 5.5 M. These results are in good agreement with the Hofmeister series [67]. From the point of view of method development, added salts should be interchangeable if they are in proximity to each other in the Hofmeister series. At the same time, their salting-out strength should be adjusted accounting for their concentration. Another alternative for HIC method development has been described considering the separation of cytochrome c, ribonuclease A, and lysozyme exploiting an ammonium sulfate inverse concentration gradient from 1.5 M to 0.5 M. The results demonstrated that such kind of gradient does not behave linearly when compared to the LSS gradient model due to a mixed mode of interaction. At higher salt concentration, linearity improves according to the model. The best results were observed at a concentration of 1.8 M at the beginning of the gradient [68]. Retention of protein species in HIC not only depends on the concentration of kosmotropic salt, its type, and concentration gradient program, but also on the stationary phase to be exploited. Hydrophobic stationary phases, as well as salt concentration in the mobile phase, can induce protein conformational changes, as well as aggregation. Stationary phases in HIC can be either amphiphilic or hydrophilic, with hydrophobic ligands linked by spacers. HIC stationary phases differ in chemical nature, ligand surface concentration, its chemistry, and matrix particle size. A comprehensive overview of the main stationary phases exploited in HIC can be found in [67]. The LSS model is a function taking into account sets of experimental conditions affecting the separation of proteins. Such conditions, such as flow rate, column dimensions, and salt gradient, lead to a predictable effect on the separation (while

307

308

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

keeping constant temperature and pH of the mobile phase). The effect of these conditions can be predicted by the LSS model, in order to obtain a proper method optimization for protein resolution and analysis.

12.7 Surfactant additives The word surfactant is a contraction of “surface-active agent,” referring to the main property of this class of compounds: their absorption at any interface, modifying the internal tension or surface. Surfactant molecules are also called “detergents” and “amphiphiles.” The amphiphilic properties of surfactant molecules are due to the association inside the same molecule of two parts with very differing polarities. One part of the surfactant molecule is highly non-polar, hydrophobic, or lipophilic (usually an alkyl chain). Another part is polar or hydrophilic, and it can be an ionic (anionic or cationic) group, or non-ionic chain with polar groups, such as alcohol, ether, or amine groups. Some surfactants possess two polar heads or two non-polar tails. The nature of the surfactant polar head usually determines the classification of the molecule. Based on the electrical charge of their polar head, the three main surfactant classes are anionic surfactants, cationic surfactants, and non-ionic surfactants. A fourth class is considered for amphoteric and/or multifunctional surfactants [71]. In water and aqueous solutions, anionic surfactants dissociate providing an anion which carries amphiphilic properties and an inactive cation. Soaps, sulfonated compounds, alkylsulfates, and alkylphosphates are the main families of anionic surfactants. Soaps: Saponification of natural oils and fats produces glycerol and soaps. Soaps are more often a mixture of several sodium and/or potassium fatty acid salts. They absorb weakly UV radiation at short wavelengths. Their water solubility is very low, and solutions of soaps are moderately basic (pH  9); thus, at low pH values, the corresponding fatty acids separate in a supernatant layer. Sulfonated compounds: Aqueous sulfonate solutions are neutral and stable. In sulfonated surfactant molecules, the sulfur atom is directly connected to a carbon atom of the hydrophobic residue, usually an alkyl chain ranging from 8 to 24 carbon atoms. Alkylsulfonates do not absorb UV light at wavelengths higher than 200 nm. Alkylbenzenesulfonates include an aromatic ring which absorbs UV light. Sulfosuccinates show two hydrophobic chains and are dialylsulfosuccinate ester salts. They slightly absorb UV light at short wavelengths. However, the ester group is sensitive to hydrolysis, so they can only be effectively exploited in almost neutral solutions. Alkylsulfates: This family of surfactants is salts of sulfuric acid esters, produced by sulfation of linear alcohols. Alkylsulfate aqueous solutions are neutral and stable in a wide pH range. Slight hydrolysis occurs in highly acidic solutions. Sodium dodecyl sulfate (SDS, C12H25OSO3Na) is the best-known member of this family. Its physico-chemical properties have been extensively investigated [72].

12.7 Surfactant additives

Alkylphosphates: These surfactant molecules include salts of the mono- and dialkylesters of phosphoric acid. Alkylphosphates are low UV absorbing compounds and are stable even at extreme pH aqueous solutions [73]. Other anionic surfactants: The hydrophilic part of many surfactant molecules produced at the industrial level sees the association of an anionic moiety with a non-ionic one [74] These surfactants have the general behavior of anionic surfactants. Two examples of such composite surfactants are alkylethersulfates and carboxymethylethoxylates. In water solutions, cationic surfactants are ionized in a cation with amphiphilic properties associated with an inactive anion, such as Cl or Br. The cationic group is usually a quaternary ammonium group. Cetyl or hexadecyl trimethyl ammonium bromide (CTAB or HTAB) is the most known and studied cationic surfactant. Alkyl trimethylammonium salts and diallyl dimethylammonium salts do not absorb UV radiation and are very stable in aqueous solutions. Other cationic surfactants are derived from pyridine and imidazole, such as alkyl pyridinium and alkyl imidazolidinium salts. Both provide stable aqueous solutions, but possess UV absorption properties [74]. Non-ionic surfactants obviously do not generate ions in solutions. The hydrophilic portion of their structure contains polar moieties such as alcohol, carbonyl, ether, or amino groups. Non-ionic surfactants are stable in aqueous solutions and are little sensitive to water hardness and ionic strength of the aqueous medium. Alkyl ethoxylates: These compounds are non-UV absorbing. Due to the ethylene oxide polymerization step in their synthesis, alkyl ethoxylates are mixtures of molecules with the same hydrophobic alkyl chain and a hydrophilic portion having a different number of ethylene oxide units [75]. At the production level, it is possible to adjust both the hydrophobic chain length and the ethylene oxide condensation in the non-ionic surfactant synthesis, finally tuning the properties of the final molecule to the needs. Physico-chemical properties such as solubilization, emulsification, wetting, cleansing, foaming, and anti-foaming are very different among the members of this wide family. The hydrophilic-lipophilic balance (HLB) value of a surfactant molecule is defined as the ratio of the molecular weight of the hydrophilic portions divided by the overall molecular weight of the compound, multiplied by a factor of 20. HLB values are used to sort the surfactants on a 0–20 scale, from the most hydrophobic ones (low HLB values) to the more hydrophilic ones (high HLB values). Two surfactants with similar HLB values may show different behavior due to very different molecular weights. Copolymers: This is another family of non-ionic, non-UV absorbing surfactants, in which the hydrophilic chain is an ethylene oxide polymer and the hydrophobic chain is a propylene oxide polymer. By changing the oxyethylene and oxypropylene unit numbers, it is possible to obtain homologues with a wide variety of properties. Other non-ionic surfactant molecules include ethoxylated alkyl phenols, which strongly absorb UV radiation, and ethoxylated fatty acids, fatty esters, and alkanolamides, which are slightly UV-absorbing surfactants. Amine oxides are non-ionic surfactants in basic and neutral solutions.

309

310

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

Amphoteric surfactants are ionic compounds exposing positive and negative charges on the same molecule [76]. They can be true amphoteric ions, such as betaines, having a cationic nature in strongly acidic media and an amphoteric behavior in neutral and basic solutions. Aminocarboxylic acids are ampholytes: They are cationic surfactants in strongly acidic media, and anionic surfactants in basic media, while showing an isoelectric point at an intermediate pH value with an inner salt amphoteric structure. Natural L-amino-acids can also be ascribable to this class. Other amphoteric surfactants include betaines derived from imidazolines, which absorb UV light and lecithins or phosphatidylcholines, which are naturally occurring phospholipids [77]. Multifunctional surfactants associating a non-ionic moiety with a charged one are considered ionic surfactants. Surfactant molecules are water-soluble and produce solutions possessing unique properties, due to their amphiphilic nature. The two main properties of surfactant solutions are the adsorption at any interface and the micelle formation, as consequences of both polar and non-polar interactions. Molecular interactions and surfactant adsorption: Interactions among molecules can be classified following an increasing polarity order: van der Waals forces, hydrogen bonding interactions, and electrostatic forces. On the other hand, for hydrophobic interactions, when a non-polar solute is dissolved in water, some hydrogen bonds are disrupted, and the solute tends to locally distort water structure and restrict the motion of water molecules. The amphiphilic character of surfactant molecules explains their trend to absorb at any interface. Two phases of different polarities are separated at an interface. This polarity difference attracts the surfactant molecules because this can minimize the entropy change by orienting their polar portion in the more polar phase and their non-polar part in the less polar phase. The adsorption of surfactant molecules at an interface decreases its interfacial tension. When surfactant solutions are exposed to air, the decrease of the waterair superficial tension explains the foaming property. The addition of a surfactant into a biphasic liquid system allows the formation of emulsions, thanks to the decreases of liquid–liquid interfacial tension. Figure 12.18 shows how several physico-chemical properties solution containing a surfactant (one among all, surface tension) undergoes an abrupt change over a narrow range of concentrations. This concentration is defined as critical micelle concentration (CMC) of the surfactant. Above this concentration, micelles formation can be observed. Figure 12.19 shows a typical micelle structure for an ionic surfactant [79]. The surfactant molecules arrange in such a way that their polar heads are oriented toward the water phase and their non-polar tails aggregate in the micelle core. This micelle structure minimizes the molecule energy: The large entropy increases of water molecules associated with the removal of non-polar surfactant tails from the aqueous solution are the main micelle formation driving force. Moreover, electrostatic forces tend to separate polar heads with the same charge, and the whole micelle is an equilibrium between these forces, although very sensitive to any chemical additive or parameter that can act on any of these forces, such as salts, solutes, temperature, and pressure.

FIGURE 12.18 Physico-chemical changes related to micelle formation at the critical micelle concentration (CMC). Adapted with permission from Al-Soufi et al. [78].

FIGURE 12.19 Representation of an ionic micelle. The Gouy-Chapman layer corresponds to the diffuse layer under the electric field due to the micelle. The Stern layer corresponds to the ionic groups. Adapted with permission from Pharr [79].

312

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

Ionic surfactants below the CMC value have been added to the mobile phase RPLC since long time in IPC. In this chromatographic setup, monomers of surfactant molecules on the mobile phase adsorb on the stationary phase and act as hydrophobic counter-ions, leading to a modified retention of charged solutes. In the 80s, the use of aqueous solutions of surfactants of different nature (with either ionic or non-ionic head groups) was suggested at concentrations exceeding the CMC as mobile phases [80]. Therefore, RPLC modes with mobile phases containing micelles in addition to surfactant monomers were named micellar liquid chromatography (MLC) [25]. The unique properties of micelle-based mobile phases are due to the ability of micelles to selectively sort and organize analyte species at the molecular level. However, as an additional phenomenon, the association between the surfactant monomers and the bonded stationary phase, forming a surface similar to the exterior of micelles, leads to important modifications in terms of retention and selectivity. In fact, analytes in solution are resolved on the basis of their different partitions between the bulk aqueous phase and the micelles in the mobile phase, and between the bulk aqueous phase and the surfactant-modified stationary phase. For water-insoluble species, equilibrium can also take place through direct transfer of the analyte included in the micellar pseudo-phase to the surfactant-bonded stationary phase. In the very first reports on MLC, only pure aqueous micellar solutions were proposed as mobile phases. Despite the attractiveness of this approach, it suffers two main problems compared to RPLC with aqueous-organic mobile phases: the excessive retention of apolar compounds and the reduced efficiency observed for solutes spanning wide ranges of polarities. For these reasons, small amounts of short-chain alcohols to the micellar mobile phase were suggested in order to enhance efficiency. Then, the use of organic modifiers added to micellar mobile phases to decrease total analysis run time was also proved. Indeed, most of the analytical methods published based on MLC use as mobile phase micellar solutions of surfactants above the CMC mixed with an organic solvent. Despite the problems found in its initial development and after more than three decades of experience, MLC appears to be a possible alternative to conventional RPLC with aqueous-organic mobile phases with possible developments within green chemistry approaches. The advantages and capabilities derived from the use of micellar mobile phases in RPLC can be summarized in the following points: - Both charged and neutral analytes can be separated with the same mobile phase/ stationary phase combination; - The interactions among analytes, stationary phase, aqueous phase, and micelles can yield unique selectivity for many compounds; - The analysis of samples containing compounds in a wide range of polarities is possible using isocratic elution. However, in case of gradient elution, equilibration times are usually shorter than those with RPLC aqueous-organic mobile phases; - The high solubilization capability of micelles facilitates dissolution of several components of complex matrices, potentially avoiding time-consuming sample preparation protocols. Thus, for example, the direct on-column injection of biological fluids is a possible approach;

12.7 Surfactant additives

- The compartmentalization of organic compounds by micelles leads to enhanced luminescence detection; - The ratio of organic solvent needed in hybrid micellar-based mobile phases is appreciably lower than in conventional RPLC mobile phase mixtures. This is translated in a lower toxicity and costs, and the reduction of the environmental impact of hazardous wastes in a green chemistry view; - Organic solvents are highly retained in the micellar medium, leading to a decreased evaporation and making micellar mobile phases stable over time. As a consequence, retention behaviors are highly reproducible and can be accurately tailored to predict changes in retention times with mobile phase composition (concentration of surfactant and volume fraction of organic modifier). This allows for more feasible method development approaches; and - MLC exploits the same hardware (pumps, tubing, injectors, detectors, etc.) and chromatographic columns as conventional RPLC. However, the solid nature of the surfactants usually exploited in MLC should be considered in order to avoid their precipitation within the column. Despite this, the long life of hardware and columns used in MLC routines has been reported. The only real limitation to the applicability of the technique in the analytical laboratories is related to the use of mass spectrometry (MS) detection, since direct online coupling to MLC is hindered by the presence of high concentrations of surfactant in the mobile phase. A suitable surfactant for MLC should have low CMC, aggregation number (the number of molecules present in a micelle once the CMC has been reached), and, for ionic surfactants, a low Kraft point, which is defined as the temperature at which the ionic surfactant solubility corresponds to its CMC [81]. This should be preferably much smaller than the ambient temperature. A high CMC value would imply the use of high concentration of surfactant, which in turn would result in high-viscosity solutions with consequent, undesirable high column pressure, and high background noise in spectrophotometry-based detectors. Since these detectors are the most exploited in MLC, a suitable surfactant should also have small absorption at the operating wavelength. Being the size of micelles a few nanometer, no problem caused by micelles is expected in light scattering detectors. Several surfactants of different nature fulfill the aforementioned conditions, but the number of surfactant compounds that have been exploited in MLC is limited. The anionic SDS (also known as sodium lauryl sulfate) is, by far, the most common surfactant in MLC, followed by CTAB and the non-ionic polyoxyethylene-(23)-dodecyl ether (Brij-35). Other surfactants include ionic, non-ionic, and zwitterionic surfactants. SDS is usually selected due to its high purity and relatively low-cost commercial availability. SDS is highly exploited in the manufacturing of cosmetics and detergents, and in several scientific fields. It allows the setup of cost-effective procedures with respect to RPLC with aqueous-organic mixtures. Also, when dealing with biological samples (urine, plasma, serum, etc.), SDS efficiently dissolves proteins allowing for direct sample injection in the chromatographic system with

313

314

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

feasible and streamlined sample pretreatment (e.g., filtration could be sufficient) [82]. On the other hand, this is not possible with cationic surfactants. Conventional SDS-modified C18 columns can accommodate hundreds of injections of biological matrices without appreciable decrease in column performance or increase in back pressure. Several studies demonstrated how the stationary phase is altered when the mobile phase contains a surfactant, leading to analyte solute-modified stationary phase interactions. This explains the interest in investigating the modification of alkyl- and cyanopropyl-bonded stationary phases when SDS or CTAB (the most common surfactants in MLC) is added to the mobile phase. Differences in selectivity between MLC and such surfactant additives on alkyl-bonded phases were attributed to the differing nature of the SDS- and CTAB-bonded phase interaction [83] (Figure 12.20). For SDS, it has been observed that the hydrophobic tail was associated with the C18 alkyl chain bonded to the silica stationary phase, the sulfate group oriented away from the surface. This creates a negatively charged hydrophilic layer on the C18 stationary phase. This charged layer affects the penetration depth of analytes in the bonded phase due to strong hydrogen-bonding interactions. The result is a decrease in hydrophobic interactions between the analytes and the C18 stationary phase. It would explain the superior resolution achieved by SDS micellar mobile phases. In the case of CTAB mobile phases, the surfactant association leads to a more hydrophobic bulk stationary phase, because the positive nitrogen head group is partially incorporated into the C18 phase. In the first MLC reports, mobile phases were proposed containing only waterbased surfactant solutions and, occasionally, buffering salts compound. As early as 1983, Dorsey et al. [84] recommended the addition of a short-chain alcohol (e.g., 1-propanol or butanol), to the micelle-based eluent to enhance efficiency. Then, the potential use of organic solvents to increase the elution strength of mobile phases was investigated. Propanol is the most usual alcohol additive in MLC applications. More recently, acetonitrile, a common solvent in aqueous-organic RPLC, has drawn some attention. This kind of organic modifier added to MLC mobile phases (i) lowers the polarity of the aqueous solution, (ii) alters the micelle structure, and (iii) acts on the stationary phase changing the amount of surface-adsorbed surfactant. Also, organic solvent molecules wet the bonded phase, changing its physicochemical structure (rigidity) and hydrophobicity. Hybrid micellar-organic mobile phases were found to have great potential, and most applications in MLC are done with them today. Although the separation mode is still predominantly micellar in nature, the micelle itself is perturbed by the organic solvent, which can cause changes in micellar parameters, such as the CMC and the surfactant aggregation number. Indeed, a high percentage of organic solvent is not desirable, because the role of the micelle as a modifier would be decreased or null. Mobile phases with high percentages of organic solvent may also flush away the surfactant molecules adsorbed on the surface of the bonded phase. The concentration of organic solvent that usually preserves micelle integrity is approximately 15% for propanol and acetonitrile, 10% for butanol, and 6% for pentanol [85].

12.8 Conclusions

FIGURE 12.20 Three-phase systems in MLC: (A) C18-bonded silica-SDS; (B) C18-bonded silica-CTAB; (C) bare silica-CTAB. Adapted with permission from Ruiz-A´ngel et al. [81].

12.8 Conclusions In the present chapter, we have described some examples about the importance that certain additives in the mobile phase as well as some co-solvents can have to improve the LC analysis of compounds belonging to different chemical classes. Accordingly, we have provided evidence about the impact that some additive and co-solvents can have on both the thermodynamic and the kinetic features of the LC process. In all the sub-sections of the chapters, it appears readily evident that mobile phase should not be regarded as a passive transporter of the analyte(s) along the column, but it is instead intimately involved in the overall chromatographic process at multiple levels. A careful optimization of the eluent composition is therefore a quintessential step in the setting of research and development programs.

315

316

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

References [1] Karlsson G, Gellerfors P, Persson A, Noren B, Edlund PO, Sandberg C, Birnbaum S. Separation of oxidized and deamidated human growth hormone variants by isocratic reversed-phase high-performance liquid chromatography. J Chromatogr A 1999;855:147–55. https://doi.org/10.1016/s0021-9673(99)00669-x. [2] Chaimbault P, Petritis K, Elfakir C, Dreux M. Determination of 20 underivatized proteinic amino acids by ion-pairing chromatography and pneumatically assisted electrospray mass spectrometry. J Chromatogr A 1999;855:191–202. https://doi.org/10.1016/s00219673(99)00685-8. [3] Chaimbault P, Petritis K, Elfakir C, Dreux M. Ion-pair chromatography on a porous graphitic carbon stationary phase for the analysis of twenty underivatized protein amino acids. J Chromatogr A 2000;870:245–54. https://doi.org/10.1016/s0021-9673(99) 00863-8. [4] Adoubel AA, Guenu S, Elfakir C, Dreux M. Separation of underivatized small peptides on a porous graphitic carbon column by ion-pair chromatography and evaporative light scattering detection. J Liq Chromatogr Relat Technol 2000;23:2433–46. https://doi.org/ 10.1081/JLC-100100499. [5] Flieger J. Application of perfluorinated acids as ion-pairing reagents for reversed-phase chromatography and retention-hydrophobicity relationships studies of selected betablockers. J Chromatogr A 2010;1217:540–9. https://doi.org/10.1016/j. chroma.2009.11.083. [6] Yamamoto E, Ishihama Y, Asakawa N. Application of partially fluorinated carboxylic acids as ion-pairing reagents in LC/ESI-MS. Talanta 2014;127:219–24. https://doi.org/ 10.1016/j.talanta.2014.03.060. [7] Lajin B, Goessler W. Fluorinated carboxylic acids as “ion repelling agents” in reversedphase chromatography. J Chromatogr A 2020;1631, 461575. https://doi.org/10.1016/j. chroma.2020.461575. [8] Lajin B, Goessler W. Fluoroalkylamines: novel, highly volatile, fast-equilibrating, and electrospray ionization–mass spectrometry signal-enhancing cationic ion-interaction reagents. Anal Chem 2020;92:10121–8. https://doi.org/10.1021/acs.analchem.0c02138. [9] Kipper K, Herodes K, Leito I. Fluoroalcohols as novel buffer components for basic buffer solutions for liquid chromatography electrospray ionization mass spectrometry: retention mechanisms. J Chromatogr A 2011;1218:8175–80. https://doi.org/10.1016/j. chroma.2011.09.025. [10] Basiri B, van Hattum H, van Dongen WD, Murph MM, Bartlett MG. The role of fluorinated alcohols as Mobile phase modifiers for LC-MS analysis of oligonucleotides. J Am Soc Mass Spectrom 2017;28:190–9. https://doi.org/10.1007/s13361-016-1500-3. [11] Veigure R, Lossmann K, Hecht M, Parman E, Born R, Leito I, Herodes K, Kipper K. Retention of acidic and basic analytes in reversed phase column using fluorinated and novel eluent additives for liquid chromatography-tandem mass spectrometry. J Chromatogr A 2020;1613, 460667. https://doi.org/10.1016/j.chroma.2019.460667. [12] He L, Zhang W, Zhao L, Liu X, Jiang S. Effect of 1-alkyl-3-methylimidazolium-based ionic liquids as the eluent on the separation of ephedrines by liquid chromatography. J Chromatogr A 2003;1007:39–45. https://doi.org/10.1016/s0021-9673(03)00987-7. [13] Marszałł MP, Ba˛czek T, Kaliszan R. Reduction of silanophilic interactions in liquid chromatography with the use of ionic liquids. Anal Chim Acta 2005;547:172– 8. https://doi.org/10.1016/j.aca.2005.05.045.

References

[14] Cui G, Yu H, Ma Y-J. Ionic liquids as mobile phase additives for determination of thiocyanate and iodide by liquid chromatography. J Sep Sci 2019;42:1733–9. https://doi.org/ 10.1002/jssc.201801277. [15] Liu Y, Ma Y, Yu H, An Y. Rapid determination of pyrrolidinium cations by ion-pair chromatography with imidazolium ionic liquids. J Chromatogr Sci 2018;56:202– 8. https://doi.org/10.1093/chromsci/bmx100. [16] Tan T, Zhang M, Wan Y, Qiu H. Utilization of deep eutectic solvents as novel mobile phase additives for improving the separation of bioactive quaternary alkaloids. Talanta 2016;149:85–90. https://doi.org/10.1016/j.talanta.2015.11.041. [17] Gao F, Liu L, Tang W, Row KH, Zhu T. Optimization of the chromatographic behaviors of quercetin using choline chloride-based deep eutectic solvents as HPLC mobilephase additives. Sep Sci Technol 2018;53:397–403. https://doi.org/10.1080/ 01496395.2017.1388257. [18] Sutton AT, Fraige K, Leme GM, da Silva Bolzani V, Hilder EF, Cavalheiro AJ, Arrua RD, Funari CS. Natural deep eutectic solvents as the major mobile phase components in high-performance liquid chromatography-searching for alternatives to organic solvents. Anal Bioanal Chem 2018;410:3705–13. https://doi.org/10.1007/s00216-018-1027-5. [19] Tan T, Qiao X, Wan Y, Qiu H. Deep eutectic solvent: a new kind of mobile phase modifier for hydrophilic interaction liquid chromatography. Se Pu 2015;33:934–7. https://doi. org/10.3724/sp.j.1123.2015.05002. [20] Loeser E, Babiak S, DelaCruz M, Karpinski P. Triflic acid and sodium triflate as chaotropic Mobile phase additives in RP-LC. J Chromatogr Sci 2011;49:57–62. https://doi. org/10.1093/chrsci/49.1.57. [21] Flieger J, Siwek A, Pizon M. Usefulness of chaotropic salt additive in RP-HPLC of organic nonionized compounds. J Sep Sci 2013;36:469–76. https://doi.org/10.1002/ jssc.201200798. [22] Shulyak N, Piponski M, Kovalenko S, Stoimenova TB, Drapak I, Piponska M, Rezk MR, Abbeyquaye AD, Oleshchuk O, Logoyda L. Chaotropic salts impact in HPLC approaches for simultaneous analysis of hydrophilic and lipophilic drugs. J Sep Sci 2021;44:2908–16. https://doi.org/10.1002/jssc.202100168. [23] Florez C, Kazakevich Y. Influence of ionic mobile phase additives with low charge delocalization on the retention of ionic analytes in reversed-phase HPLC. J Liq Chromatogr Relat Technol 2013;36:1138–48. https://doi.org/10.1080/10826076.2012.670183. [24] Alpert AJ. Effect of salts on retention in hydrophilic interaction chromatography. J Chromatogr A 2018;1538:45–53. https://doi.org/10.1016/j.chroma.2018.01.038. [25] El-Shaheny RN, El-Maghrabey MH, Belal FF. Micellar liquid chromatography from green analysis perspective. Open Chem 2015;13. https://doi.org/10.1515/chem-20150101. [26] Fischer J, Jandera P. Chromatographic behaviour in reversed-phase high-performance liquid chromatography with micellar and submicellar mobile phases: effects of the organic modifier. J Chromatogr B Biomed Appl 1996;681:3–19. https://doi.org/ 10.1016/0378-4347(95)00538-2. [27] Peris-Garcı´a E, Rodrı´guez-Martı´nez J, Baeza-Baeza JJ, Garcı´a-Alvarez-Coque MC, Ruiz-Angel MJ. Search of non-ionic surfactants suitable for micellar liquid chromatography. Anal Bioanal Chem 2018;410:5043–57. https://doi.org/10.1007/s00216-0181161-0. [28] Ianni F, Lisanti A, Marinozzi M, Camaioni E, Pucciarini L, Massoli A, Sardella R, Concezzi L, Natalini B. Hydrophobic amino acid content in onions as potential fingerprints

317

318

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39] [40]

[41]

[42]

[43]

of geographical origin: the case of Rossa da Inverno sel. Rojo Duro. Molecules 2018;23: E1259. https://doi.org/10.3390/molecules23061259. Sardella R, Lisanti A, Marinozzi M, Ianni F, Natalini B, Blanch GP, Ruiz del Castillo ML. Combined monodimensional chromatographic approaches to monitor the presence of d-amino acids in cheese. Food Control 2013;34:478–87. https://doi.org/10.1016/j. foodcont.2013.05.026. de Person M, Chaimbault P, Elfakir C. Analysis of native amino acids by liquid chromatography/electrospray ionization mass spectrometry: comparative study between two sources and interfaces. J Mass Spectrom 2008;43:204–15. https://doi.org/10.1002/ jms.1287. Petritis K, Chaimbault P, Elfakir C, Dreux M. Parameter optimization for the analysis of underivatized protein amino acids by liquid chromatography and ionspray tandem mass spectrometry. J Chromatogr A 2000;896:253–63. https://doi.org/10.1016/s0021-9673 (00)00582-3. Bassler BJ, Kaliszan R, Hartwick RA. Retention mechanisms on metallic stationary phases. J Chromatogr A 1989;461:139–47. https://doi.org/10.1016/S0021-9673(00) 94283-3. Forga´cs E, Cserha´ti T. Retention of some ring-substituted aniline derivatives by porous graphitized carbon. Dependence on physico-chemical parameters. Chromatographia 1992;33:356–60. https://doi.org/10.1007/BF02275918. Berthod A, Ruiz-Angel MJ, Huguet S. Nonmolecular solvents in separation methods: dual nature of room temperature ionic liquids. Anal Chem 2005;77:4071–80. https:// doi.org/10.1021/ac050304+. Garcı´a-Alvarez-Coque MC, Ruiz-Angel MJ, Berthod A, Carda-Broch S. On the use of ionic liquids as mobile phase additives in high-performance liquid chromatography. A review. Anal Chim Acta 2015;883:1–21. https://doi.org/10.1016/j.aca.2015.03.042. Li L, Liu Y, Wang Z, Yang L, Liu H. Development and applications of deep eutectic solvent derived functional materials in chromatographic separation. J Sep Sci 2021;44:1098–121. https://doi.org/10.1002/jssc.202000523. Li G, Zhu T, Lei Y. Choline chloride-based deep eutectic solvents as additives for optimizing chromatographic behavior of caffeic acid. Korean J Chem Eng 2015;32:2103– 8. https://doi.org/10.1007/s11814-015-0054-6. Makarov A, LoBrutto R, Kazakevich Y. Liophilic Mobile phase additives in reversed phase HPLC. J Liq Chromatogr Relat Technol 2008;31:1533–67. https://doi.org/ 10.1080/10826070802125918. Hofmeister F. Zur Lehre von der Wirkung der Salze, Archiv f experiment. Pathol u Pharmakol 1888;24:247–60. https://doi.org/10.1007/BF01918191. Flieger J. The effect of chaotropic mobile phase additives on the separation of selected alkaloids in reversed-phase high-performance liquid chromatography. J Chromatogr A 2006;1113:37–44. https://doi.org/10.1016/j.chroma.2006.01.090. Kazakevich YV, LoBrutto R, Vivilecchia R. Reversed-phase high-performance liquid chromatography behavior of chaotropic counteranions. J Chromatogr A 2005;1064:9– 18. https://doi.org/10.1016/j.chroma.2004.11.104. Flieger J. Effect of mobile phase composition on the retention of selected alkaloids in reversed-phase liquid chromatography with chaotropic salts. J Chromatogr A 2007;1175:207–16. https://doi.org/10.1016/j.chroma.2007.10.036. Roberts JM, Diaz AR, Fortin DT, Friedle JM, Piper SD. Influence of the Hofmeister series on the retention of amines in reversed-phase liquid chromatography. Anal Chem 2002;74:4927–32. https://doi.org/10.1021/ac0256944.

References

[44] Flieger J, Siwek A, Pizon M. Usefulness of chaotropic salt additive in RP-HPLC of organic nonionized compounds. J Sep Sci 2013;36:469–76. https://doi.org/10.1002/ jssc.201200798. [45] Zhao H. Are ionic liquids kosmotropic or chaotropic? An evaluation of available thermodynamic parameters for quantifying the ion kosmotropicity of ionic liquids. J Chem Technol Biotechnol 2006;81:877–91. https://doi.org/10.1002/jctb.1449. [46] Komsta Ł, Skibinski R, Bojarczuk A, Rado n M. Salting-out chromatography—a practical review. Acta Chromatogr 2011;23:191–203. https://doi.org/10.1556/ achrom.23.2011.2.1. [47] Sˇtrop P. Hydrophobic chromatography of proteins on semi-rigid gels: effect of salts and interferents on the retention of proteins by spheron P 300. J Chromatogr A 1984;294:213–21. https://doi.org/10.1016/S0021-9673(01)96128-X. [48] Kou R, Zhang J, Wang T, Liu G. Interactions between polyelectrolyte brushes and Hofmeister ions: Chaotropes versus Kosmotropes. Langmuir 2015;31:10461–8. https://doi. org/10.1021/acs.langmuir.5b02698. [49] Staahlberg J, Joensson B, Horvath C. Theory for electrostatic interaction chromatography of proteins. Anal Chem 1991;63:1867–74. https://doi.org/10.1021/ac00017a036. [50] Gama MS, Santos MS, Lima ERA, Tavares FW, Barreto AGB. A modified PoissonBoltzmann equation applied to protein adsorption. J Chromatogr A 2018;1531:74– 82. https://doi.org/10.1016/j.chroma.2017.11.022. [51] Pegram LM, Record MT. Thermodynamic origin of Hofmeister ion effects. J Phys Chem B 2008;112:9428–36. https://doi.org/10.1021/jp800816a. [52] Henry M. Hofmeister series: the quantum mechanical viewpoint. Curr Opin Colloid Interface Sci 2016;23:119–25. https://doi.org/10.1016/j.cocis.2016.08.001. [53] Pollard TP, Beck TL. Toward a quantitative theory of Hofmeister phenomena: from quantum effects to thermodynamics. Curr Opin Colloid Interface Sci 2016;23:110– 8. https://doi.org/10.1016/j.cocis.2016.06.015. [54] Xie WJ, Gao YQ. A simple theory for the Hofmeister series. J Phys Chem Lett 2013;4:4247–52. https://doi.org/10.1021/jz402072g. [55] Zhang Y, Cremer PS. Interactions between macromolecules and ions: the Hofmeister series. Curr Opin Chem Biol 2006;10:658–63. https://doi.org/10.1016/j. cbpa.2006.09.020. [56] Grover PK, Ryall RL. Critical appraisal of salting-out and its implications for chemical and biological sciences. Chem Rev 2005;105:1–10. https://doi.org/10.1021/cr030454p. [57] Ewonde RE, De Vos J, Broeckhoven K, Eβer D, Eeltink S. Assessment of the resolving power of hydrophobic interaction chromatography for intact protein analysis on nonporous butyl polymethacrylate phases. J Chromatogr A 2021;1651, 462310. https:// doi.org/10.1016/j.chroma.2021.462310. [58] Sta˚hlberg J, J€onsson B, Horva´th C. Combined effect of coulombic and van der Waals interactions in the chromatography of proteins. Anal Chem 1992;64:3118–24. https:// doi.org/10.1021/ac00048a009. [59] Tsumoto K, Ejima D, Senczuk AM, Kita Y, Arakawa T. Effects of salts on protein–surface interactions: applications for column chromatography. J Pharm Sci 2007;96:1677–90. https://doi.org/10.1002/jps.20821. [60] Liapis AI, Grimes BA. The coupling of the electrostatic potential with the transport and adsorption mechanisms in ion-exchange chromatography systems: theory and experiments. J Sep Sci 2005;28:1909–26. https://doi.org/10.1002/jssc.200500240. [61] Hyde AM, Zultanski SL, Waldman JH, Zhong Y-L, Shevlin M, Peng F. General principles and strategies for salting-out informed by the Hofmeister series. Org Process Res Dev 2017;21:1355–70. https://doi.org/10.1021/acs.oprd.7b00197.

319

320

CHAPTER 12 Co-solvents and mobile phase additives in HPLC

[62] Mazzini V, Craig VSJ. Specific-ion effects in non-aqueous systems. Curr Opin Colloid Interface Sci 2016;23:82–93. https://doi.org/10.1016/j.cocis.2016.06.009. [63] Zhou H-X. Interactions of macromolecules with salt ions: an electrostatic theory for the Hofmeister effect, proteins: structure. Funct Bioinform 2005;61:69–78. https://doi.org/ 10.1002/prot.20500. [64] Fekete S, Veuthey J-L, Beck A, Guillarme D. Hydrophobic interaction chromatography for the characterization of monoclonal antibodies and related products. J Pharm Biomed Anal 2016;130:3–18. https://doi.org/10.1016/j.jpba.2016.04.004. [65] Cecchi T. Retention mechanism for ion-pair chromatography with chaotropic reagents. From ion-pair chromatography toward a unified salt chromatography. Adv Chromatogr 2011;49:1–35. https://doi.org/10.1201/b10721-2. [66] Wu S-L, Karger BL. Hydrophobic interaction chromatography of proteins. In: Methods in enzymology. Academic Press; 1996. p. 27–47. https://doi.org/10.1016/S0076-6879 (96)70004-6. [67] Cusumano A, Guillarme D, Beck A, Fekete S. Practical method development for the separation of monoclonal antibodies and antibody-drug-conjugate species in hydrophobic interaction chromatoraphy, part 2: optimization of the phase system. J Pharm Biomed Anal 2016;121:161–73. https://doi.org/10.1016/j.jpba.2016.01.037. [68] Tyteca E, De Vos J, Tassi M, Cook K, Liu X, Kaal E, Eeltink S. Generic approach to the method development of intact protein separations using hydrophobic interaction chromatography. J Sep Sci 2018;41:1017–24. https://doi.org/10.1002/jssc.201701202. [69] Snyder LR, Dolan JW. Gradient elution fundamentals. In: High-performance gradient elution. John Wiley & Sons, Ltd; 2006. p. 23–73. https://doi.org/10.1002/ 9780470055526.ch2. [70] Rodriguez-Aller M, Guillarme D, Beck A, Fekete S. Practical method development for the separation of monoclonal antibodies and antibody-drug-conjugate species in hydrophobic interaction chromatography, part 1: optimization of the mobile phase. J Pharm Biomed Anal 2016;118:393–403. https://doi.org/10.1016/j.jpba.2015.11.011. [71] Rosen MJ, Kunjappu JT. Surfactants and interfacial phenomena. 4th Edition. New York: John Wiley & Sons Ltd.; 2012. https://doi.org/10.1002/9781118228920. [72] Takei K, Tsuto K, Miyamoto S, Wakatsuki J. Anionic surfactants: lauric products. J Am Oil Chem Soc 1985;62:341–7. https://doi.org/10.1007/BF02541402. [73] Taxvig C, Rosenmai AK, Vinggaard AM. Polyfluorinated alkyl phosphate ester surfactants – current knowledge and knowledge gaps. Basic Clin Pharmacol Toxicol 2014;115:41–4. https://doi.org/10.1111/bcpt.12208. [74] Tadros T. Surfactants. In: Tadros T, editor. Encyclopedia of colloid and Interface science. Berlin, Heidelberg: Springer; 2013. p. 1242–90. https://doi.org/10.1007/978-3642-20665-8_40. [75] Sarkar B, Alexandridis P. Alkyl propoxy ethoxylate “graded” surfactants: micelle formation and structure in aqueous solutions. J Phys Chem B 2010;114:4485–94. https://doi. org/10.1021/jp910939q. [76] Sarkar R, Pal A, Rakshit A, Saha B. Properties and applications of amphoteric surfactant: a concise review. J Surfact Deterg 2021;24:709–30. https://doi.org/10.1002/jsde.12542. [77] Pichot R, Watson R, Norton I. Phospholipids at the interface: current trends and challenges. Int J Mol Sci 2013;14:11767–94. [78] Al-Soufi W, Pin˜eiro L, Novo M. A model for monomer and micellar concentrations in surfactant solutions: application to conductivity, NMR, diffusion, and surface

References

[79] [80]

[81]

[82]

[83]

[84] [85]

tension data. J Colloid Interface Sci 2012;370:102–10. https://doi.org/10.1016/j. jcis.2011.12.037. Pharr DY. Green analytical chemistry – the use of surfactants as a replacement of organic solvents in spectroscopy. Phys Sci Rev 2017;2. https://doi.org/10.1515/psr-2017-0006. Armstrong DW, Henry SJ. Use of an aqueous micellar Mobile phase for separation of phenols and polynuclear aromatic hydrocarbons via HPLC. J Liquid Chromatogr 1980;3:657–62. https://doi.org/10.1080/01483918008060181. ´ ngel MJ, Garcı´a-A ´ lvarez-Coque MC, Berthod A. New insights and recent develRuiz-A opments in micellar liquid chromatography. Sep Purif Rev 2009;38:45–96. https://doi. org/10.1080/15422110802178876. Stępnik KE. A concise review of applications of micellar liquid chromatography to study biologically active compounds. Biomed Chromatogr 2017;31. https://doi.org/10.1002/ bmc.3741. Mao C, McGill KE, Tucker SA. Optimization of micellar liquid chromatographic separation of polycyclic aromatic hydrocarbons with the addition of second organic additive. J Sep Sci 2004;27:991–6. https://doi.org/10.1002/jssc.200301737. Dorsey JG, DeEchegaray MT, Landy JS. Efficiency enhancement in micellar liquid chromatography. Anal Chem 1983;55:924–8. https://doi.org/10.1021/ac00257a024. Ruiz-Angel MJ, Caballero RD, Simo´-Alfonso EF, Garcı´a-Alvarez-Coque MC. Micellar liquid chromatography: suitable technique for screening analysis. J Chromatogr A 2002;947:31–45. https://doi.org/10.1016/s0021-9673(01)01595-3.

321

CHAPTER

Method development in liquid chromatography

13

John W. Dolan and Lloyd R. Snyder† LC Resources, McMinnville, OR, United States

There are many different ways to develop a new high-performance liquid chromatography (HPLC) method, and one is not necessarily better than another as long as it attains the goals of the developer in a timely manner. Here, we share one approach based on our experience in a method development environment and years of interacting with clients in the pharmaceutical industry and working with widely accepted scientific principles. It should be noted that method development for liquid chromatography is a complex process that involves a number of steps, so this discussion should be considered an overview rather than a step-by-step instruction manual. For in-depth information, refer to Refs. [1–3], as well as the scientific literature and column manufacturer’s technical notes.

13.1 Introduction In recent years, the pharmaceutical industry has been applying quality by design (QbD) to various tasks in the laboratory and manufacturing environment. QbD is based on an ICH (International Committee on Harmonization) document [4], which states that to have a high-quality product (e.g., method), quality must be designed into the product, not tested into it. Another concept of QbD is the design space, which is the multidimensional space of operational variables within which a method is valid and does not require revalidation. In practical terms, changes in conditions are allowed within the design space to meet system suitability requirements (without revalidating the method). For HPLC, the design space encompasses the range of allowed values of various conditions (% of organic solvent, pH, °C, etc.) where changes in any combination of these variables are allowed. This approach provides the flexibility to adjust a method to restore performance, if necessary. A requirement †

Deceased.

Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00032-1 Copyright # 2023 Elsevier Inc. All rights reserved.

323

324

CHAPTER 13 Method development in liquid chromatography

of QbD is that the effects of various conditions on the separation must be defined so that the design space limits can be identified. QbD, although new in name, is not a new concept to experienced chromatographers. It does, however, provide a practical organizational structure for method development, which we apply in this chapter. The method development process comprises six consecutive steps: 1. 2. 3. 4. 5. 6.

Define the goals of the method (Section 13.2). Determine the method development approach (Section 13.3). Develop the method (Section 13.4). Perform prevalidation experiments (Section 13.5). Validate the method (Section 13.6). Document the process (Section 13.7).

We emphasize step 2 but also examine each step in more or less detail. Reversedphase separation is assumed unless otherwise noted. A survey of the scientific literature, existing in-house methods, or other resources may provide leads on how to proceed with a particular sample. Such information can be helpful at the outset, but be careful because information about method robustness is rarely available; also, starting method development from a poorly developed existing method seldom is a good approach. It is better to use available information for choosing starting conditions, such as the initial column, organic solvent, and (maybe) mobile-phase pH.

13.2 Goals Before method development can start, the goals (or practical application) of the method must be delineated. Related questions may include the following: • • • • • •

How will the method be used (research, production, quality control, random generic samples, high throughput, etc.)? Who will use the method (location, training, special communication problems, etc.)? What are the chromatographic goals (resolution, run time, number of samples to analyze per batch, detection and quantification limits, linearity, range, etc.)? Are any restrictions or limitations placed on the method (laboratory environment, isocratic only, UV detection only, etc.)? What level of validation is required (R&D method, regulatory approval, etc.)? Are sufficient resources available for adequate method development (time, personnel, budget, equipment, etc.)?

For example, a method for the content assay of a pharmaceutical product for regulatory purposes has different requirements than a method used to support a synthetic chemist or one used for in-process monitoring. Once the goals are established, the method development process can proceed.

13.3 A structured approach to method development

13.3 A structured approach to method development Adequate resolution, Rs, between adjacent peaks of interest is the primary goal of most HPLC methods. For method development, a fundamental resolution equation for isocratic separation can serve as a useful guide: Rs ¼ 0:25 N 0:5 ½k1 =ð1 + k1 Þðα  1Þ, i

ii

(13.1)

iii

where N is the column plate number, k1 is the retention factor, k, for the first peak, and α is the separation factor (selectivity): α ¼ k2 =k1 ,

(13.2)

where k2 is the retention factor of the second peak. We recommend using Eq. (13.1) as a guide for method development. First, start with a column that has an adequate value of N. The value N  10,000 is recommended unless other factors suggest larger or smaller N values. Usually, a C8 or C18 column is chosen at the start because these columns often can provide a successful separation (see Section 13.4.2 for column-type screening). Next, use a gradient scouting run to determine whether isocratic or gradient conditions should be used (Section 15.3 and the discussion of Fig. 15.7 in Chapter 15). The adjustment of either the isocratic percent of the organic solvent, %B, or gradient time, tG, may be sufficient to obtain the desired separation. If greater resolution is needed, explore each of the various factors that influence α. The simplest approach is to use a combination of tG (or %B) and temperature (°C) (e.g., Fig. 15.8), then change solvent or column type if necessary. It is usually prudent to select a pH (e.g., pH 2.5) for initial experiments and reserve changes in pH for later. After changes in α have been explored, the value of N can be revisited. If there is excess resolution, the run time (and N) can be reduced by using a shorter column and increased flow rate. Conversely, a limited increase (generally, no more than 25–40%) in Rs can be gained by using a longer or smaller particle column. Additional choices are discussed in Section 13.4 and Chapter 15.

13.3.1 Column plate number, N: Term i of Eq. (13.1) For most separations, values of N fall within a range of 5000  N  20,000, corresponding to a maximum twofold change in resolution. Larger values of N require longer run times, so changes in α are often preferable. N increases for longer columns, smaller particles, and lower flow rates—but flow rate usually has a relatively small effect on plate number and resolution. For conventional HPLC operation with a maximum pressure of 400 bar (6000 psi), a 100  4.6-mm column packed with 3-μm particles represents a good starting point (N ¼ 10,000). For ultrahigh–performance liquid chromatography (UHPLC) operation and a maximum pressure of 1000 bar, shorter columns with smaller particles ( 10, excessive run times and undesirable peak broadening can occur. When the range of k values exceeds 0.5 < k < 20, gradient elution is usually recommended (Chapter 15). The retention factor is controlled most easily by adjustment of the mobile-phase strength (% B solvent). For isocratic conditions, this can be achieved by progressively reducing %B in a sequence of 90% B, 80% B, 70% B, and so on until the desired k range is reached. An alternative approach is to use gradient scouting runs (Sections 13.3.4 and 15.3). Fine-tuning k often provides additional benefits (Section 13.3.3).

13.3.3 Selectivity, α: Term iii of Eq. (13.1) Selectivity, which defines the spacing of two peaks, is influenced by different chromatographic variables. Unfortunately, without prior knowledge (experimental data, sample-structure information, etc.), it is not possible to predict the influence of a particular variable on α for a given pair of peaks. It is possible, however, to make general statements about the influence of different variables on α. One such study examined 67 chemically diverse solutes in this regard, with the results summarized in Table 13.1. The study determined the average change in α (jδ log αj) for the sample

Table 13.1 Comparison of orthogonal power of chromatographic variables. Variablea

Change

Example

%B tG °C ACN (MeOH) Column pH [Buffer]

10% 3 20 °C To MeOH (ACN)

50% ACN to 60% ACN 10–30 min 35–45°C Replace ACN by MeOH (or vice versa)

Fs > 65c; Fs > 100d 5 units 2

pH 2.5–7.5 25–50 mM

Orthogonal power (OP)b 0.08 0.07 0.07 0.20 0.19 ≫0.7e 0.02

%B, %-organic solvent; tG, gradient time; °C, column temperature;[buffer], concentration of buffer. Average jδ log αj; OP  0.1 needed for “orthogonal” conditions. c Fs ¼ F value in Ref. [5]; for ionic or ionizable compounds. d For nonionized compounds. e Ionic samples only. a

b

13.3 A structured approach to method development

set for a defined change in a variable; we refer to this as the orthogonal power (OP) for that variable. If OP 0.1, it is likely that a significant change in selectivity will occur. This, of course, does not guarantee the separation of any particular peak pair, but it is a good starting point. We can approximately rank the values of OP: ½bufferðleast effectiveÞ ≪ %B  tG  °C < solvent type  column type ≪ pHðmost effectiveÞ

where tG is the gradient time (Chapter 15). The OP values of different variables will be examined next (in the order presented in Table 13.1). •







%B, tG. According to the linear-solvent-strength model [3], %B and tG (or gradient steepness) are equivalent variables for controlling a separation. A change of 10% B (e.g., from 50% ACN to 60% ACN) changes k values by about 2.5-fold (Ref. [2, p. 58]). Similarly, a 2.5-fold change in tG (e.g., from a 10-min to a 25-min gradient) changes retention about 2.5-fold. Either such change has OP  0.07–0.08 (Table 13.1), slightly less than the target minimum of OP 0.1. However, these variables are easy to change while maintaining k values in an acceptable range (Section 13.3.2). Furthermore, changes in %B or tG may provide sufficient changes in α to obtain adequate Rs. For these reasons, we recommend that %B or tG should be investigated early in the method development process, despite their relatively lower OP values. °C. The value of OP ¼ 0.07 for a 20°C change in column temperature (Table 13.1) suggests that the temperature is somewhat limited in its ability to increase resolution. However, for partially ionized solutes, a change in column temperature can have a dramatic effect on selectivity [6]. Furthermore, the nature of the selectivity change for °C may be different than that of %B or tG so that a combined change of °C and %B or tG may be especially effective. The convenience of temperature changes leads us to recommend simultaneous changes in °C and either %B or tG at an early stage in method development (in this connection, see also Fig. 15.8 in Chapter 15). The column temperature should be controlled in all cases (usually slightly above room temperature, e.g., 30–35°C). Solvent type. A change in the B solvent (e.g., methanol, MeOH vs acetonitrile, ACN) can be effective in changing α during method development. According to Table 13.1, replacing ACN with MeOH (or vice versa) has OP ¼ 0.2, double the minimum desired OP 0.1. Any of the three popular organic solvents [ACN, MeOH, and tetrahydrofuran, (THF)] can be blended for improved control of selectivity [6]. One approach to method development is to screen two or more solvents early in the development process to see which one separates more peaks. Then, the chosen solvent can be fine-tuned (as previously) by adjusting %B or tG; the use of mixtures of two or more B solvents can also be considered. Column type. For years, it has been known that changing from one column type to another (e.g., C18 to cyano) can result in a significant change in α; however,

327

328

CHAPTER 13 Method development in liquid chromatography





changing from one C18 column to another can sometimes also provide an adequate change in selectivity. Recent developments (the hydrophobicsubtraction model [7]) have led to a better understanding of column selectivity, as well as its implementation by means of free column comparison software (USP-PQRI database [5]). Using the latter software, we can identify similar (equivalent) or different (orthogonal) columns by means of a derived comparison function, Fs. Two columns with Fs  3 can be assumed to be equivalent. As Fs increases, the columns become more different. For a maximum change in column selectivity (OP  0.19 in Table 13.1), a value of Fs > 65 is sufficient for ionizable solutes (acids or bases), while Fs > 100 is adequate for neutral or nonionized compounds. pH. A change in mobile-phase pH can be one of the most powerful ways to change α if the analytes are ionizable. For such samples, a 5-unit change in pH (e.g., pH 2.5–7.5) can have OP > 0.7 (Table 13.1); an operating range of 2 < pH < 8 generally is advised for silica-based columns. At pH 8, the silica dissolves. For most columns, a low pH buffer of 2.5  pH  3 is a good starting place. Low pH suppresses the ionization of column silanols and acidic analytes, providing a better peak shape. Many basic analytes have sufficiently high pKa values that they remain ionized at pH 8, several manufacturers offer silica-based columns that are stable at pH >8. Buffer concentration. For most reversed-phase separations, a change in mobilephase buffer concentration has little effect on selectivity (OP ¼ 0.02, Table 13.1). Exceptions exist for mixed mode or HILIC separations (Chapter 5), where ionic or electrostatic interactions play a significant role in the separation. A buffer concentration of 5–10 mM (measured in the total mobile phase) is recommended. Higher buffer concentrations (e.g., >50 mM) can result in buffer solubility problems.

13.3.4 Gradient elution Many samples have a sufficiently wide polarity range that 1 < k < 20 is not possible for any isocratic condition. Furthermore, even when isocratic separation is possible, identifying those conditions by stepwise changes in %B can be time consuming. An initial gradient separation is instead recommended prior to method development to determine whether isocratic separation is possible—and if so, what %B provides 1 < k < 20 for the sample. A free calculator [8] can use the results of this initial gradient to determine approximate isocratic separation conditions. If only gradient elution is feasible, the calculator also can be used to trim “wasted” time off the beginning or end of the gradient. Resolution-modeling software (Section 13.4.1) can further increase the information content of a limited number of experimental runs. We recommend starting method development with gradient runs that can be used with resolution-modeling software.

13.4 Method development in practice

13.4 Method development in practice Implementation of the method development approach of Section 13.3 involves several additional choices, as presented in this section. The method development process should represent the best compromise among the factors that affect method development for a given sample.

13.4.1 Resolution-modeling software A linear relationship exists between retention (log k) and mobile-phase %B: log k ¼ a + b%B

(13.3)

where a and b are constants for a given solute and separation conditions. Similar relationships exist between values of k and °C; other curve fits can be used to describe the relationship between k and other variables (pH, ion-pair-reagent concentration, etc.). These relationships allow an accurate prediction of retention as a function of separation conditions, based on two or more experimental measurements for changes in each condition. This in turn allows predictions of Rs for simultaneous changes in one to three variables, such as temperature and gradient time. It is convenient to display the results of such calculations as resolution maps, where Rs is plotted vs one to three conditions, using resolution-modeling (“computer simulation”) software (e.g., DryLab, Molnar Institute, Berlin). Using data from the initial “calibration runs,” resolution maps allow optimum conditions for a separation to be determined quickly. So, for example, 12 experimental runs (2 tG values  2°C values  3 pH values) can give a three-dimensional model (cube) allowing prediction of Rs under any combination of these three variables, as well as any isocratic %B–°C–pH combination. Thus, just a few runs can answer the following questions: • • • •

Can an adequate separation be obtained using the tested variables? If so, what conditions should be used? How sensitive is the separation to each (or a combination of) variables? What conditions should be tested to demonstrate the robustness in QbD (Sections 13.5–13.6)?

A further benefit of resolution-modeling software is that it requires high-quality input data for accurate predictions. This adds discipline to the method development process so that, even if the software is not used, the quality of the experimental data—and the results of method development—tend to be better. We strongly recommend using resolution-modeling software during method development for both improved productivity and higher-quality methods.

13.4.2 Priority of column screening All the variables listed in Table 13.1 can be varied in a continuous manner—except column selectivity. Optimization of these “continuous” variables can be achieved by incrementally changing the variable or using the resolution-modeling software

329

330

CHAPTER 13 Method development in liquid chromatography

(Section 13.4.1). Column selection, on the other hand, requires a choice between one column and another—columns cannot be blended conveniently for intermediate results. Historically, the predictability of differences in column selectivity was poor, so successfully changing a separation by changing columns was often more luck than skill. Today, column selectivity differences can be predicted (Section 13.3.3), improving the chance of changing a separation by using a different column. This leads to two general approaches: •



Screen continuous variables first. This is the traditional approach, where a single column is chosen and then a systematic investigation of other variables (e.g., tG, and °C, pH) is performed. This approach is easily automated with modern HPLC equipment, and the number of experiments can be reduced if resolution-modeling software is used. This can be an efficient way to conduct method development. Column screening first. An alternative approach is to screen two or more columns of different selectivity at the beginning of the method development process, to pick a column for further method development. The problem with this approach historically is that candidate columns were chosen for reasons that may not have reflected the orthogonal nature of the column; each lab had a favorite column set but often could not offer a solid rationale for the selection. With recent advances in the understanding of column selectivity [7] and the availability of a free database for selecting orthogonal columns [5], column screening now makes more sense. A simple switching valve system can facilitate screening several columns in an unattended manner. Visual inspection or peak counting can facilitate choosing the most promising column for further method development.

13.4.3 HPLC vs UHPLC A thorough investigation of several variables can be time consuming. Consider first a conventional HPLC system (0.3 (as in the present example), gradient elution is preferred. Note that peaks 5 and 6 in Figure 15.7 overlap (Rs ¼ 0.1); further improvement of the separation is next explored by varying conditions to change selectivity. Among the various ways to alter selectivity noted previously, it is usually best to first determine the effects of a change in gradient time and temperature, as shown in Figure 15.8A–D. Although a change in gradient time, Figure 15.8B, does not improve the separation of peaks 5 and 6, a change in temperature does, Figure 15.8A and C. With the higher temperature and a change in gradient time, the separation in Figure 15.8D is seen to provide a baseline resolution of the sample (Rs ¼ 1.9). Because the last peak leaves the column at 12.2 min, the final %B for the separation can be reduced to 45% B; the gradient time must be reduced to maintain the same value of k* ¼ 15, or tG ¼ 12 min (Figure 15.8E). If this procedure does not result in an adequate separation, further changes in conditions that affect selectivity (α) can be explored or the column length can be increased (N). While such experiments can be carried out manually (by trial and error), it is usually more efficient to use computer simulation; see the further discussion of Ref. [3].

15.4 Problems associated with gradient elution Some commonly observed problems include: • • •

baselines column equilibration dwell volume changes

Baseline problems include drift and artifact peaks. With UV detection, the absorbance of the A and B solvents (water and organic) often differs, resulting in baselines that usually drift upward during the separation. Artifact peaks (those not associated with the sample) may also be observed in the chromatogram. It is recommended to begin each series of runs (for either routine operation or method development) with a blank gradient, that is, a gradient program run without injection of a sample. Any problems with drift or “false” peaks are then apparent. Although most data systems can correct for a moderate gradient drift, the drift can be corrected in other ways (e.g., the addition of a nonretained, UV-absorbing compound such as thiourea to the A solvent). Artifact peaks are often the result of solvent impurities. To ensure column equilibration, before injecting each sample, the initial mobile phase (e.g., 5% B in Figure 15.6) should be allowed to flow for enough time that the column becomes equilibrated and successive separations of the same sample are identical. Typically, 10 column volumes of the initial mobile phase are sufficient for equilibration (i.e.,  10 mL for a 100  4.6 mm column). Dwell volume, VD, refers to the volume of the gradient equipment between the point at which the A and B solvents are mixed and the column inlet. The gradient

379

9

5%–100% B in 10 min; 30°C; k* ≈ 5, Rs = 0.1

2

1

4

8

5 + 6

0

2

11

7

3

(A)

10

4

6

Time (min) 1

5%–100% B in 30 min; 30°C; k* ≈ 15, Rs = 0.1

0

2

2

4

6

5%–100% B in 10 min; 50°C; k* ≈ 5, Rs = 0.1

1

0

2

4

5%–45% B in 12 min; 50°C; k* ≈ 15, Rs = 1.9

6 Time (min)

8

3

5

6

10

11

12

9

8 4

11

10

5 -7

2

1

10

9

8

(D)

9

4

2

4

14

5 -7

Time (min)

3

12

8 4 3

5%–100% B in 30 min; 50°C; k* ≈ 15, Rs = 1.9

10

2

2

11

7

8 Time (min)

1

0

(C)

10

5 + 6

4

3

(B)

9

8

10

11

60% B 40%

7

20% 0%

0

(E)

2

4

6

8

10

12

Time (min)

FIGURE 15.8 Gradient separations of the “irregular” sample of Figure 15.6 as a function of gradient time and temperature (A–D). Conditions: 100  4.6 mm (3 μm) C18 column, 5–100% acetonitrilepH 2.6 phosphate buffer; 2.0 mL/min; gradient times and temperatures indicated in the figure; (E) shows the final separation of (D), after shortening the gradient.

References

at the column inlet then is delayed by this volume. Different gradient systems have different values of VD, so the chromatogram is delayed by the time equivalent to this volume (dwell time ¼ VD/F). For some samples and gradient conditions, this can result in a loss of sample resolution; a shift in retention also results when systems with different values of VD are used. Various means for dealing with this problem are discussed in Refs. [2,3].

References [1] Zhu PL, Snyder LR, Dolan JW, Djordjevic NM, Hill DW, Sander LC. Combined use of temperature and solvent strength in reversed-phase gradient elution. I. Predicting separation as a function of temperature and gradient conditions. J Chromatogr A 1996;756:21. [2] Snyder LR, Dolan JW. High-performance gradient elution. New York, NY: WileyInterscience; 2007. [3] Snyder LR, Kirkland JJ, Dolan JW. Introduction to modern liquid chromatography. 3rd ed. New York, NY: Wiley-Interscience; 2010 [chapter 9]. [4] Leroy F, Presle B, Verillon F, Verette E. Fast generic-gradient reversed-phase highperformance liquid chromatography using short narrow-bore columns packed with small nonporous silica particles for the analysis of combinatorial libraries. J Chromatogr Sci 2001;39:487. [5] Ayrton J, Dear GJ, Leavens WJ, Mallet DN, Plumb RS. Use of generic fast gradient liquid chromatography-tandem mass spectroscopy in quantitative bioanalysis. J Chromatogr B 1998;709:243. [6] Braumann T, Weber G, Grimme LH. Quantitative structure–activity relationships for herbicides: reversed-phase liquid chromatographic retention parameter, log kw, versus liquid–liquid partition coefficient as a model of the hydrophobicity of phenylureas, s-triazines, and phenoxycarbonic acid derivatives. J Chromatogr 1983;261:329.

381

CHAPTER

Fundamentals of enantioselective liquid chromatography

16

P. Peluso1 and Bezhan Chankvetadze2 1

Istituto di Chimica Biomolecolare ICB, CNR, Sede Secondaria di Sassari, Sassari, Italy, 2Institute of Physical and Analytical Chemistry, School of Exact and Natural Sciences, Tbilisi State University, Tbilisi, Georgia

16.1 Introduction Chirality is one of the fundamental properties of nature. In chemistry, we mostly deal with molecular chirality that has its roots in the revolutionary work of Pasteur in 1848 [1]. On the other hand, molecular chirality quite often transfers to macromolecular and supramolecular levels and leads to the specific effects which are matter of researches in various branches of chemistry. In addition, molecular chirality finds its reflection in pharmacy, biology, medicine, food, and environmental sciences. This chapter focuses mostly on molecular chirality since the enantiomers are nonsuperimposable molecules with the same chemical composition and constitution but different arrangement of atoms in space. Enantioselectivity mentioned in the title of this chapter is required for the separation of enantiomers and their analysis. Let us firstly discuss why do we need to separate the enantiomers from each other. As mentioned above, the enantiomers have the same chemical composition and constitution (set of chemical bonds between the atoms), and thus, they are very similar (actually identical) to each other in isotropic medium. However, the different spatial arrangement in space causes significant differences in their behavior in anisotropic (chiral) environment. Such environment is quite common in nature, and living bodies (among them the human body) belong to this category. Thus, for instance, two enantiomers of a chiral pharmaceutical may exhibit in our body different pharmacology, metabolism, pharmacokinetics, and toxicity. The same applies to chiral agrochemicals, food additives, and many other groups of chemical compounds. Based on their different properties in a chiral medium, the enantiomers are considered as two different chemical entities, and therefore, their separation and quantification are necessary. The next question is how to separate the enantiomers. The abovementioned similarity of enantiomers makes their separation challenging, and conceptually, this is possible only by using chiral auxiliary, medium, or interaction counterpart. The latter has to be present in enantiomerically pure or enriched form. The necessity Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00024-2 Copyright # 2023 Elsevier Inc. All rights reserved.

383

384

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

of having a chiral counterpart for recognizing and eventually separating the enantiomers from each other was recognized from very early days of development of chromatographic techniques. The chiral compound used (in enantiomerically pure or enriched form) to enantioselectively bind (and thus recognize) the enantiomers of a chiral compound to be separated is called chiral selector (CS). In initial studies, the CSs were used in a pure form in liquid chromatography (LC). Later, the CS got attached to the inert carrier, and this combined material is called chiral stationary phase (CSP) and can be used for the separation of enantiomers among others also in high-performance liquid chromatography (HPLC). The CSs used for LC separation of enantiomers until 1960s were basically the chiral compounds available in nature (Table 16.1) [2–7]. Although natural CSs exhibited useful enantiomer recognition ability, around the 1960s it became clear that for the preparation of highly efficient CSs either their synthesis based on a rational design or adequate derivatization of natural compounds was necessary. The major goal of the rational design should be obtaining a structure containing a set of interaction sites able to maximize the enantioselective intermolecular noncovalent interactions underlying selector-selectand complex formation (and if possible a suppression of nonenantioselective ones). In gas chromatography, the initial strategy was the synthesis of new CSs by mimicking peptides in order to promote hydrogen bonding (HB)-type noncovalent selector-selectand interactions, and this led to the success. In particular, in 1966 Gil-Av and co-workers reported the first baseline separation of enantiomers of amino acid derivatives in gas chromatography [8]. The initial approach in LC was derivatization, in particular acetylation, of chiral polymers (cellulose) available in nature [9]. Actually, in this specific study, the goal was not the development of a CS for general use, but the authors were looking for the material that would allow them to investigate some, at that time unusual, behavior of atropisomeric compounds. It has also to be mentioned that the CSs used until the late 1960s were at the same time also CSP since the packing material used did not contain any inert carrier. These materials were used in low-pressure column LC and not in HPLC mode. In over 50 years of intense development, many CSs were developed and quite many of them were commercialized. The effect of mobile phase and mobile phase additives on the separation of enantiomers has been studied in detail. Traditional LC Table 16.1 Chiral stationary phases (CSPs) used in early liquid chromatographic (LC) enantioseparations. Chiral selector/CSP

Year of publication

References

Lactose Lactose Cellulose paper Starch Cellulose

1939 1944 1951 1956 1961

[2] [3] [4] [5,6] [7]

16.2 Chiral selector

techniques were extended to supercritical fluid chromatography, nano-LC, and capillary electrochromatography. Chiral LC has been established not only for the analysis of enantiomers but also for their preparative- and product-scale separations. Impressive studies for better understanding of enantioselective recognition mechanisms have been performed and used to develop even more universal highperformance materials, huge separation factors, and separation of enantiomers within a few seconds, which have been reported. However, there are still further challenges in the field and the problems waiting for their solution. In this chapter, the basic developments and mechanisms of enantioseparations in HPLC are summarized. In particular, the major type of CSs and inert carriers used for the preparation of a CSP are overviewed. The effects of mobile phases and mobile phase additives on a separation process are highlighted. Currently available tools for the thermodynamic treatment of the separation process, their advantages, and bottlenecks are discussed, and kinetic approaches are also briefly overviewed. The final part of the chapter deals with the experimental and theoretical tools currently in use for better understanding of chiral recognition mechanisms at the molecular level.

16.2 Chiral selector As already mentioned above, CS is an essential component of an enantioseparation system. In theory, any chiral molecule that is able to interact with other molecules enantioselectively through noncovalent interactions can be used as a CS. However, there are many other requirements a CS has to meet in order to have any practical value, and these requirements become even more strict if a CS has to be commercialized as a part of CSP. CS has to be able to participate in multiple noncovalent enantioselective interactions and recognize the enantiomers of diverse structural classes of chiral analytes. In addition, a versatile CS has to function with various solvents, under multimodal elution conditions, have energetically rather homogenous (uniform) interaction sites with chiral analytes, and form selector-selectand complexes reversibly in a kinetically favorable way. Finally, CS has to be available in well-characterized pure form, not be exotic or very expensive, and the process of its production has to be reproducible. Currently available CSs can be classified as based on natural compounds (small molecules, macrocyclic oligomers, or polymers), biopolymers or entirely synthetic ones. Of CSs available in nature, only quinine (QN) and quinidine (QD) behave in the separation process as pseudoenantiomers to each other and, thus, make a designed adjustment of the enantiomer elution order (EEO) possible, while other macrocyclic (cyclodextrins, cyclofructans) or polymeric (amylose and cellulose) and biopolymeric (bovine serum albumin (BSA), α1-glycoprotein, ovomucoid) CSs are available only in one stereochemical configuration. This applies also to glycopeptidetype CSs as, for example, vancomycin, teicoplanin, and ristocetin. This is a certain disadvantage because the adjustment of the desired EEO with these CSs/CSPs is not always warranted.

385

386

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

The major LC CSs/CSPs developed over the last 50 years and commercialized are listed in Table 16.2 [10–41]. Some of them may be of a little use, or obsolete currently from the commercial point of view, but their development contributed significantly to advance the field of LC enantioseparations. The text below is built in a way to underline the rationale that led the scientists to propose these materials. As mentioned above, until the late 1960s, mostly natural CSs such as disaccharides and polysaccharides, or their partially modified forms (for instance, partially acetylated cellulose), were used as CS. The CS and CSP proposed by Davankov and Rogozin in the late 1960s to the early 1970s were also based on amino acids as naturally available compounds, but with significantly stronger elements introduced on the basis of a rational design [10–12]. There is a very interesting notice in this early publication stressing the advantage of chromatographic separation mechanism over other single-step separation processes. Specifically, the authors noticed “A large difference between the effective free energies of adsorption of D-proline and L-proline (ΔΔG0  400 cal/mole) indicates a high sensitivity of ligand-exchange processes with respect to the steric structures of the isomers to be separated. In a similar experiment, with a cation-exchange resin containing L-proline residues, the difference in the sorption energies of the α-methylbenzylamine enantiomers (according to the ion-exchange mechanism) was found to be ca. 5 cal/ Table 16.2 Major chiral selectors of commercially available CSPs and chiral columns for HPLC [10–41]. CS/CSP

Year

References

Ligand-exchange Crown ethers Brush-type/Pirkle-type Polymethacrylates Bovine serum albumin α1-acid glycoprotein Polysaccharide esters Polysaccharide phenylcarbamates with electrondonating or electron-withdrawing substituents Cyclodextrins Poly-[N-acryloyl amino acid esters], (Polyamides) Ovomucoid Polysaccharide phenylcarbamates containing both electron-donating and electron-withdrawing substituents Glycopeptides/macrocyclic antibiotics Cinchona alkaloids Polysaccharide derivatives “covalently” attached to silica Cyclofructans

1971 1975 1980 1980 1983 1983 1984 1984

[10–12] [13–15] [16] [17,18] [19,20] [21] [22–24] [25,26]

1984 1986 1987 1993

[27,28] [29] [30] [31,32]

1994 1985, 1996 1986, 1992, 1996 2009

[33,34] [35,36] [37–40] [41]

16.2 Chiral selector

mole” [11]. The term “effective free energies” deserves special attention in this quote since it means that the authors have recognized the advantage of chromatography as a cumulative multistep process over other batch-type processes already in 1972. Davankov has expressed this idea more clearly in his later publication [42]. This point deserves special attention and seems to be quite overlooked in most of the studies also today. Only in a few publications, attention is paid to this aspect of separation techniques [43,44]. Ligand-exchange chromatography is suitable for the separation of enantiomers of bidentate chiral compounds (mostly amino acids, diamines, amino alcohols, etc.). Later, the same mechanism was used for performing the very first separation of enantiomers in capillary electrophoresis (actually in capillary electrokinetic chromatography) [45]. Very recently, Davankov provided a very interesting view on his pioneering study [12]. Especially interesting seems the reason why a complexforming metal, in this particular case the copper cation, was introduced in the separation system in order to avoid excessive hydration of both CS and chiral analyte (amino acids). Thus, the significant difference between chiral ligand-exchange chromatography and most of the other chromatographic enantioseparation techniques is that in the former technique the interaction between the enantiomers and chiral selector is mediated by the metal ion. The metal component of the selector has the function to simultaneously coordinate both, on one side, a CS as one of the ligands and, on the other side, the enantiomers of a chiral analyte (Figure 16.1) [12]. Thus, the selector does not directly interact with the enantiomers to be separated. Rather, the enantiomers of the analyte have to enter a complex formation process with the metal cation that is mostly fixed on the surface through complexation with a chiral entity acting as a chiral selector. This complexation step is mostly slow and represents one of the bottlenecks of the chiral ligand-exchange chromatography causing its commonly low plate numbers. Later, Davankov has analyzed the peak broadening and other mechanistic aspects of chiral ligand-exchange chromatography, providing his view on how to overcome some of these problems [46].

FIGURE 16.1 First CSP developed for chiral ligand-exchange chromatography [12]. Redrawn with permission from Davankov [12].

387

388

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

Another innovation in the studies by Davankov and co-workers was that, for the first time, a CS was attached to the inert carrier (cross-linked styrene-divinylbenzene copolymer) in order to enhance separation efficiency [10–12,46]. The synthesis of the first series of cyclic polyethers having a cavity of molecular size and their ability to form complexes with metal cations were reported by Pedersen in 1967 [47]. The “ion-dipole interaction between the cation and the negatively charged oxygen atoms of the polyether ring” was considered to be a driving force of complex formation. This interaction led to the inclusion of cations in the cavity of these molecules. Although clearly shown, Pedersen did not emphasize complexforming ability of his new molecules with ammonium cations perhaps because, at that time, separation of metal cations was very hot topic. Actually, the ability of crown ether cavity to include ammonium cation warranted its application for complexing organic molecules containing primary amino group(s) and thus, to impressive application of this family of organic molecules in various branches of chemistry and, among them, also in separation science (mostly for separation of enantiomers). Initially, it was reported the stoichiometry of the complexes to be 1:1, independent of the size of the metal cation. However, later the author reported the complexes of other stoichiometry, too, depending on the size of a polyether cavity that was adjustable based on the number of oxygen atoms in the cycle. Pedersen’s initial cyclic polyethers were not chiral but, later, chiral crown ethers have been developed by incorporating appropriate chiral units as chiral barrier(s) into crown ethers [48–51]. Currently, many chiral crown ethers are known and, among them, those containing chiral binaphthyl or tartaric acid, as a distinctive unit, found special application for the separation of enantiomers of chiral compounds containing primary amino group, advancing to the commercialization stage. The first crown ether-based CSP reported by Sogah and Cram in 1975–1976 provided separation of enantiomers of chiral α-amino acids and α-amino acid ester salts [13,14]. However, the separation efficiency of the CSPs based on optically active bis-(1,10 -binaphthyl)-22-crown-6 compounds was not satisfactory for general use. In 1987, Shinbo and co-workers dynamically coated (3,30 -diphenyl-1,10 -binaphthyl)-20-crown-6 on octadecyl silica. The lipophilic interaction between the octadecyl groups of silica gel and the chiral crown ether, which contains a highly lipophilic 3,30 -diphenyl-1,10 -binaphthyl group, most likely warranted a dynamic attachment of a CS onto the hydrophobic surface of silica gel [15]. Actually, this rather simple technology became a basis for the first commercially available crown ether-based chiral column. A tentative mechanism for the resolution of chiral primary amino compounds on crown ether-based CSPs is the tripodal complexation of the protonated primary amino group (R-NH+3 ) inside the cavity of 18-crown-6 ring via three +NdHO hydrogen bonds. Considering the structural variability of chiral crown ethers used as CS currently, it is of course impossible to talk about one unique mechanism that operates in all cases. However, the abovementioned hypothesis of complexation mechanism through intermolecular HB sounds quite plausible. The synthesis of chiral crown ether incorporating tartaric acid unit, (+)-(18-crown-6)-2,3,11,12-tetracarboxylic acid was reported by Lehn and

16.2 Chiral selector

co-workers in 1980 [52]. This crown ether was first used as a CS in capillary electrophoresis in the early 1990s by Kuhn and co-workers [53]. Perhaps this inspired Japanese [54] and Korean [55] scientists to prepare CSPs by covalent immobilization of this crown ether onto the surface of microporous silica. This material became later commercially available. The concept of donor-acceptor-type CSPs became most advanced in the late 1970s, and the first example of such kind of CSP based on binaphthyl-2,20 -diyl hydrogen phosphate was described by Mikes and Boshart [56]. One year later, Hara and Dobashi published their first paper on donor-acceptor-type CSPs, emphasizing the role of HB for enantioselective recognition by the novel CS they proposed [57]. In the 1960s to the early 1970s, Pirkle and co-workers published a series of papers on the use of nuclear magnetic resonance (NMR) spectroscopy in chiral solvents for generating chemical shifts nonequivalence of diastereotopic signals and proposed the rules for determining absolute stereochemical configuration of chiral compounds [58]. Later, they used for the same purpose enantiomerically pure reagents, such as 2,2,2-trifluorophenylethanol, the compound that is also known as Pirkle’s alcohol [59]. These authors elegantly used their knowledge collected in NMR spectroscopy for developing a new CS for LC. On this basis, the first chiral column for HPLC separation of enantiomers was commercialized in 1980 [16]. After a few years of the introduction of the first multipurpose chiral column, Pirkle noticed: “Some years ago, we introduced the use of chiral solvating agents (CSA) for the NMR determination of enantiomeric purity and absolute configuration. Subsequent studies on the mode of operation of these CSAs led to the design of a number of chiral stationary phases (CSPs) on which one may effect the liquid chromatographic separation of many enantiomers” [60]. Pirkle-type CSs were designed with a purpose to target specific chiral interactions with the analyte. They consist of single strands of a chiral selector covalently bonded to the silica surface. For this reason, these CSPs are known also as “brush-type” CSPs. This type of CS commonly contains either π-donor or π-acceptor aromatic groups, as well as HB donors or acceptors, and dipole-stacking inducing functional groups (Figure 16.2A). Shortly before Pirkle’s publication in 1980, Hara and Dobashi published their article where they emphasized the importance of a pair of HBs between a chiral analyte and selector for HPLC separation of enantiomers [57]. Perhaps this was the reason why Pirkle in his studies highlighted again the three-point interaction principle introduced by Dalgliesh in 1952 [63]. In ref. [16], the following notice was made: “It should be generally appreciated that a minimum of three reference points are needed to distinguish the handedness of a chiral object. In molecular terms, this means that a CSP, in order to interact preferentially with one solute enantiomer, must undergo a minimum of three simultaneous interactions with that enantiomer. Since the ease of enantiomer separation stems from the magnitude of the stability difference of the diastereomeric solvates, the strength of the stereochemically dependent interaction should be appreciable”. In addition, the crucial role of HB, significance of other structural elements (size, charge distribution), and noncovalent interactions such as hydrophobic interactions and π-π donor-acceptor interactions were emphasized.

389

390

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

FIGURE 16.2 (A) Chiral recognition model based on multipoint interaction between a Pirkle-type CSP and N-(2-naphthyl)alanine undecenyl ester [61]; (B) illustration of the concept of reciprocity: a single enantiomer of a racemate which separates well on the CSP (left), and, when used to produce a second CSP (right), will usually afford separation of the enantiomers of analytes which are structurally similar to the CS of the first CSP [62]. Redrawn (A) and reproduced (B) with permission from Pirkle and Pochapsky [61] and Pirkle et al. [62].

Another interesting notice from the same early paper by Pirkle’s group is advocating of reciprocal chiral recognition principle (Figure 16.2B) [62], postulated earlier by Mikesˇ et al. [64] for the synthesis of useful CSs. The authors mentioned: “Chiral recognition being a reciprocal event, chiral stationary phases modeled after solutes resolvable upon the fluoroalcoholic columns successfully separate the enantiomers of a number of fluoroalcohols” [16]. In the early 1980s, Allenmark and co-authors reported a preparation of the first protein-based CSP suitable for HPLC separation of enantiomers [19,20]. The CSP was made by immobilization of the BSA on the surface of silica. In the authors’ opinion, hydrophobic interactions were the major contributor to enantioselective recognition, but they also noticed the role of electrostatic interactions [20]. HB between selector and selectand was not considered to be involved in enantioselective recognition, most likely because the aqueous mobile phases were used. Indeed, earlier perception, in our personal opinion not completely correct was that the HB cannot operate in aqueous medium. Almost in parallel with the Allenmark’s group, Hermansson reported a preparation of novel CSP for HPLC enantioseparations by immobilization of α1-acid glycoprotein onto microporous silica particles [21]. Hermansson shared the optimism regarding protein-based CSPs in the following way: “Such a chiral phase can be used for many different types of applications beside the analytical approach. It can be used to separate small amounts of isotope-labelled racemic drugs to be used as tracers in metabolic studies in vitro. Probably, it can also be used for the determination of binding constants of enantiomers or non-chiral

16.2 Chiral selector

drugs. Many racemic drugs can be screened, in short times, for stereoselective protein binding without the need for the pure enantiomers. Normally the protein-binding studies are performed using equilibrium dialysis. Such studies are time-consuming and often require labelled drugs” [21]. Discussing briefly the chiral recognition mechanism, the author mentioned that ionic and hydrophobic interactions as well as HB may be involved in the retention between analytes and the selector. This, in our opinion, correct mention of HB as a possible selector-selectand interaction seems interesting since the mobile phases used in this study were mostly aqueous. Cyclodextrins (CD) are the only CSs which are successfully used for separation of enantiomers in all separation techniques such as gas chromatography, HPLC, supercritical fluid chromatography, capillary electrophoresis, capillary electrochromatography, and also in the lab-on-a-chip separations. CDs offer several distinguished characteristics making them so universal as CS. First of all, as first noticed by Cramer in 1952, CDs can form complexes enantioselectively due to presence of chiral carbon atoms in their structure. He noticed: “Cyclodextrins distinguish not only molecules with different shape but optical antipodes too”. This statement was supported by the enantiomeric enrichment of mandelic, chlorophenylacetic, and bromophenylacetic acids [65]. Another important property of CDs, leading to their widespread application as CS, is their rather universal chiral recognition ability from the viewpoint of analyte coverage. The first application of β-CD as a CS in HPLC was reported in 1982 by Debowski et al. [66]. They used β-CD as a chiral additive to the mobile phase and observed partial separation of the enantiomers of mandelic acid and analogues. Perhaps these studies inspired Armstrong and co-workers who reported the first CD-based CSPs for HPLC in 1984 [27]. Major mechanism for enantioselective recognition with CD-based CSPs was considered to be an inclusion complex formation between a chiral analyte and CD-type CS. For this reason, the application of aqueous-organic mobile phases was recommended since water favors inclusion complex formation with CDs. In addition, it was assumed that the more nonpolar component of the mobile phase would occupy the hydrophobic cavity of CD and, thus, hinder inclusion complex formation. In the early 1990s, Armstrong and co-workers reported a synthesis of several CSPs based on CD derivatives [28]. The derivatization of CDs was performed in order to introduce into their structure new sites for π–π, dipole–dipole, or repulsive noncovalent interactions which were considered to be absent in the native CDs. These new CSPs also appeared useful for separation of enantiomers under so-called normal-phase conditions [28]. Currently, CD-based columns for HPLC separations of enantiomers are commercially available and used for separation of enantiomers of various classes of chiral analytes. Although CDs are quite universal CSs from the viewpoint of chiral analyte coverage, their enantiomer resolving capability expressed in terms of separation factor is moderate. For this reason, CD-based CSs are more successful in the separation techniques such as gas chromatography and, especially, capillary electrophoresis, where the modest chiral recognition power of the CS can be easily compensated by higher separation efficiency of the technique.

391

392

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

In 1984, along with CD-based CSs, polysaccharide esters [22–24] and phenyl carabamates [25,26] were also introduced as CS for HPLC separation of enantiomers. Among many successful groups of CSs which have been proposed in the last 50 years, polysaccharide phenylcarbamates and esters seem to be by far the most successful CSs for analytical and preparative-scale separation of enantiomers. Di- and polysaccharides, as well as cellulose acetyl esters [9,22] and starch (Table 16.1) [67,68], were used as CSs also before the 1980s. Of these studies, especially close to the new materials introduced in the 1980s was the one used in 1973 by Hesse and Hagel [22]. In early studies by Okamoto and co-workers, the effect of the type of polysaccharide and its linkage with pendant groups were studied. In the first study, polysaccharides such as cellulose, amylose, inulin, curdlan, chitosan, xylan, and dextran were converted into their phenylcarbamates and evaluated as CS for HPLC [25]. Among these polysaccharides, cellulose and amylose were found to be the most useful polysaccharides considering the chiral recognition ability of their derivatives, availability in pure form, and simplicity with regard to their derivatization. Later, it was found that chitin phenylcarbamate may also appear of some interest for certain groups of chiral analytes [69]. However, only cellulose and amylose derivatives were commercialized and established as the most widely used CSs for separation of enantiomers in HPLC, super/subcritical fluid chromatography, and capillary electrochromatography [70–72]. Of various derivatives of cellulose and amylose, basically arylesters and arylcarbamates have been studied intensively. Both of these kinds of derivatives, with some preference of arylcarbamates, appear to be suitable as CSs for liquid-phase enantioseparation techniques mentioned above. Already in early studies by Okamoto and his co-workers, attempts were made to find correlations between the spectral properties of polysaccharide derivatives and their enantiomer resolving ability. Intramolecular HB between adjacent carbamate moieties was observed in the phenylcarbamate derivatives of cellulose and amylose. In this perspective, the nature and position of substituents on the phenyl moiety, which were either electron-withdrawing or electron-donating, affect intra- and intermolecular HB capability of the carbamate moieties. This feature impacts structure and function of polysaccharide carbamate-based CSPs in terms of solubility in many organic solvents, higher ordered secondary structure, and intermolecular interaction with chiral analytes. For instance, cellulose tris(3,5-dichlorophenylcarbamate) (CDCPC) exhibited for structurally diverse 10 test racemic compounds, commonly used for evaluation of polysaccharide-based CSs, better enantiomer resolving ability compared to cellulose tris(3,5-dimethylphenylcarbamate) (CDMPC) [26]. However, CDCPC is soluble in n-hexane/propan-2-ol mixtures and, since the first generation of polysaccharide-based CSPs were prepared by physical coating of a CS on porous silica, the solubility of CDCPC in one of the most popular mobile phases did not allow its application as CS for normal-phase chromatography. At the same time, the CDMPC became one of the most successful CSs in HPLC. The interaction of chiral analytes with polymeric CSs, such as polysaccharide derivatives, may be a quite slow process compared to low-molecular-weight CSs. This may cause the significant band broadening in HPLC. Thus, for polysaccharide-based

16.2 Chiral selector

CSs the well-ordered secondary structure and the presence of uniform interaction sites with chiral analytes seem to be a very important characteristics responsible for peak efficiency in HPLC separations [31,32,71–74]. Thus, in polysaccharide phenylcarbamates the carbamate moiety has a dual function which contributes to enantioseparation: a) It is the most likely interaction site with chiral analytes as it was already noticed by Okamoto and co-authors in earlier publications, and b) due to the involvement in intramolecular HB, the same carbamate moieties significantly determine the solubility of polysaccharide derivatives in certain organic solvents, as well as their higher order structure (i.e., the uniformity of the adsorption sites) [31,32,71–74]. Since both of the abovementioned properties are desirable for CSs in HPLC, polysaccharide phenylcarbamates having a good balance between free carbamate moieties, available for interaction with chiral analytes, and carbamate moieties, involved in intramolecular HB, could be the most promising CSs [31,32,71–74]. The infrared spectroscopy (IR) represents the most convenient, simple, and fast tool for observing a formation of HB between adjacent carbamate moieties in polysaccharide phenylcarbamates. This is commonly done by the measurement of the IR spectrum in the NH region of the phenylcarbamates [26,31,32,73,74]. Such measurements indicate that electron-donating substituents on the phenyl ring promote the involvement of the carbamate moieties in intramolecular HB, while electronwithdrawing substituents on the phenyl ring diminish their involvement in intramolecular HB [26,31,32,73,74]. This is clearly seen from the NH region of IR spectra of CDCPC and CDMPC shown in Figure 16.3 [75]. One has also to consider the possibility of formation of HB between different strands of polysaccharide phenylcarbamates that may also affect the physical–chemical properties and enantiomer resolving ability of these materials.

FIGURE 16.3 NH bands in the FT-IR spectra of cellulose tris(3,5-dimethylphenylcarbamate) (CDMPC) (A) and cellulose tris(3,5-dichlorophenylcarbamate) (CDCPC) (B). Reproduced with permission from Chankvetadze et al. [75].

393

394

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

It was considered that a lower degree of involvement of carbamate moieties of CDCPC in intramolecular HB to be the reason why it was soluble in n-hexane/ propan-2-ol mixtures while CDMPC was not. At the same time, the availability of more free carbamate moieties for interactions with chiral analytes in CDCPC compared to CDMPC could be the explanation of more universal chiral recognition ability of the former compared to the latter CS [31,32,73,74]. Thus, based on all abovementioned considerations Chankvetadze in cooperation with Okamoto’s group proposed that the polysaccharide phenylcarbamate derivatives simultaneously possessing both, alkyl and halogen, substituents on the phenyl moiety could be useful CSs for liquid-phase separation of enantiomers. These new derivatives were synthesized in 1990s, and many of them became commercialized later by various companies and represent nowadays powerful materials and columns for separation of enantiomers on analytical and preparative scale by LC and supercritical fluid chromatography [71,72,76–79]. In addition, these CSs are also widely used in nano-LC and capillary electrochromatography [77,80,81]. As already mentioned above, the first generation of polysaccharide-based CSPs was prepared by physical adsorption of a CS onto the surface of wide-pore silica. This is rather easy, one-step, and fast process. However, the materials obtained by this way may suffer from instability problems and some organic solvents cannot be used as mobile phase components or modifiers, as well as for sample dissolution. In order to overcome these problems, quite many strategies were employed since 1987 for a covalent attachment of polysaccharide derivatives onto the surface of silica [37–40,82–85]. As the result of these efforts, highly efficient polysaccharidebased chiral columns are available today containing CS which are not soluble in organic solvents. We avoid here using a term “covalently immobilized CS” on purpose because, at least in some of these technologies, it is not clear if a CS is actually covalently linked to silica surface, and therefore cannot be stripped from it, or just due to a cross-linking CS becomes simply insoluble in organic solvents and, thus, remains coated on the surface of silica. Since 1971, Blaschke and co-workers systematically used chiral polyamide derivatives for preparative-scale separation of enantiomers of chiral drugs which were impossible, or very difficult, to be synthesized enantioselectively [86,87]. In 1986, they attached this type of chiral selector to mesoporous silica and used this CSP for separation of enantiomers of various kinds of chiral compounds [29]. This CSP and corresponding column were also commercialized and quite widely used in academia and industry for analytical and preparative-scale separation of enantiomers. Some pharmaceutical companies, such as Bayer AG in Germany, developed their own versions of polyamide-type CSs [88] for large-scale separations of enantiomers [89]. Another member of protein-type CS based on ovomucoid was proposed by Miwa and co-authors in 1987 [30,90]. The chiral columns based on this CS also got commercialized. Over last 3 decades, Haginaka published series of excellent research and review articles on chiral recognition mechanisms of protein- and glycoprotein-based CSs [91–93].

16.2 Chiral selector

Since 1994, Armstrong and co-workers successfully linked several macrocyclic antibiotics (in the first line, vancomycin, teicoplanin, and ristocetin) to the surface of silica and introduced these materials as multimodal CSPs for enantioseparations in HPLC [33,34]. Regarding the mechanism of chiral recognition and a potential of these new materials, the authors mentioned that: “The diversity of functionality of some of these chiral selectors is only approached by that of glycoproteins. Consequently, enantioseparation may be possible via several different mechanisms including π-π; complexation, hydrogen bonding, inclusion in a hydrophobic pocket, dipole stacking, steric interactions, or combinations thereof. While all other CSPs avail themselves of the same type of interactions, they are not all necessarily available in a single chiral selector and in relatively close proximity to one another. Macrocyclic antibiotics seem to have many of the useful enantioselectivity properties of proteins and other polymeric chiral selectors without their inherent problems of instability and low capacities” [33]. Several of macrocyclic antibiotic-based CSPs and chiral columns got commercialized and established over the last almost three decades as the second most popular materials after polysaccharide derivatives for HPLC separation of enantiomers. Another successful group of CSPs for HPLC separation of enantiomers, especially of ionic chiral analytes, was developed by L€ammerhofer and Lindner since 1996 [36]. Few attempts of using Cinchona alkaloid-based CSs for liquid chromatographic separation of enantiomers have been reported since 1954 [94–96]. In 1985, Rosini and co-authors reported the first application of a silica-supported Cinchona alkaloid-based material as CSP for HPLC [35]. In the years 1985–1996, CSPs based on Cinchona alkaloids were described by several groups. However, all these CSPs were characterized by low enantioselectivities and a limited coverage of chiral analytes. L€ammerhofer and Lindner achieved significant improvements of performance (selectivity and analyte coverage) of Cinchona alkaloid-based CSPs by carbamoylation of the secondary hydroxy group of the native alkaloids [36]. Later, in order to improve the analyte coverage of Cinchona alkaloid-based CSPs, chiral weak anion exchangers (WAX) (Figure 16.4A) were synthetically combined with aminosulfonic acid-based chiral strong cation exchangers (SCX), and thus, CSPs containing a single zwitterionic (ampholytic) CS were developed (Figure 16.4B) [98]. L€ammerhofer and Lindner have published several excellent reviews on chiral recognition mechanisms of Cinchona alkaloid-based selectors [99,100]. As a major part, a quinuclidine moiety, with a tertiary amino group which has a pKa about 9.8, is present in all these CSs and gets protonated under acidic conditions and functions as the positively charged site of the chiral WAX-type CS. This enables the CS to interact with negatively charged sites of chiral analytes through long-range electrostatic interactions (Figure 16.4A). Three of the four chiral centers of the quinuclidine moiety (1S, 3R, 4S) are stereochemically identical in QN and QD, while the configuration of chiral centers of the C8 and C9 atoms of the Cinchona backbone changes, respectively, from (S)-C8 and (R)-C9 for QN to (R)-C8 and (S)-C9 for QD. The absolute configuration of the chiral C8 and C9 atoms appeared to be of critical importance for the overall stereochemically driven molecular recognition, leading to the

395

396

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

FIGURE 16.4 Structures of Cinchona alkaloid-based selectors: (A) WAX; (B) ZWIX. Reproduced with permission from Peluso et al. [97].

formulation of the term “pseudo-enantiomeric” behavior. Thus, QN- and QD-based CSs also behave pseudo-enantiomerically to each other resulting in a reversed elution order of the resolved pair of enantiomers. This has a significant practical importance for desired adjustment of EEO in HPLC. Together with the abovementioned long-range electrostatic interactions, which are important to drive approaching chiral analyte to a CS, HB and/or a π-π-type stacking are also involved in selector-selectand interactions in the case of WAX-type

16.2 Chiral selector

CSs. These two latter noncovalent interactions, characterized by directionality, are of crucial importance for enantioselective recognition. Cinchona-based WAX-type CSPs are popular for separation of enantiomers of charged chiral analytes, especially of free or derivatized amino acids, amines, and carboxylic acids [36,98–101]. Cyclofructan-based CSs were introduced by Armstrong and co-workers in 2009 [41]. Native cyclofructans, as macrocyclic molecules, cannot form inclusion-type complexes and, thus, behave differently compared to CDs. This seems to be caused by intramolecular HB hindering the analytes to enter the cavity. This limits the capability of native cyclofructans to be used as CS in separation science. However, aliphatic- and aromatic-functionalized cyclofructans separate a very broad range of racemic compounds and act as complementary CSs. In particular, aliphatic-derivatized cyclofructans, specifically CF6s, with a low substitution degree, baseline, separate the enantiomers of chiral primary amines. As the authors noticed, partial derivatization on the CF6 molecule disrupts the intramolecular HB, thereby making the core of the molecule more accessible. In contrast, highly aromatic functionalized CF6 stationary phases lose most of the enantioselective capabilities toward primary amines; however, they gain broad selectivity for most other types of analytes. Later, cyclofructan-based CSPs and columns got commercialized, which are used predominantly for separation of enantiomers of chiral analytes containing primary amino group [102]. Since the number of commercially available CSP and chiral columns changes dynamically, we cannot warrant that there is no other CS-based material commercially available currently. Below, few (groups) among hundreds of CS and CSPs described in the literature over last 40 years but not (yet) commercialized due to various reasons are shortly overviewed. The selection of this material is mostly based on the personal vision of the authors and should no way indicate to any kind of priority of these materials. Due to certain limits of this book chapter, it is simply impossible to cover the materials evaluated as CS in LC over the last 5 decades more or less completely. Tartaramide-based CSs initially proposed by Hara and Dobashi [103], and further developed by Allenmark and co-workers, are definitely of certain interest [104]. The interesting concept for synthesis of entirely synthetic CS proposed in 1990s by Still and co-authors also deserves a great attention [105–108]. The inspiration for these studies was a mimicking of biological receptors. According to these authors, the ability of biological receptors, such as antibodies, to bind their ligands with exceptional selectivity is related to the conformational rigidity, large size (ability to include the ligand as much as possible), and multiplicity of well-defined interaction sites. Actually, this group demonstrated the validity of their approach not only through enantioselectivity exhibited by their newly developed selectors but also through their functional group selectivity and biooligomer residue selectivity [106]. Still and co-workers also proposed a synthetic receptor binding elucidation strategy based on an encoded combinatorial library [107]. This solid-phase methodology enabled to screen fast over 50,000 ligands on their affinity to synthetic receptors, and to distinguish binding energies as small as 1 kcal/mol [107,108].

397

398

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

Gasparrini and co-workers in cooperation with Still’s group prepared a CSP by immobilization of a synthetic C3-symmetric receptor onto silica. The CS used in these studies was characterized by a concave, rigid binding site and an array of alternating HB donors and acceptors located around the macrocycle periphery [109,110]. A separation factor (α) of 43 obtained for the enantiomers of threonine derivative, corresponding to a difference in binding affinities of enantiomers ΔΔG ¼ 2.2 kcal/mol, was exceptional in 1990s. However, today significantly higher separation factors are reported in few separations of enantiomers with both, synthetic and natural compound-based CSs [111–113]. Molecularly imprinted polymers have attracted the attention as useful CSPs since the 1970s [114]. Since the 1980s, these materials found application in HPLC for separation of enantiomers [115]. Since 1990s, these materials including their mechanisms of imprinting, as well as enantioselective recognition, were intensely studied by Haginaka’s research group [116–120]. No CSP or chiral column based on molecularly imprinted polymers got commercialized, perhaps mostly due to their too high specificity to a target molecule, rather low efficiency and some issues related to a residual target in the CSP and its potential “bleeding”. Of the newest generation of CS and CSPs described in the literature, materials based on chiral cages [121], covalent organic frameworks [122,123], and highly ordered (chiral) mesoporous silica materials [124] may become of certain interest in the future after some refinement.

16.3 Inert carrier As already mentioned above, until the 1970s CSs and CSPs used in LC were actually the same, while the CS was commonly used in planar chromatography [4,63], or packed in chromatographic column with the mobile phase propulsed through the packing bed either by gravitation or applying a low pressure [2,3,5–7]. In these early studies, chiral LC was used mostly for preparative purposes. Since the 1970s, the analytical potential of HPLC was recognized and hence, high efficiency of the columns became a prerequisite. Initially, Davankov used cross-linked styrenedivinylbenzene copolymer as an inert carrier in order to enhance the separation efficiency in LC [10–12]. Later, similar organic copolymers were used also for covalent attachment of crown-ether-based CS by Cram and co-workers [13]. However, in the 1970s the high-purity silica with relatively narrow particle-size distribution and wide range of pore sizes also became available. Soon, silica became mostly used material as inert carrier for CSs [14]. Silica-based CSPs commonly provide higher plate numbers compared to organic polymer-based CSPs. The particulate fully porous silica materials used as carriers for CS, together with availability in different particle size and pore diameter, may also have various surface chemistry, as well as offer high mechanical and pressure stability. In addition, monolithic silica, as well as particulate but superficially porous silica materials, can be also used. Among other inert carriers, those based on zirconia [125,126] were also evaluated but did not

16.3 Inert carrier

established, not offering any significant advantages over silica. This subsection shortly summarizes mostly silica-based and few organic [127] inert carriers used for preparation of CSPs. As mentioned above, silica particles have number of advantages to be used as inert carriers for preparation of CSPs. Since the samples to be chromatographically separated become increasingly multicomponent and complex (especially for proteomic and metabolomic studies, food and environmental analysis with complex matrices), the higher peak efficiency is in a hot demand in chromatography. Higher peak efficiency can be generated with smaller particles. Indeed, based on the wellknown fundamental Eq. (16.1), it is obvious that the column efficiency increases with decreasing particle diameter (dp). H¼

1 Ce dp



1 +

Dm Cm d 2p u

+

Csm d 2p u Cd Dm B 1 + Cu + ¼ Au3 + u u Dm

(16.1)

This is related to a smaller A-term (while the voids between particles are reducing with decreasing particle size, dp) and a smaller C-term due the shorter diffusion path length inside the particle. In the case of particle-packed column, the plate height H is roughly proportional to dp, whereas the backpressure, ΔP, is inversely proportional to d2p [128]. Thus, a linear increase in column efficiency due to the use of smaller particles is accompanied by a quadratic increase in column backpressure. This limits the application of small particles in HPLC to 1.0–2.0 μm at present. Monoliths offer following advantages from this point of view. The flow through monolithic channels is laminar. Thus, monoliths have nearly no void volume and do not develop eddies. The transport of an analyte through the monolithic bed is basically perfusive and, in the monolithic media with a low portion of mesopores, the analyte diffusion into and out of the pores does not significantly contribute to the band broadening. Thus, monolithic materials enable to achieve higher peak efficiencies by alternative mechanisms (different flow path) compared to the approach used in particle-packed columns. Due to reducing the role of diffusion (which is slow) in solute transfer through the monolithic bed, the peak efficiency does not decrease as sharply as in particle-packed columns when increasing linear flow rate of the mobile phase. This allows the application of higher flow rates and thus shortens the analysis time without a significant sacrifice of plate numbers. Thus, monolithic materials offer advantages such as low backpressure and a flatter dependence of the plate numbers on the linear flow rate of the mobile phase [129,130]. This combination enables fast separations even with conventional HPLC systems [131]. Monoliths based on organic polymers are typically prepared from acrylamide, methacrylate, or styrene monomers or by ring-opening polymerization. Organic monoliths for the separation of enantiomers have been applied mostly in the capillary format and basically for capillary electrochromatographic enantioseparations [132,133]. Only very few articles deal with capillary- or nano-LC [134]. Therefore, CSPs based on organic monoliths are not discussed in detail in this chapter.

399

400

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

The monolithic silica-based chiral columns were described in literature since 2003 [131,135–138]. Chankvetadze et al. reported a column prepared by in situ coating of a commercially available achiral monolithic column with cellulose tris(3,5-dimethylphenylcarbamate) enabled a baseline separation of the enantiomers of 2,2,2-trifluoro-1-(9-anthryl)ethanol within 30 s [131]. Coated-type polysaccharide-based monolithic columns have the disadvantage of relatively low stability and not all kinds of mobile phases are compatible with such a column. Therefore, a technology was developed for in situ covalent immobilization of cellulose (3,5-dimethylphenylcarbamate) onto monolithic silica [135]. Monolithic silica columns containing β-cyclodextrin [136], tert-butylcarbamoylquinine chiral anion exchanger [137], and penicillin G acylase [138] as CSs were reported at around the same time. In this chapter, CSPs prepared by modification of silica monoliths confined in fused-silica capillaries are not discussed in details and the interested reader can find these materials overviewed in references 129 and 130. Monolithic support definitely offers certain theoretical advantages over particulate silica for preparation of CSPs. However, a CS has to be attached to monolithic silica in situ by either coating or covalent immobilization of CS. This is a complicated process and cannot be easily performed in a reproducible way. In addition, characterization of monolithic CSP that is cladded within the column is not easy without damaging a column. Thus, the quite limited advantages monolithic CSPs offer over particulate CSPs, and the difficulties associated with the column preparation and characterization, did not allow this concept to see a commercialization for chiral separations. Another type of silica-based materials attracting special attention for preparation of CSPs is superficially porous silica. This kind of silica particles contain nonporous core covered with porous shell. Based on the morphology of these particles, they can offer kinetic advantage over totally porous silica particles mostly by affecting A-term and C-term in van Deemter equation (16.2): H¼A+

B + Cu u

(16.2)

where H is a plate height, u is a linear (interstitial) flow rate of the mobile phase, and A, B, and C coefficients describe chromatographic band broadening due to eddy diffusion, longitudinal diffusion, and mass transfer between mobile and stationary phases, respectively. As shown in Figure 16.5, the decrease of A coefficient in the case of superficially porous particles is mostly caused by more uniformity of the size (diameter) of superficially porous particles compared to fully porous particles related to their production technology [139]. The decrease in A coefficient is caused by more uniform path of the analyte through the packed bed while the decrease in C coefficient is caused by shortening the diffusion path through the porous structure of silica particles (since the core is nonporous). The negative effect of superficially porous structure on silica particles is reduction of a specific surface area that may become critical for some analytical but mostly for preparative application of these materials (causing lower loadability with the sample to be separated and, thus, lower productivity).

16.4 Mobile phase

C-term advantage of superficially porous silica

A-term advantage of superficially porous silica

FIGURE 16.5 Comparison of analyte paths in the case of superficially porous and fully porous particles. Redrawn with permission from Gumustas et al. [139].

For the first time, CSP prepared based on superficially porous silica was mentioned in 2011 [140], and the first on-purpose study illustrating the advantages of these materials for preparation of CSPs was published in 2012 by Chankvetadze and co-workers [141]. Later, almost all kinds of more or less popular CSs were attached to superficially porous silica and the advantages of these CSPs especially for obtaining very fast separation of enantiomers has been demonstrated [142– 166]. Some of these materials are commercialized and available from leading manufacturers of chromatographic columns. Currently available superficially porous silica materials need further optimization to be ideally suited for attachment of high-molecular-weight CSs such as polysaccharide derivatives [146].

16.4 Mobile phase The mobile phase in HPLC is the medium where selector-selectand interaction and, thus, an enantioselective recognition take place. On this basis, its nature is as important for the final result as structure and chemistry of a CS and analyte are. The CSs/ CSPs overviewed in the Section 16.2 were initially proposed for one or other kind of mobile phase but almost for all of them the portfolio of applicable mobile phases was significantly expanded over the last 40 years. Before we discuss the effect of a mobile phase on HPLC separation of enantiomers, let us make following remark: the terminologies such as normal-phase (NP) and reversed-phase (RP) are commonly used in enantioselective HPLC in analogy with achiral separations. Thus, if one uses less polar mobile phase such as hydrocarbon containing alcohol, one calls NP separation, and when the polar mobile phase, such as aqueous-organic mixtures, are used, the

401

402

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

separation is called RP. Based on our experience over last 2 decades, as well as based on publications by several other groups, it is obvious that due to multiple mechanisms involved in selector-selectand interactions in case of enantioselective chromatography using nonpolar/less polar mobile phase does not a priori mean that the separation follows the NP separation mechanism. Even more frequently, when aqueous-organic mobile phase is used, the separation does not always follow RP separation mechanism. Therefore, in enantioselective chromatography, using the terminology just indicating the mobile phase or its type/composition (for instance, hydrocarbon-alcohol or aqueous-organic) seems preferable rather than something indicating to a separation mechanism (NP or RP). When discussing a mobile phase, one has to consider its major component, a mobile phase modifier and a mobile phase additive. Each of these components is involved in selector-selectand interaction, can selectively enhance or suppress one or another kind of noncovalent interactions, and thus affect retention and separation of enantiomers on a given CSP. Below, just typical effects that a mobile phase may have on enantioseparations are mentioned with the references for further reading. Thus, considering the abovementioned remark regarding NP and RP modes, one can consider following four basic separation modes: less polar mobile phase, aqueous-organic polar mobile phase, polar organic (PO), and hydrophylic interaction chromatography (HILIC). Within each mode, further submodes are also mentioned in few studies based on the additives to the mobile phase. There are some theoretical and practical reasons for using one or another mobile phase for HPLC separation of enantiomers. For instance, alkane-alcohol mobile phases were used if one believed that HB was the most important noncovalent interaction contributing to enantioselective recognition. In early days of enantioselective LC, it was believed that intermolecular HB is hindered in aqueous medium (mobile phase). However, later studies confirmed that HB-based recognition is also possible in aqueous solvents [44,71,72]. Another argument for using the alkane-alcohol mobile phase can be a better solubility (or hydrolytic stability) of analyte in organic solvents rather than in aqueous solvents. This type of mobile phase is also better suitable for preparative-scale separations, since removing organic solvents from the purified fractions appears to be less energy-demanding compared to aqueous solutions. The aqueous-organic phase also offers certain advantages such as an injection of biological samples on column after very simplified or even without sample preparation, a better imitation of stereoselective drug-receptor noncovalent interactions. In addition, watercontaining mobile phases are better compatible with MS detectors and safer in this application (HPLC-MS coupling). Polar organic mobile phases commonly offer rapid analysis, sharp chromatographic peaks, and many chiral analytes are better soluble in organic solvents such as methanol (MeOH) or acetonitrile (ACN) than in alkane-alcohol mixtures [44,71,167]. The major criteria for selection of the mobile phase composition are following: (1) The reversible adsorption of the analyte from the mobile to a stationary phase has to be thermodynamically favored; (2) Chiral selector has to be able to recognize chiral analyte enantioselectively in the given mobile phase; (3) Analysis time and peak

16.4 Mobile phase

shape have to be acceptable; (4) The hazardous effects (toxicity, explosivity, inflammability) of the mobile phase on the personal and environment have to be minimized and green chemistry approach considered; (5) Purity, availability, and price of the mobile phase components are also important. Variation of a mobile phase composition can lead to: (a) Appearance of enantioseparation; (b) Disappearance of enantioseparation; (c) Improvement or worsening of enantioseparation; (d) Change of peak dispersion and symmetry; and (e) Change of EEO. All of these changes are obviously important for separation of enantiomers. For instance, EEO is critical as from practical, as well as from theoretical points of view. It is well proved that eluting the minor enantiomeric impurity in front of major enantiomer in their mixture is favorable for obtaining better analytical characteristics (lower limit of detection and limit of quantification) for the enantioselective method [168–170]. The EEO can be easily adjusted by alternative use of two CSs with opposite stereochemical configurations. The problem is that some CSs such as, for instance, polysaccharide-, cyclodextrin-, protein-, and glycopeptide-based CSs are not available in both stereochemical configurations, and the EEO cannot be a priori adjusted on the columns with using CSs of opposite stereochemical configuration (that is available with synthetic CSs). Therefore, adjustment of the EEO by using mobile phase becomes very important [44,71,72]. At molecular level, a reversal of EEO, observed in HPLC as the analytical conditions change, means that the “enantioselective” sum of the interactions, which was in favor of one enantiomer in the first separation environment, changes in the new separation system, where a new noncovalent interaction system favors the opposite enantiomer. Thus, understanding the molecular origins of these phenomena may provide useful information for understanding of chiral recognition mechanism at molecular level. The very first example of EEO reversal based on the mobile phase composition was reported by Pirkle and co-workers in 1984, on brush-type CSPs, when the alkane-/alcohol-type mobile phase was replaced with an aqueous alcohol-based mobile phase [171]. Many examples on all major kinds of CSPs and by various modification of a mobile phase are described in the literature over the last almost 4 decades [44]. Just one example is discussed below in order to illustrate the effect of the mobile phase composition on separation of enantiomers in HPLC [172]. In this particular example, the effect of the water content in ACN on separation of enantiomers of chiral arylpropionic acid-derivative ketoprofen was studied on amylose tris (5-chloro-2-methyl-phenylcarbamate)-based chiral column Lux Amylose-2 (Figure 16.6) [172]. With water content up to around 20% in ACN, the chiral column exhibits rather “HILIC-like” than RP behavior. HILIC-like behavior is associated with a decrease in analyte retention with increasing content of water in the mobile phase made primarily of ACN. At the water contents around 10% (v/v), this trend reverts and the retention of analytes increases with increasing content of water in the mobile phase. It is worth mentioning that such behavior is commonly not observed in mobile phases made of MeOH, and in this case with increasing content of water in the mobile phase, only increase in retention is observed. At the molecular level, in both mobile phase

403

404

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

FIGURE 16.6 (A) Dependence of retention of ketoprofen enantiomers on the content of water in ACN on Lux Amylose-2 column; (B) Separation of enantiomers of ketoprofen on Lux Amylose-2 with ACN, ACN-water (60–40, v/v), and ACN-water (20–80, v/v) as MPs, (T ¼ 5°C). Adapted with permission from Matarashvili et al. [172].

systems, ACN/water and MeOH/water, analyte retention is dependent on the balance between hydrophilic (HBs) and hydrophobic noncovalent interactions operating between selector and selectand. Hydrophobic interactions play a primary role in retention as far as the nature and composition of the mobile phase facilitate such interactions, as well as HBs get suppressed. Since MeOH, as a protic solvent, forms HBs with CS, as well as eventually with analyte, it does not allow significant HBtype interactions between a CS and analyte. Therefore, the addition of water can only reduce the strong eluting strength of MeOH (a common mobile phase modifier in RP), thus causing an increase in analyte retention. Contrary to this, ACN does not strongly interfere with CS-analyte HB-type interactions. The addition of water to ACN will first significantly reduce HB-induced retention, which is favored in pure ACN, and at higher water contents reduce the eluting strength of ACN (as in RP separations). Thus, HBs are prevalent at low water content in ACN (typically below 10–20%), while hydrophobic interactions take over commonly above 10–20% water content. Switching of separation modes between HILIC and RP on the same chiral column just depending on the mobile phase composition is an elegant way to illustrate that a selectand-selector-mobile phase combination represents one integral and dynamic recognition system. It depends actually on our ability to properly understand and tune it. What we discuss in this section relates to such increase of a water content that it gradually becomes a major component of a mobile phase. As a result of such transition from ACN to aqueous ACN, a U-shaped dependence of analyte retention on the water content in the mobile phase is commonly observed (Figure 16.6A) [167] that indicates a significant redistribution of noncovalent interactions in a separation

16.5 Thermodynamics of a separation process

system. In quite many cases, a reversal of EEO was also observed as a result of this HILIC to RP transition (Figure 16.6B) [172], indicating a drastic change in the system in terms of interactions which are responsible for enantioselective recognition. The major question to be answered here is if such a drastic change in the recognition pattern is caused due to a structural/conformational change in a selector or selectand, in both of them, or just by the redistribution of noncovalent selector-selectand interactions due to change of mobile phase. For further effects of the mobile phase composition on separation of enantiomers, the reader is referred to a recent article [44].

16.5 Thermodynamics of a separation process From the resolution (Rs) Eq. (16.3) in chromatography, it is obvious that this major characteristic of a quality of separation depends on separation selectivity (α), separation efficiency/plate numbers (N), and analyte retention (k). R¼

pffiffiffiffi   N k α1 4 k+1 α

(16.3)

Of these characteristics, α and k are thermodynamic characteristics while theoretical plate numbers (N) is a kinetic characteristic. As shown in Figure 16.7 [173], of these parameters the influence of retention (k) on peak resolution is minor (especially if k > 2), while that of the plate numbers (N) is intermediate and that of separation 3.0

Resolution (R)

2.5

Efficiency Retention

a

Selectivity

2.0 1.5

N

1.0

k 0.5 0 1.0

1.05

1.10

1.15

1.20

1.25

a

0

5.000

10.000

15.000

20.000

25.000

N

0

5

10

15

20

25

k

FIGURE 16.7 Impact of selectivity (α), efficiency (N), and retention (k) on peak resolution (R). Adapted with permission from Ravisankar et al. [173].

405

406

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

selectivity (α) is major. Thus, for optimization of chromatographic separation it is favorable to optimize a separation selectivity (α) rather than efficiency (N). At the same time, it is important to note that optimizing selectivity may appear not only more favorable but also more challenging. Optimization of selectivity, in an ideal case, may require understanding of selector-selectand interactions on a molecular level. Without this knowledge, some empirical rules can be used to adjust separation selectivity based on the variation of the CSP, mobile phase composition, and separation temperature. In opposite to this, a dependence of plate numbers on separation parameters (such as flow rate and viscosity of a mobile phase, temperature) and characteristics of CSP (its particle size, morphology, etc.) is better known. Thus, the thermodynamics and kinetics of a separation process both are important for its understanding and optimization. In this section, some thermodynamic aspects of enantioseparations are discussed, while in the next one the kinetic aspects are shortly overviewed. One of the major terms of thermodynamics, the standard Gibbs free energy (ΔG0), characterizes the equilibrium of the process. On the other hand, ΔG0 is directly related to the noncovalent interactions involved in selector-selectand contact. The experimental tools enabling direct measurement of thermodynamic quantities, especially the standard molar enthalpy (ΔH0) of the process, are thermal analysis methods such as differential scanning calorimetry (DSC) and isothermal titration calorimetry (ITC). However, these methods are not easily applicable to the binary (two-phase) heterogeneous systems commonly used in chromatographic separations. Therefore, for calculation of thermodynamic quantities related to a chromatographic separation process, the indirect method mostly based on the van’t Hoff’s equation adapted to chromatographic separations is used. Although widely applied, following limitations of this approach have to be mentioned: (1) van’t Hoff’s approach was developed for a true equilibrium and in chromatographic process we do not have a true but just a pseudo-equilibrium; (2) ΔG0 is applicable to isobaric system at a constant temperature. Its analogue for the isochoric system at a constant temperature is the standard Helmholtz free energy F. However, a chromatographic column represents an open system with pressure gradient. Therefore, it is neither isobaric nor isochoric [44,174]; (3) On CSP there are at least two kinds of adsorption sites (nonenantioselective and enantioselective) [44,174,175]. Therefore, thermodynamic characteristics determined from van’t Hoff’s approach provide some overall information and not the true thermodynamic quantities for each kind of adsorption site separately. The van’t Hoff equation can be used for describing the temperature dependence of the retention factor k: lnk ¼ 

ΔH0 1 ΔS0  + + lnϕ R T R

(16.4)

16.5 Thermodynamics of a separation process

ΔH0 and the standard molar entropy (ΔS0) of the analyte transfer process from the mobile phase to the stationary phase (analyte adsorption) can be calculated by using the plot of natural logarithm of k versus the reciprocal of the absolute  0temperature.  ΔH0 of solute transfer relates to the slope of the line, while the sum ΔSR + lnϕ can be calculated from its intercept. Assuming the phase ratio ϕ to be known, the standard molar entropy can be calculated. The difference between the free energies of transfer of two enantiomers of a given chiral analyte from the mobile phase to the stationary phase (ΔS,RΔG0) can be calculated by Eq. (16.5): ΔS,R ΔG0 ¼ ΔS,R ΔH0  TΔS,R ΔS0 ¼ RTln

KS KR

(16.5)

where ΔS,RΔH0 is the difference between molar enthalpies, and ΔS,RΔS0 is the difference between the molar entropies of the phase transfer of two enantiomers from the mobile phase to the stationary phase, KS and KR are distribution coefficients of two enantiomers between the mobile and stationary phases. The subscript S arbitrary corresponds to the most retained enantiomer, while the subscript R to the less retained enantiomer. At a certain temperature, the so-called isoenantioselective temperature (Tiso), the enthalpic and entropic terms in Eq. (16.5) compensate each other, ΔS,RΔG0 equals zero, and the enantiomers are not separated. T iso ¼

ΔS,R ΔH 0 ΔS,R ΔS0

(16.6)

The EEO is opposite above and below Tiso. Commonly, the enantioseparations below Tiso are enthalpy-driven, while above Tiso they are entropy-driven. This is defined based on the relative contribution of the ΔS,RΔH0 or the ΔS,RΔS0 terms, respectively, to the free energy of the process, according to Eq. (16.5). The differences between the molar enthalpy of phase transfer of two enantiomers of a given chiral compound ΔS,RΔH0, as well as between the molar entropy of phase transfer of the same enantiomers ΔS,RΔS0, can be also calculated by plotting the natural logarithm of the separation factor α, versus the reciprocal of the absolute temperature T [44,175]. lnα ¼ 

ΔS,R ΔH 0 1 ΔS,R ΔS0  + T R R

(16.7)

As it is evident from Eq. (16.5), the adsorption of analyte enantiomers from the mobile to the stationary phase can be favored either by enthalpy or by entropy, as well as in rare cases by both of them. Based on Eq. (16.7), the same applies to the separation (i.e., selective adsorption) of enantiomers. Thus, in theory multiple scenarios are possible. For nonlinear van’t Hoff behavior, dependence of ln K and 1/T can be estimated by a quadratic form (Eq. 16.8) [176]: lnK ¼ A +

B C + T T

(16.8)

407

408

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

where the coefficients A, B, and C can be estimated by nonlinear regression methods. ΔH° and ΔS° can then be obtained by Eqs. (16.9), (16.10), which show that ΔH° and ΔS° are temperature dependent with nonlinear van’t Hoff behavior [176].   2C ΔH° ¼ R B + T

(16.9)

  C ΔS° ¼ R A  2 T

(16.10)

In theory, any reversible process which alters the enthalpy or entropy of phase transfer can result in nonlinear van’t Hoff plots. These processes may affect the analyte, the stationary phase, or the mobile phase, or more than one of them. The processes such as ionization, changes in conformation, or changes in the degree to which the mobile phase interacts with either the analyte or stationary phase can be among such reversible processes. Once again, we would like to stress that the thermodynamic data obtained based on van’t Hoff’s approach have to be used with a great care. On the other hand, thus obtained thermodynamic quantities driving the analyte transfer from the mobile phase to the stationary phase provide reasonable ideas regarding the chiral recognition mechanism, help explain some uncommon experimental observations, and have a certain practical value, too. Thus, for instance, the opposite EEO of ketoprofen on the polysaccharide-based CSPs containing alternatively, coated or covalently immobilized CSs, (Figure 16.8A), can be well explained based on the value of Tiso derived from the corresponding van’t Hoff plots (Figure 16.8B and C) [71,177]. For the column with the coated CS, Tiso was in the range of 55°C, while for the column with immobilized CS it was between 2 and 34°C. This indicates that the EEO of ketoprofen on the column with covalently immobilized CS reverts due to enthalpy-entropy compensation, that is actually an extrathermodynamic relationship of thermodynamic quantities, observed already at subzero temperatures. On the column with coated CS, the same enthalpy-entropy compensation takes place at around 55°C. Thus, the EEO of ketoprofen at 25°C is opposite on the two columns mentioned above. The values of Tiso also explain why on the column with coated CS temperature-dependent reversal of EEO was observed in the studied temperature range, while this reversal was not there in the case of the column with covalently immobilized CS in the same temperature range [177]. Using basic thermodynamic approach for explaining some unusual observations in enantioselective chromatography seems impressive and particularly useful. However, some open questions still remain. For instance, why is Tiso so different for CSPs with such rather small structural difference, as it is in the columns with, chemically, the same chiral selector but in one case just coated, and in other case covalently attached onto the silica? Determining adsorption isotherms and deconvoluting the overall enthalpy and entropy terms into their contributions may provide answers to some still unanswered questions.

16.5 Thermodynamics of a separation process

FIGURE 16.8 Separation of enantiomers of ketoprofen on the columns with coated and immobilized CS under NP elution conditions (A), and plots of natural logarithm of k versus the reciprocal of absolute temperature for separation of enantiomers of ketoprofen on the column with coated (B) and immobilized (C) CS. Adapted with permission from Maisuradze et al. [177].

409

410

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

16.6 Kinetics of a separation process As already mentioned above, kinetic performance of the entire separation process is also important for optimization of separation. The major quantitative characteristic of the kinetic part of separation is a plate numbers (N) that is directly related to the theoretical plate height (H). Both of these parameters (they are of course dependent) can be experimentally measured from the chromatogram and describe sample zone (peak) broadening during the separation due to various contributions. The wellknown van Deemter Eq. (16.2) can be rewritten in more detailed form as follows (Eq. 16.11) [178]: h ¼ aðvÞ +

b + CS v + Cads v + hheat v

(16.11)

where h is the reduced plate height and is defined as H/dp, where H is the height equivalent to a theoretical plate and dp is the particle diameter. In the Eq. (16.11), v is the reduced interstitial velocity, expressed as: v¼

Fv d p πr 2 εe Dm

(16.12)

where Fv is the flow rate, r the column radius, εe the external porosity, and Dm the bulk molecular diffusion coefficient. The four terms in Eq. (16.11) account for all the contributions to band broadening. a(v) is the eddy dispersion, b the longitudinal diffusion, cs the solid–liquid mass transfer resistance across the stationary phase, and cads accounts for a slow adsorption–desorption kinetics. This last term is usually negligible except for chiral separations (especially for the second eluted enantiomer) and separation of large biomolecules. Finally, hheat is an additional source of band broadening due to the friction between the eluent at high flow rates and the packed bed especially when it is made of very small particles (sub-2 μm), otherwise it can be neglected. As mentioned above, the major issue in chiral chromatography is that the adsorption–desorption kinetics cannot be neglected and it is also quite difficult to estimate [178]. We want to pay readers’ attention to the fact that the van Deemter and related equations adequately describe the intracolumn band-broadening (peak dispersion) and do not address the extra column contribution to this phenomenon. Therefore, the first issue the operator has to deal in order to optimize a separation process is to minimize extra column contribution (void volumes in a separation system starting from sample injection to its detection) to the peak dispersion and afterwards further improve peak performance taking care of intracolumn contributions. In this book, there is Chapter 8 on the ultra-fast and ultra-high-performance liquid chromatography addressing these issues in more details. Therefore, the discussion there will be focused on more general issues and on rather detailed discussion of one single example [178]. As mentioned above, reducing a void volume in the separation system is necessary for improving peak performance. The dead volume has not only extracolumn

16.6 Kinetics of a separation process

but also an intracolumn component related to the empty space between the (mostly) spherical particles of the packing material. Reducing this space is possible by reducing particle size, as well as by reducing inhomogeneity of particle size and shape, and these are common approaches in chromatography for increasing the plate numbers. Reducing the particle size of packing material reduces not only the voids in the column but also the intraparticle diffusion path for the analyte molecules and thus also a C coefficient van Deemter equation (16.2) (more specifically, cs in Eq. 16.11). The limiting effect of particle size reduction, together with the technical issues of adequately packing the columns with submicron size particles, is also fundamental issue of increasing backpressure in the system as discussed in Section 16.3. As the solution for the increasing backpressure, the use of monolithic packing materials was viewed in 1990s but did not work as successfully as expected for chiral separations [129,130]. Alternative solution for achieving shorter intraparticle diffusion path without enormous increase of backpressure in the system was introduction of superficially porous particles. This concept was more successful compared to monoliths for chiral separations [140–166] but still has some limitations for specific type of CSs [145], as well as eventually for preparative-scale applications due to reduced specific surface area of packing material. Beyond these rather general approaches, a fine tuning of a separation process sometimes may require more deep understanding of peak dispersion phenomena as discussed in the example below [178]: as shown in Figure 16.9, the sulfoxide-2 in difference to sulfoxide-1 has a methyl substituent on the amide function and a halogen substituent on the benzyl group. Despite these apparently minor changes, the chromatographic behaviors of secondly eluted enantiomers of these molecules are dramatically different, as it can be seen in Figure 16.9 [178]. It is evident that the more retained enantiomer of sulfoxide-1 has a high affinity for the CSP while the two firstly eluted enantiomers of both sulfoxides retain on this chiral column quite similarly. Accordingly, the enantioselectivity is more than 10 times larger in the case of sulfoxide-1 (α ¼ 30.0) compared to sulfoxide-2 (α ¼ 2.5), as reported in Table 16.3 [178].

FIGURE 16.9 Chromatographic traces of the enantioseparation of sulfoxides 1 and 2. Adapted with permission from Felletti et al. [178].

411

412

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

Table 16.3 Enantioselectivity values (α) and Bi-Langmuir isotherm parameters calculated through inverse method for sulfoxide-1 and sulfoxide-2 [178]. Selective site

Nonselective site

Compound

α

qssel(g/L)

b1, sel(L/g)

b2, sel(L/g)

qsnsel(g/L)

bnsel(L/g)

Sulfoxide-1 Sulfoxide-2

30.0 2.5

1.6 1.7

0.4 0.3

33.7 2.1

0.4 2.0

1.7 0.5

From a molecular viewpoint, the different affinity of the secondly eluted enantiomers can be explained in light of the abovementioned structural differences between the sulfoxides [143,144,179]. Figure 16.10 reveals that the van Deemter curve for sulfoxide-1 exhibits an unusual convex upward shape, while the van Deemter curves for the other enantiomers present the regular trend (convex downward). Convex upward van Deemter plots were also reported by Armstrong and coworkers [148,149] and Asnin and co-workers [180]. Armstrong and co-workers explained this observation as the result of frictional heating due to the percolation of fluid through finely packed beds at (relatively) high flow rates [148], while Asnin’s group with a combination of large retention and imperfect packing (hence, eddy dispersion). In addition, Asnin’s group pointed out the possible influence of slow adsorption–desorption kinetics.

FIGURE 16.10 Reduced van Deemter plots of sulfoxide-1 (blue) and sulfoxide-2 (red) enantiomers. Void circles refer to first enantiomers, while full circles refer to second ones. Reproduced with permission from Felletti et al. [178].

16.6 Kinetics of a separation process

In order to investigate the origin of this phenomenon for chiral sulfoxides mentioned above, the different contributions to mass transfer for these species have been investigated. Firstly, longitudinal diffusion (b-term of van Deemter curve) was estimated. Calculated b-terms are reported in Table 16.4 [178]. The second eluted enantiomer of sulfoxide-1 is characterized by a roughly 50% smaller b-term compared to the first one. This is apparently counter-intuitive since the zone retention factor k1 of the second eluted enantiomer of sulfoxide-1 (see Table 16.4) is approximately seven times higher. Actually, the larger the zone retention factor, the larger the b-term is expected. However, this finding can be explained by considering that the effective diffusion coefficient of the second eluted enantiomer of sulfoxide-1 is one order of magnitude smaller than that of the first one (Table 16.4) [178]. As it can be pointed out from this table, Dp for the second eluted sulfoxide-1 enantiomer is one order of magnitude smaller compared to the others. As a reflection of this, cs of the second enantiomer of sulfoxide-1 is one order of magnitude higher than those of the others, which are very close to each other (Table 16.4) [178]. It has to be mentioned that in achiral chromatography the eddy diffusion term a(v) in van Deemter equation can be determined by the subtraction of b/v and cv terms from accurately measured h values. However, this approach cannot be applied for chiral chromatography because the contribution of adsorption–desorption kinetics to the overall mass transfer term cannot be neglected. In order to estimate the individual contributions coming from a(v) and cadsv in ref. [178], an attempt was made to differentiate between a(v) and cadsv by studying the adsorption isotherms of both racemates on the CSP. Adsorption isotherms provide the information about the energetics of the process and thus indirectly also about its kinetics (commonly, the stronger the binding, the slower the adsorption–desorption process). The importance of measuring adsorption isotherm is that it enables to differentiate between chiral and achiral contributions to retention. The experimental peaks could be satisfactorily Table 16.4 Effective diffusion coefficients (Deff), molecular diffusion coefficients (Dm), particle diffusivity (Dp), retention factor (k), zone retention factors (k1), longitudinal diffusion (b), and solid–liquid mass transfer resistance term (cs) for the enantiomers of the two sulfoxides [178]. Sulfoxide-1

Sulfoxide-1

Parameter

1st

2nd

1st

2nd

Deff (cm2/s) Dm (cm2/s) Dp (cm2/s) k k1 b cs

1.95  105 3.39  105 1.91  105 0.2 1.6 3.0 0.015

1.66  105 3.39  105 2.84  105 6.0 14.8 1.6 0.233

1.43  105 1.48  105 1.40  105 0.2 1.6 3.0 0.015

1.25  105 2.48  105 1.70  105 0.5 2.3 3.3 0.016

413

414

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

predicted by using a competitive Bi-Langmuir isotherm as the adsorption model. The calculated adsorption isotherm parameters are listed in Table 16.3 [178]. Contrary to what usually happens in chiral chromatography, saturation capacity of selective sites is higher than that of nonselective ones for sulfoxide-1. In the context of this study, the most striking result was a binding constant of the second eluted sulfoxide-1 enantiomer. Remarkably, its binding constant on selective sites (33.7 L/g) was found to be more than 15 times larger than that of more retained sulfoxide-2 enantiomer (2.1 L/g) and even more than 80 times larger than that of the less retained sulfoxide-1 enantiomer (0.4 L/g). These findings support the previous statement regarding the importance of the adsorption–desorption kinetics to explain the shape of the curve a(v) + cadsv for the more retained sulfoxide-1 enantiomer. On the other hand, they validate the hypothesis of localized enantioselective (mostly one enantiomer) adsorption of this molecule. The very large binding constant is indeed consistent with a scenario of perfect matching between the chiral molecule and a portion of the CSP, thanks to specific HB and other noncovalent interactions. This arrangement, which can be exclusively achieved by the second eluted sulfoxide-1 enantiomer, prevents the diffusion of the molecule on the surface. It is not excluded also that when polymeric CSs are coated on silica, they could arrange as a complex structure where chiral cavities are present at a molecular level [181]. In this context, it is likely that only molecules with specific steric arrangement could fit them. This would explain also the very large enantioselectivity of the CSP toward sulfoxide-1. Independent molecular modeling study perfectly supports this observation [179].

16.7 Enantioselective recognition mechanisms 16.7.1 Experimental techniques Understanding enantioselective recognition mechanisms is of critical importance for optimization of the enantioseparation process, and even more important for development of novel, powerful CSs. Understanding of chiral recognition mechanisms means identifying those noncovalent interactions between a chiral analyte (selectand) and a CS which are responsible for selector-selectand transient complex formation and enantioselective recognition in these interactions. It has to be mentioned that forces responsible for the selector-selectand interactions are not a priori the same which are responsible for enantioselective recognition or in other words, strong (tight) selector-selectand interaction does not a priori mean better enantioselective recognition. More or less complete overviewing of the experimental techniques applied for understanding of enantioselective recognition and separation mechanisms over last four decades is impossible in a chapter like this. Therefore, the strategy here is to mention the major techniques (Table 16.5) and provide references for further reading on this topic. Thus, the techniques such as infrared (IR) spectroscopy, nuclear magnetic resonance (NMR) spectroscopy, and electronic circular dichroism (ECD) spectroscopy

16.7 Enantioselective recognition mechanisms

Table 16.5 Major instrumental techniques used for mechanistic studies in enantioselective liquid chromatography. Experimental technique 1

Infrared Spectroscopy

2

Nuclear magnetic resonance spectroscopy

3

Electronic circular dichroism

4

Solid-state nuclear magnetic resonance spectroscopy Vibrational circular dichroism

5

6

Adsorption studies

7 8

Mass spectrometry Thermal analysis

9

X-ray diffraction

Application (Information provided)

References

Following the synthesis of chiral selectors, presence of HB donors and acceptors in CS, intra- and intermolecular HB in CS, selectand and between selector and selectand Following the synthesis of chiral selectors, selector-selectand interactions, strength of HB sites, tentative structure of selectorselectand complexes (based on nuclear Overhauser effect) Secondary structure and regularity of CS, information about uniformity of adsorption sites, certain information regarding stereochemical configuration of CS and selectands Structure of CS, selector-selectand interactions

[26,32,73– 76,182–185]

Selector-selectand interactions, absolute stereochemical configuration of selectands and fine nuances of structure of CS Energetics and adsorption capacities (surface density of adsorption sites) of CSPs Selector-selectand interactions Thermodynamic quantities, phase transitions Crystallinity of CS and CSP, effect of the analyte on a CS

[191–193]

[26,32,60,73– 76,186,187]

[26,32,73–76]

[182,188– 190]

[194–207]

[208,209] [189,210,211] [182,212]

have been used already in the early days of development of CS/CSPs for following the synthesis of CS, their structural characterization, understanding chiral recognition mechanisms, and for optimizing the performance of CSs [44,71,72]. For instance, in early 1990s a combination of IR, NMR, and ECD spectroscopy was used for better understanding the properties of polysaccharide-based CSs and in developing new generation of this CSs containing both, electron-donating and electronwithdrawing substituents [31,32,44,71–74]. Later, other experimental techniques such as vibrational circular dichroism (VCD) [191,192], solid-state NMR spectroscopy [188–190], as well as for specific cases adsorption [194–207], mass spectrometry [208,209], thermal analysis [189,210,211,213], X-ray diffraction [182,212], and

415

416

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

several other techniques were used. Among these techniques, the standard techniques used for characterizing packing materials in chromatography such as porosimetry, surface area determination, particle-size analyzers, electron microscopy are not discussed specifically. IR spectroscopy has been used in order to get information about the extent of intramolecular HB especially in polysaccharide-based CSs since early 1980s. This fast, easy, and widely available technique provides information regarding the advancement of carbamoylation or esterification reactions of polysaccharides, as well as regarding the distribution of carbamate moieties in the final product(s) between free (able to interact with chiral analytes) and involved in intramolecular HBs (responsible for higher order structure of polysaccharide derivative) [26,32,73–76,182–185]. In the Section 16.2, we reported a quote from one of the pioneers of enantioselective liquid chromatography made in 1985 [60] that perfectly describes the symbiotic interplay between NMR spectroscopy and chiral HPLC as two powerful techniques for studying enantioselective intermolecular recognition. In the 1970s–1980s, Pirkle and co-workers used NMR spectroscopy to get idea about chiral recognition in selector-selectand pair [60]. In particular, in 1986, for the first time, Pirkle’s group used the nuclear Overhauser effect (NOE) to provide direct support for a chromatographically derived chiral recognition model [214], paving the way to the utilization of NMR spectroscopy for structural characterization of CSs as well as classifying interaction sites with chiral analytes based on their nature and energetics [26,32,73,74,76]. In late 1990s and early 2000s, Okamoto and co-workers reported in several papers on the use of NMR spectroscopy for confirmation of the chiral recognition by polysaccharide derivatives observed in solution. These studies were focused at disclosing possible proximities between various structural fragments of CSs and chiral analytes [186,187]. NMR spectroscopy is one of the most elegant tools for the study of the enantioselective recognition mechanism in solution at a molecular level. Despite this, the correlations between the enantiomer resolving ability of CS attached to the solid surface and the enantioselective recognition in solution must be drawn with some caution. The fact is that the most of polysaccharide phenylcarbamate derivatives possessing high enantioselective recognition ability are only soluble in organic solvents, such as acetone, terahydrofuran, pyridine, and DMSO, which strongly interact with the polar carbamate groups of the phenylcarbamate derivatives of polysaccharides. In these solvents, the polysaccharide phenylcarbamate derivatives are not able to sufficiently interact with enantiomers for efficient enantioselective recognition. Therefore, clarifying the enantioselective recognition mechanism of polysaccharide phenylcarbamate derivatives by NMR spectroscopy experiments performed in a solution is difficult. Solid-state NMR spectroscopy was intensively used by Wang and co-workers for elucidating the structural changes in amylose tris(3,5-dimethylphenylcarbamate, ADMPC) caused by the influence of solvent and temperature. The aim of these studies was to explain the unusual behavior of this material, as a CS, caused by the change of mobile phase modifier and separation temperature [188–190]. Solid-state

16.7 Enantioselective recognition mechanisms

cross-polarization magic angle spinning (CP/MAS) NMR experiment (1H/13C CP/ MAS) was employed to reveal structural differences in an ADMPC-based CSP (Chiralpak AD) as a function of the mobile phase composition [190]. Chiralpak AD CSP in dry state showed an amorphous CP/MAS NMR spectrum. However, after flushed Chiralpak AD adsorbent with organic mobile phases, CP/MAS spectra displayed evidence of solvent-CSP complexes. Chiralpak AD CSP flushed with nonpolar n-hexane exhibited solvent-CSP association with minimal structural perturbation of CS. However, for Chiralpak AD CSP in contact with n-hexane/alcohol mixtures, solvent incorporation into CS caused marked alteration of conformation distribution. This was evidenced by the increased resolution of 13C peaks in the CP/MAS spectrum of the CSP. Propan-2-ol exhibited more efficient displacement ability of incorporated n-hexane, while forming relatively more ordered solvent complexes with Chiralpak AD CS, in comparison with ethanol as a modifier. Reversed EEO and unusual retention behavior as a function of mobile phase modifier, observed earlier on Chiralpak AD, were considered to be caused by different alterations of the steric environment of the chiral cavities in the CS by the different mobile phase modifiers. In addition, based on the chemical shift of the C-1 carbon on the amylose backbone, the authors arrived to the conclusion that the structure of Chiralpak AD is helical with a number of folds less than six [188]. Vibrational circular dichroism (VCD) was established in the studies related to enantioseparations mostly by Grinberg and co-authors. In one of their studies, they applied VCD to evaluate the effect of polar organic solvents on the conformation of CS [192]. The VCD spectra were taken of the intact ADMPC film in the solid state, as well as in the contact with n-hexane containing various amounts of alcohols, such as ethanol, n-propanol, and propan-2-ol. The significant alteration of ADMPC conformation was observed due to the effect of solvents studied. Other applications of VCD are reported in the references [191–193]. Applications of X-ray diffraction (XRD) in studies related to noncovalent selector-selectand interactions were reported by Kasat et al. [182,212]. In the frame of enantioselective HPLC, this technique was used by noticing that the polymer crystallinity increases significantly upon absorption of enantiomers [182]. X-ray crystallography can be used for deriving solid-state structure of selectands, sometimes for selectors, and selector-selectand co-crystals, if these latter can be obtained. Thermal analysis methods, such as DSC and ITC, can provide valuable information regarding the energetics of noncovalent selector-selectand interactions. These techniques may enable observing of possible phase transitions and calculation of thermodynamic quantities. The major advantage of ITC in studies of noncovalent selector-selectand interactions is that this technique provides the possibility of a direct measurement of the thermodynamic quantity such as enthalpy and entropy and free energy of interaction can be calculated, while when using other techniques (among them also separation techniques), these quantities are derived indirectly from the dependences of related parameters, among them also retention and selectivity of separation, on temperature. The problem is that highly precise ITC instruments operate ideally for the homogeneous single-phase systems, while in chromatography one

417

418

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

deals with biphasic heterogeneous systems. Few studies on the application of ITC for highly dispersed heterogeneous systems are already published but these are not yet routinely established. Adsorption studies provide valuable information regarding the surface density and energetic profile of selector-selectand noncovalent interaction sites and enable the determination of the thermodynamic quantities of this process. Since the adsorption of a selectand on a selector is a necessary precondition for enantioselective recognition and enantiomer separation in chromatographic techniques, collecting the information about the fine mechanisms of adsorption process can help significantly in better understanding the entire recognition/separation process. The major advantage of the adsorption studies is that they enable a clear differentiation between nonenantioselective and enantioselective adsorption sites that is not easily available by other approaches [174,175,194–197]. The major contribution to the adsorption studies related to separation of enantiomers in liquid chromatography was made by Guiochon, Felinger, Cavazzini, Fornstedt, Kaczmarski, Tsui, and few other groups. Only one earlier example is discussed in more details here, while other publications illustrating the power of this fundamental approach for better understanding of enantioselective separation mechanisms in liquid chromatography are listed in Table 16.5. The thermodynamics of adsorption of (R)- and (S)-propranolol from the acetic acid buffer (pH ¼ 4.7 and 5.5) on the immobilized cellobiohydrolase I was studied by Guiochon and co-workers between 5°C and 45°C [195]. The equilibrium data were best described by a bi-Langmuir adsorption isotherm. One of the two Langmuir contributions described the nonspecific interactions between propranolol enantiomers and most sites on the surfaces (type-I, nonenantioselective sites), with a large saturation capacity. The second contribution accounts for the enantioselective interactions (type-II sites). The type-II sites have a lower monolayer capacity than the first. The adsorption enthalpy and entropy on type-I sites were 1.1 kcal/mol and +0.1 cal/mol1 K1, respectively. For type-II sites, they were 1.9 kcal/mol and 2.6 cal/mol1 K1, respectively, for (R)-propranolol and +1.6 kcal/mol and +11.6 cal/mol1 K1, respectively, for (S)-propranolol, at pH ¼ 5.5. The opposite algebraic sign of the thermodynamic quantities explains why at this pH the retention time of the less-retained R-enantiomer decreased with increasing temperature, while the retention time of the S-enantiomer increased. This, of course, led to unusual increase of the separation factor when the temperature was raised from 5°C to 45°C [195]. Although van’t Hoff’s approach is easy and provides explanation for some unusual observations in enantioselective liquid chromatography [177,210], it is obvious that, for obtaining more reliable information on enantioselective and nonenantioselective sites involved in chiral chromatography, alternative adsorption techniques and approaches [194–207] have to be used.

16.7.2 Computation/molecular modeling The main objective of modeling chromatographic enantioseparation is obtaining information on noncovalent forces responsible for selector-selectand interaction and for enantioselective recognition in this interaction. Intermolecular noncovalent

16.7 Enantioselective recognition mechanisms

interactions may occur between selector and selectand, solvent and selector, solvent and selectand, and solvent and selector-selectand complex. The ensemble of noncovalent interactions may act cooperatively stabilizing the diastereoisomeric complex which releases the second eluting enantiomer. Otherwise, noncovalent interactions may act in terms of negative cooperation, destabilizing the other diastereoisomeric complex which releases the first eluting enantiomer. In the meantime, intramolecular interactions within both selector [44,72,75] and selectand [179,215–217] may stabilize conformationally the structures, impact their molecular shape, and determine the enantioseparation result. In addition, boundary conditions play a key role in the enantioseparations; thus, the real experimental system to be modeled may be rather complex. In the last few years, the interest toward the application of computational tools for exploring enantioselective recognition mechanism in chromatography at molecular level by integrating experimental and theoretical data has been growing more and more. Five types of computational studies can be identified by examining the literature published in this field [44]: (a) calculation of molecular properties of optimized geometries of the interacting partners, and correlation of these calculated properties with experimental parameters; (b) calculation of theoretical spectra and their comparison with experimental spectra to confirm structures, and to determine absolute configurations of chiral compounds and conformational chirality of complex structures; (c) determination of binding and interaction energy between selector and selectand by using molecular mechanics (MM) and quantum mechanical (QM) methods; (d) molecular dynamics (MD) simulations that shows how analyte molecules move around and/or along the selector, and interact with selector surface over time; (e) molecular docking allows for predicting both energy and geometry of selector-selectand binding. As confirmation of the interest in this field, recently several reviews were published on this topic [97,218–220], and other papers dealing with chiral recognition in separation science also examined recent applications of computational methods in enantioseparation science [44,221,222]. For this reason, in this section only the main concepts underlying the development of this field are summarized. In addition, selected applications of molecular modeling in chromatography are also summarized in Table 16.6 [179,186,187,211,212,215,216,223– 239] and, in few cases, briefly commented below. In general, the choice of a specific strategy depends on the scope of the study, computational means at disposal, structural and size features of the interacting partners, and boundary conditions. In this regard, mobile phase plays a key role in liquidphase enantioseparation. Each solvent has distinctive steric and electronic features and occupies a space in the neighborhood of selector and analyte surfaces; therefore, mobile phase may influence adsorption and enantiodifferentiation processes through solvation effects on both analyte and selector, and may stabilize or destabilize noncovalent interactions. Thus, omitting the proper solvation model in the virtual system may be questionable, given that analyte molecules compete with solvent molecules to be adsorbed onto the selector surface. It is worth mentioning that the importance of introducing solvent parameterization correctly had been stressed in some earlier studies on enantioselection modeling performed by Grinberg and co-authors [211].

419

Table 16.6 Representative modeling approaches used in enantioselective chromatography. Separation system

Polysaccharidebased CSs

Entry

Technique

Chiral analyte

Chiral selector

Solvation model

Year

References

1 2 3 4 5 6

MM, MM, MM MM, MM, MM

trans-Stilbene oxide Benoxaprofen 1,10 -Bi-2-naphthol Diol, Hydroxy ester trans-Stilbene oxide 1-(9-Anthryl)2,2,2-trifluoroethanol

CTPC ADMPC CTPC CMB CDMPC ADMPC

vacuum vacuum vacuum vacuum vacuum vacuum

1995 1996 1996 1997 1999 2002

[223] [224] [186] [211] [225] [187]

7

DFT

vacuum

2007

[212]

8

MD

CDMPC ADMPC ASMBC ADMPC

vacuum

2007

[226]

9

Docking, MD MM, MD

ADMPC

vacuum

2010

[227]

vacuum

2012

[228]

Pyrazole derivatives

cellulose and amylose carbamate-based CSs CMB

implicit solvent implicit solvent implicit solvent explicit solvent explicit solvent explicit solvent vacuum, SM8

2012

[229]

2014

[230]

2016

[231]

2017

[232]

2018

[233]

2020

[234]

2021

[179]

10 11

MD MD MD MD

p-O-tert-butyl tyrosine allyl ester Metalaxyl, benalaxyl α-Amino acid derivatives

Pyrazole derivatives

CMB

13

Docking, MD Docking, MD MD

Pyrazole derivatives

CMB

14

MD

Flavanone

ADMPC

15

MD

16

MD

CDMPC ADMPC ADMPC

17

MM, V (DFT)

Halogenated 4,40 bipyridines Benzoin, Flavanone, Thalidomide, Valsartan BSBA and derivatives

12

polysaccharide-based CSs

Macrocyclic glycopeptides

Brush-type CSs

18

Docking

Xanthonic derivatives

19

MD

Carnosine

Ristocetin A Teicoplanin Teicoplanin aglycone Vancomycin Teicoplanin A2-2

20

MD

Alcohols

Proline-based CSs

21

24

MD

Tetrahydroindazole derivatives Nevirapine, Oxcarbazepine Aminophenylalanine derivative cyclopropyl dafachronic acid derivatives

Whelk-O1

23

TD-DFT, MD DFT, Docking MD

22 Cinchona alkaloid CSs

Whelk-O1 Zwitterionic Quinine CS Quinine-based CS

vacuum

2018

[235]

explicit solvent explicit solvent explicit solvent vacuum

2020

[215]

2015

[236]

2020

[216]

2021

[237]

2015

[238]

2018

[239]

explicit solvent explicit solvent

Acronyms: ADMPC, amylose tris(3,5-dimethylphenylcarbamate); ASMBC, amylose tris[(S)-methylbenzylcarbamate]; CDMPC, cellulose tris(3,5-dimethylphenylcarbamate); CMB, cellulose tris(4-methylbenzoate); CTPC, cellulose triphenylcarbamate; DFT, density functional theory; MD, molecular dynamics; MM, molecular mechanics; V, electrostatic potential.

422

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

On the other hand, a virtual model in the vacuum may show acceptable predictive ability for a population of selector/selectand pairs large enough to be statistically significant, and compared under invariant solvation conditions, assuming that the impact of (achiral) solvent is equal for all selector/selectand pairs. The selector differentiates the two enantiomers through various set of noncovalent interactions. While a single noncovalent interaction shows rather small power in stabilizing the selector-enantiomer complex, the ensemble of different interactions may cooperatively act to promote the enantiodifferentiation. Moreover, interactions with the achiral support may also affect adsorption of the analyte on the CSP. In this regard, it is worth noting that in modeling enantioselection processes on CSP, three approaches are in general used in published literature in terms of modeling of the “inert support”: (a) chromatographic CS is modeled as a free molecule (without the inorganic support) with no limit on flexibility; (b) chromatographic CS is modeled as a fixed molecule by freezing its conformational flexibility, again without the support; (c) chromatographic CS is modeled by linking selector units to the modeled inorganic surface. This approach is usually reported for brush-type CSP [216,236], whereas few studies including support modeling are reported for polymeric selectors [234,240]. In this regard, we have to say that the real advantages of also modeling the inert support in the virtual chromatographic system involving polysaccharide-based CSs, compared to modeling of the selector exclusively, have been not actually demonstrated so far. Rather, in the modeling of polysaccharide-based CSs, although modeling small frameworks of these CSs may provide some pieces of information [241–244], these approaches neglect or underestimate the influence of backbone and conformational chirality on enantiodifferentiation, and consequently, their use may limit the reliability of the model to describe the real system. A special comment deserves the application of molecular docking in the vacuum to model enantioseparations by using oligomers as the virtual model of polysaccharide-based CSs. This has become a common and very popular practice [235,245,246]. However, the main limitation of these studies is that the solvent is not considered in the virtual model, and thus, the model results are representative of the binding affinity of selector and selectand in the vacuum exclusively. In several cases, the value of adding molecular docking calculations to the experimental study is very limited, not going beyond a simple and approximative visualization of the selector-selectand complex. It is worth mentioning that along with atomistic model, empirical fitting procedures are also reported as mathematics-based approaches integrating experimental data. In this type of approaches, mathematical models for key separation parameters are fitted to experimental observations. The topic was recently reviewed by Mangelings and co-authors for polysaccharide-based and macrocyclic antibiotic chiral selectors [220]; thus, just few lines are dedicated to this topic herein. It is worth mentioning that in 2011, the West’s group reported, for the first time, the use of linear solvation energy relationships (LSER) to predict the degree of separation between two enantiomers in the case of polysaccharide-based CSPs [247]. This LSER model comprising five Abraham descriptors (E, S, A, B, V) was augmented with two

References

additional descriptors, flexibility (F), and globularity (G), these descriptors being considered important for chiral recognition. Along with polysaccharide-based CSPs [247–250], the augmented solvation parameter model was also used to characterize macrocyclic glycopeptide [251], Cinchona-based brush-type CSPs [252], and the (R,R)-Whelk-O1 selector [253].

16.8 Conclusions and future trends After almost 50 years of intense development, enantioselective chromatography is quite matured field of research today. Special progress has been made from the practical point of view. There are multiple class of CSs, CSPs, and chiral columns available, and various elution mode(s) can be applied. Enantioseparations can be performed on various scale, starting from nano-liquid chromatography and labon-a-chip platforms to preparative- and even product-scale separations. From the theoretical point of view, there is still a major advancement to be achieved in order to properly understand the mechanisms underlying enantioselective recognition on the molecular level and enabling first of all explaining observed enantioseparation outcomes, and then designing desired enantioselectivities and performances.

References [1] Pasteur L. Recherches Sur les Proprietes Specifiques des Deux Acides Qui Composent l’Acide Racemique. Ann Chim Phys 1850;28:56–99. [2] Henderson GH, Rule HG. A new method of resolving a racemic compound. J Chem Soc 1939;1568–73. € [3] Prelog V, Wieland P. Uber die Spaltung der Tr€ oger‘schen Base in optische Antipoden, ein Beitrag zur Stereochemie des dreiwertigen Stickstoffs. Helv Chim Acta 1944;27:1127–34. [4] Kotake M, Sakan T, Nakamura N, Senoh S. Resolution into optical isomers of some amino acids by paper chromatography. J Am Chem Soc 1951;73:2973–4. € [5] Krebs J, Diewald J, Arlitt H, Wagner JA. Uber die chromatographische Spaltung von Racematen. II. Versuche zur Aktivierung von oktaederf€ ormig gebauten Komplexen. ZAAC J Inorg Gen Chem 1956;287(1–2):98–105. € [6] Krebs H, Wagner JA, Diewald J. Uber die chromatographische Spaltung von Racematen III. Versuche zur Aktivierung organischer Hydroxy- und Aminoverbindungen mit asymmetrischem C-Atom. Chem Ber 1956;89:1875–83. [7] Mayer W, Merger F. Darstellung optisch aktiver Catechine durch Racemattrennung mit Hilfe der Adsorptionschromatographie an Cellulose. Liebigs Ann Chem 1961;644:65–9. [8] Gil-Av E, Feibush B, Charles-Sigler R. Separation of enantiomers by gas liquid chromatography with an optically active stationary phase. Tetrahedron Lett 1966;7:1009–15. [9] L€uttringhaus A, Hess U, Rosenbaum HJ. Optisch aktives 4.5.6.7-dibenzo1.2-dithiacyclooctadien, Zeitschrift f€ur Naturforschung. Section B J Chem Sci 1967;22(12):1296–301.

423

424

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

[10] Rogozhin SV, Davankov VA. Ligand chromatography on asymmetric complexforming sorbents as a new method for resolution of racemates. J Chem Soc D Chem Commun 1971;490a. [11] Davankov VA, Rogozhin SV, Semechkin AV, Sachkova TP. Ligand-exchange chromatography of racemates. Influence of the degree of saturation of the asymmetric resin by metal ions on ligand exchange. J Chromatogr A 1973;82:359–65. [12] Davankov VA. 50 years of Chiral liquid chromatography. How it started. J Chromatogr A 2022;1922(1668), 462910. [13] Sogah GDY, Cram DJ. Total chromatographic optical resolutions of α-amino acid and ester salts through chiral recognition by a host covalently bound to polystyrene resin. J Am Chem Soc 1976;98:3038–41. [14] Sogah GDY, Cram DJ. Chromatographic optical resolution through chiral complexation of amino ester salts by a host covalently bound to silica gel. J Am Chem Soc 1975;97:1259–61. [15] Shinbo T, Yamaguchi T, Nishimura K, Sugiura M. Chromatographic separation of racemic amino acids by use of chiral crown ether-coated reversed-phase packings. J Chromatogr A 1987;405:145–53. [16] Pirkle WH, House DW, Finn JM. Broad spectrum resolution of optical isomers using chiral high-performance liquid chromatographic bonded phases. J Chromatogr A 1980;192:143–58. [17] Yuki H, Okamoto Y, Okamoto I. Resolution of racemic compounds by optically active poly(triphenylmethyl methacrylate). J Am Chem Soc 1980;102:6356–8. [18] Okamoto Y, Okamoto I, Yuki H, Murata S, Noyori R, Takaya H. Novel packing material for optical resolution: (+)-poly(triphenylmethyl methacrylate)coated on macroporous silica gel. J Am Chem Soc 1981;103:6971–3. [19] Allenmark S, Bomgren B, Boren H. Direct liquid chromatographic separation of enantiomers on immobilized protein stationary phases. III. Optical resolution of a series of N-aroyl D,L-amino acids by high-performance liquid chromatography on bovine serum albumin covalently bound to silica. J Chromatogr A 1983;264:63–8. [20] Allenmark S, Bomgren B, Boren H. Direct liquid chromatographic separation of enantiomers on immobilized protein stationary phases. IV. Molecular interaction forces and retention behaviour in chromatography on bovine serum albumin as a stationary phase. J Chromatogr A 1984;316:617–24. [21] Hermansson J. Direct liquid chromatographic resolution of racemic drugs using α1-acid glycoprotein as the chiral stationary phase. J Chromatogr A 1983;269:71–80. [22] Hesse G, Hagel R. A complete separation of a racemic mixture by elution chromatography on cellulose triacetate. Chromatographia 1973;6:277–80. [23] Okamoto Y, Kawashima M, Yamamoto K, Hatada K. Useful chiral packing materials for high-performance liquid chromatographic resolution. Cellulose triacetate and tribenzoate coated on macroporous silica gel. Chem Lett 1984;13:739–42. [24] Ichida A, Shibata T, Okamoto Y, Yuki Y, Namikoshi H, Toda Y. Resolution of enantiomers by HPLC on cellulose derivatives. Chromatographia 1984;19:280–4. [25] Okamoto Y, Kawashima M, Hatada K. Useful Chiral packing materials for highperformance liquid chromatographic resolution of enantiomers: phenylcarbamates of polysaccharides coated on silica gel. J Am Chem Soc 1984;106:5357–9. [26] Okamoto Y, Kawashima M, Hatada K. Controlled chiral recognition of cellulose triphenylcarbamate derivatives supported on silica gel. J Chromatogr A 1986;363:173–86.

References

[27] Armstrong DW, DeMond W. Cyclodextrin bonded phases for the liquid chromatographic separation of optical, geometrical, and structural isomers. J Chromatogr Sci 1984;22(9):411–5. [28] Armstrong DW, Chang S-C, Stalcup AM, Hilton ML, Duncan JD, Faulkner JR. Derivatized cyclodextrins for normal-phase liquid chromatographic separation of enantiomers. Anal Chem 1990;62(15):1610–5. [29] Blaschke G, Br€oker W, Fraenkel W. Enantiomeric resolution by HPLC on silicagel-bound, optically active polyamides. Angew Chem Int Ed Engl 1986;25(9):830–1. [30] Miwa T, Ichikawa M, Tsuno M, Hattori T, Miyakawa T, Kayano M, Miyake Y. Direct liquid chromatographic resolution of racemic compounds. Use of ovomucoid as a column ligand. Chem Pharm Bull 1987;35(2):682–6. [31] Chankvetadze B, Yashima E, Okamoto Y. Tris(chloro- and methyl-disubstituted phenylcarbamate)s of cellulose as chiral stationary phases for chromatographic enantioseparation. Chem Lett 1993;22(4):617–20. [32] Chankvetadze B, Yashima E, Okamoto Y. Chloromethylphenylcarbamate derivatives of cellulose as chiral stationary phases for high-performance liquid chromatography. J Chromatogr A 1994;670(1–2):39–49. [33] Armstrong DW, Tang Y, Chen S, Zhou Y, Bagwill C, Chen JR. Macrocyclic antibiotics as a new class of chiral selectors for liquid chromatography. Anal Chem 1994;66 (9):1473–84. [34] Armstrong DW, Liu Y, Ekborgott KH. A covalently bonded teicoplanin chiral stationary phase for HPLC enantioseparations. Chirality 1995;7(6):474–97. [35] Rosini C, Bertucci C, Pini D, Altemura P, Salvadori P. Cinchona alkaloids for preparing new, easily accessible chiral stationary phases. I. 11-(10,11-dihydro-60 -methoxy-cinchonan-9-OL)-tiopropylsilanized silica. Tetrahedron Lett 1985;26(28):3361–4. [36] L€ammerhofer M, Lindner W. Quinine and quinidine derivatives as chiral selectors. I. Brush type chiral stationary phases for high-performance liquid chromatography based on cinchonan carbamates and their application as chiral anion exchangers. J Chromatogr A 1996;741(1):33–48. [37] Okamoto Y, Aburatani R, Miura S-i, Hatada K. Chiral stationary phases for HPLC: cellulose tris(3,5-dimethylphenylcarbamate) and tris(3,5-dichlorophenylcarbamate) chemically bonded to silica gel. J Liq Chromatogr 1987;10(8–9):1613–28. [38] Kimata K, Tsuboi R, Hosoya K, Tanaka N. Chemically bonded chiral stationary phase prepared by the polymerization of cellulose p-Vinylbenzoate. Anal Methods Instrum 1993;1:23–9. [39] Francotte E. Photochemically cross-linked polysaccharide derivatives as supports for the chromatographic separation of enantiomers. WO 9627615 A1 Patent; 1996. [40] Francotte E. Thermally immobilized polysaccharide derivatives. WO1997EP03225 EU Patent; 1997. [41] Sun P, Wang C, Breitbach ZS, Zhang Y, Armstrong DW. Development of new HPLC chiral stationary phases based on native and derivatized cyclofructans. Anal Chem 2009;81(24):10215–26. [42] Davankov VA. In: Krstulovic AM, editor. Chiral separations by HPLC. New York: Ellis Horwood Limited; 1989. p. 175–93. Chapter 8. [43] Blaschke G, Chankvetadze B. Chapter 6: Resolution of racemic mixtures. In: Gualtieri F, editor. New trends in synthetic medicinal chemistry. Weinheim, Germany: WileyVCH; 2000. p. 139–73.

425

426

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

[44] Peluso P, Chankvetadze B. Recognition in the domain of molecular chirality: from noncovalent interactions to separation of enantiomers. Chem Rev 2022;122:13235–400. [45] Gassmann E, Kuo JE, Zare RN. Electrokinetic separation of chiral compounds. Science 1985;230(4727):813–4. [46] Davankov VA. Chiral selectors with chelating properties in liquid chromatography: fundamental reflections and selective review of recent developments. J Chromatogr A 1994;666:55–76. [47] Pedersen CJ. Cyclic polyethers and their complexes with metal salts. J Am Chem Soc 1967;89(26):7017–36. [48] Wudl F, Gaeta F. A versatile synthesis of chiral crown amino-ethers. J Chem Soc Chem Commun 1972;0(2):107a. [49] Kyba EP, Sousa LR, Sogah GDY, Siegel MG, Cram DJ. Chiral, hinged, and functionalized multiheteromacrocycles. J Am Chem Soc 1973;95(8):2691–2. [50] Hyun MH. Development and application of crown ether-based HPLC chiral stationary phases. Bull Korean Chem Soc 2005;26(8):1153–63. [51] Paik M-J, Kang JS, Huang B-S, Carey JR, Lee W. Development and application of chiral crown ethers as selectors for chiral separation in high-performance liquid chromatography and nuclear magnetic resonance spectroscopy. J Chromatogr A 2013;1274: 1–5. [52] Behr J-P, Girodeau J-M, Hayward RC, Lehn J-M, Sauvage J-P. Molecular receptors. Functionalized and chiral macrocyclic polyethers derived from tartaric acid. Helv Chim Acta 1980;63(7):2096–111. [53] Kuhn R, Erni F, Bereuter T, H€ausler J. Chiral recognition and enantiomeric resolution based on host-guest complexation with crown ethers in capillary zone electrophoresis. Anal Chem 1992;64(22):2815–20. [54] Machida Y, Nishi H, Nakamura K, Nakai H, Sato T. Enantiomer separation of amino compounds by a novel chiral stationary phase derived from crown ether. J Chromatogr A 1998;805(1–2):85–92. [55] Hyun MH, Jin JS, Koo HJ, Lee W. Liquid chromatographic resolution of racemic amines and amino alcohols on a chiral stationary phase derived from crown ether. J Chromatogr A 1999;837:75–82. [56] Mikes F, Boshart B. Resolution of optical isomers by high-performance chromatography. A comparison of two selector-selectand system. J Chromatogr A 1978;149:455–64. [57] Hara S, Dobashi A. Liquid chromatographic resolution of enantiomers on chiral amide bonded-silica gel normal phase: separation of racemic α - amino acid derivatives by nacetyl-l-valyl-aminopropyl-silanized silica (AVA) phase. J Liq Chromatogr 1979;2 (6):883–9. [58] Pirkle WH, Beare SD, Burlingame TG. Optically active nuclear magnetic resonance solvents. VIII. Resolution of 2,2,2-trifluorophenyIethanol. J Org Chem 1969;34(2): 470–1. [59] Pirkle WH, Sikkenga DL, Pavlin MS. Nuclear magnetic resonance determination of enantiomeric composition and absolute configuration of γ-lactones using chiral 2,2,2-trifluoro-1-(9-anthryl)ethanol. J Org Chem 1977;42(2):384–7. [60] Pirkle WH, Tsipouras A. 3,5-Dinitrobenzoyl amino acid esters. Broadly applicable chiral solvating agents for NMR determination of enantiomeric purity. Tetrahedron Lett 1985;26:2989–92.

References

[61] Pirkle WH, Pochapsky TC. Considerations of chiral recognition relevant to the liquid chromatographic separation of enantiomers. Chem Rev 1989;89:347–62. [62] Pirkle WH, Welch CJ, Lamm B. Design, synthesis, and evaluation of an improved enantioselective naproxen selector. J Org Chem 1992;57:3854–60. [63] Dalgliesh CE. The optical resolution of aromatic amino-acids on paper chromatograms. J Chem Soc 1952;3940–2. [64] Mikesˇ F, BoshartG G-AE. Resolution of optical isomers by high-performance liquid chromatography, using coated and bonded chiral charge-transfer complexing agents as stationary phases. J Chromatogr A 1976;122:205–21. [65] Cramer F. Einschlußverbindungen der Cyclodextrine. Angew Chem 1952;64:136. [66] Dębowski J, Sybilska D, Jurczak J. β-Cyclodextrin as a chiral component of the mobile phase for separation of mandelic acid into enantiomers in reversed-phase systems of high-performance liquid chromatography. J Chromatogr A 1982;237:303–6. [67] Hess H, Burger G, Musso H. Complete enantiomer separation by chromatography on potato starch. Angew Chem Int Ed 1978;17:612–4. [68] SteckelbergW BM, Musso H. Notiz zur Antipodentrennung von Biphenylderivaten durch Chromatographie. Chem Ber 1968;101(4):1519–21. [69] Yamamoto C, Hayashi T, Okamoto Y. High-performance liquid chromatographic enantioseparation using chitin carbamate derivatives as chiral stationary phase. J Chromatogr A 2003;1021:83–91. [70] Shen J, Okamoto Y. Efficient separation of enantiomers using stereoregular chiral polymers. Chem Rev 2016;116:1094–138. [71] Chankvetadze B. Recent trends in preparation, investigation and application of polysaccharide-based chiral stationary phases for separation of enantiomers in highperformance liquid chromatography. TrAC Trends Anal Chem 2020;122, 115709. [72] Chankvetadze B. Recent developments on polysaccharide-based Chiral stationary phases for liquid-phase separation of enantiomers. J Chromatogr A 2012;1269:26–51. [73] Chankvetadze B, Yashima E, Okamoto Y. Dimethyl-, dichloro- and chloromethylphenylcarbamate derivatives of amylose as chiral stationary phases for high performance liquid chromatography. J Chromatogr A 1995;694:101–9. [74] Chankvetadze B, Chankvetadze L, Sidamonidze S, Kasashima E, Yashima E, Okamoto Y. 3-Fluoro-, 3-bromo-, and 3-chloro-5-methylphenylcarbamates of cellulose and amylose as chiral stationary phases for HPLC enantioseparation. J Chromatogr A 1997;787:67–77. [75] Chankvetadze B, Yamamoto C, Okamoto Y. Enantioseparation of selected chiral sulfoxides using polysaccharide-type chiral stationary phases and polar organic, polar aqueous-organic and normal-phase eluents. J Chromatogr A 2001;922:127–37. [76] Chankvetadze B. Our research co-operation with professor Yoshio Okamoto. Chirality 2022;34:630–45. [77] Albals D, Vander Heyden Y, Schmid MG, Chankvetadze B, Mangelings D. Chiral separations of cathinone and amphetamine-derivatives: comparative study between capillary electrochromatography, supercritical fluid chromatography and three liquid chromatographic modes. J Pharm Biomed Anal 2016;121:232–43. [78] Riasova´ P, Ja´c P, Pola´sˇek M, Vander Heyden Y, Mangelings D. Coupling of chiral and achiral stationary phases in supercritical fluid chromatography: evaluating and improving retention prediction. J Chromatogr A 2022;1667:462883.

427

428

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

[79] La Z, Charton J, Etienne L, Bourey J, Lipka E. Supercritical fluid chromatography and liquid chromatography for isomeric separation of a multiple chiral centers analyte. J Chromatogr A 2021;1651:462270. [80] Fanali S, Chankvetadze B. History, advancement, bottlenecks and future of chiral capillary electrochromatography. J Chromatogr A 2021;1637, 461832. [81] De Cesaris MG, D’Orazio G, Gentili A, Fanali C, Takaishvili N, Chankvetadze B, Fanali S. Enantioseparation of selected chiral agrochemicals by using nano-liquid chromatography and capillary electrochromatography with amylose tris(3-chloro-5-methylphenylcarbamate) covalently immobilized onto silica. J Chromatogr A 2022;1673, 463128. [82] Oliveros L, Lopez P, Minguillon C, Franco P. Chiral chromatographic discrimination ability of a cellulose 3,5-dimethylphenylcarbamate/10-undecenoate mixed derivative fixed on several chromatographic matrices. J Liq Chromatogr 1995;18(8):1521–32. [83] Francotte E, Huynh D. Immobilized halogenophenylcarbamate derivatives of cellulose as novel stationary phases for enantioselective drug analysis. J Pharm Biomed Anal 2002;27(3–4):421–9. [84] Chankvetadze B, Ikai T, Yamamoto C, Okamoto Y. High-performance liquid chromatographic enantioseparations on monolithic silica column containing covalently attached 3,5-dimethylphenylcarbamate derivative of cellulose. J Chromatogr A 2004;1042(1–2):55–60. [85] Shen J, Ikai T, Okamoto Y. Synthesis and application of immobilized polysaccharidebased chiral stationary phases for enantioseparation by high-performance liquid chromatography. J Chromatogr A 2014;1363:51–61. [86] Blaschke G. Chromatographic resolution of racemates. Angew Chem Int Ed Engl 1971;10(7):520–1. [87] Blaschke G. Chromatographic resolution of racemates. New analytical methods. Angew Chem Int Ed Engl 1980;19(1):13–24. [88] Arlt D, B€omer B, Grosser R, Lange W. New chiral polyamide stationary phases for chromatographic enantiomer separation. Angew Chem Int Ed Engl 1991;30 (12):1662–4. [89] Schulte M, Devant R, Grosser R. Enantioseparation of gantofiban precursors on chiral stationary phases of the poly-(N-acryloyl amino acid derivative)-type. J Pharm Biomed Anal 2002;27:627–37. [90] Miwa T, Miyakawa T, Kayano M, Miyake Y. Application of an ovomucoid-conjugated column for the optical resolution of some pharmaceutically important compounds. J Chromatogr A 1987;408:316–22. [91] Haginaka J. HPLC Chiral stationary phases based on a glycoprotein. Trends Glycosci Glycotechnol 1997;9(49):399–407. [92] Haginaka J. Protein-based chiral stationary phases for high-performance liquid chromatography enantioseparations. J Chromatogr A 2001;906(1–2):253–73. [93] Haginaka J. Progress in Chiral stationary phases based on proteins and glycoproteins. Chem Pharm Bull 2022;70(7):458–68. [94] Grubhofer N, Schleith L. Die Spaltung von racemischer Mandelsaure mit einem optisch aktiven Anionenaustauscher. Biol Chem 1954;296:262–6. [95] Izumoto S, Sakaguchi U, Yoneda H. Stereochemical aspects of the optical resolution of cis(N)-[co(N)2(O)4]- complexes by reversed-phase ion-pair chromatography with cinchona alkaloid cations as the ion-pairing reagents. Bull Chem Soc Jpn 1983;56:1646–51. [96] Pettersson C. Chromatographic separation of enantiomers of acids with quinine as chiral counter ion. J Chromatogr A 1984;316:553–67.

References

[97] Peluso P, Dessı` A, Dallocchio R, Mamane V, Cossu S. Recent studies of docking and molecular dynamics simulation for liquid-phase enantioseparations. Electrophoresis 2019;40:1881–96. [98] Hoffmann CV, Pell R, L€ammerhofer M, Lindner W. Synergistic effects on enantioselectivity of zwitterionic chiral stationary phases for separations of chiral acids, bases, and amino acids by HPLC. Anal Chem 2008;80:8780–9. [99] L€ammerhofer M, Lindner W. Liquid chromatographic enantiomer separation and chiral recognition by cinchona alkaloid-derived enantioselective separation materials. Adv Chromatogr 2008;46:1–107. [100] L€ammerhofer M. Chiral recognition by enantioselective liquid chromatography: mechanisms and modern chiral stationary phases. J Chromatogr A 2010;1217(6):814–56. [101] Ilisz I, Bajtai A, Lindner W, Peter A. Liquid chromatographic enantiomer separations applying chiral ion-exchangers based on cinchona alkaloids. J Pharm Biomed Anal 2018;159:127–52. [102] Tana´cs D, Berkecz R, Shahmohammadi S, Forro E, Armstrong DW, Peter A, Ilisz I. Macrocyclic glycopeptides- and derivatized cyclofructan-based chiral stationary phases for the enantioseparation of fluorinated β-phenylalanine analogs. J Pharm Biomed Anal 2022;219, 114912. [103] Dobashi Y, Hara S. Direct resolution of enantiomers by liquid chromatography with the novel chiral stationary phase derived from (R,R)-tartramide. Tetrahedron Lett 1985;26 (35):4217–20. [104] Andersson S, Allenmark S, M€oller P, Persson B, Sanchez D. Chromatographic separation of enantiomers on N,N’-diallyl-L-tartardiamide-based network-polymeric chiral stationary phases. J Chromatogr A 1996;741(1):23–31. [105] Yoon SS, Still CW. An exceptional synthetic receptor for peptides. J Am Chem Soc 1993;115:823–4. [106] Still CW. Discovery of sequence-selective peptide binding by synthetic receptors using encoded combinatorial libraries. Acc Chem Res 1996;29:155–63. [107] Borchardt A, Still CW. Synthetic receptor binding elucidated with an encoded combinatorial library. J Am Chem Soc 1994;116:373–4. [108] Pan Z, Clark Still W. Macrocyclic oligomers of isophthalic acid and transl,2-diaminocyclohexane - building blocks for synthetic peptide receptors. Tetrahedron Lett 1996;37:8699–702. [109] Gasparrini F, Misiti D, Villani C, Borchardt A, Burger MT, Clark SW. Enantioselective recognition by a new chiral stationary phase at receptor level. J Org Chem 1995;60:4314–5. [110] Gasparrini F, Misiti D, Clark Still W, Villani C. Enantioselective and diastereoselective binding study of silica bound macrobicyclic receptors by HPLC. J Org Chem 1997;62:8221–4. [111] Chankvetadze B, Yamamoto C, Okamoto Y. Extremely high enantiomer recognition in HPLC separation of racemic 2-(benzylsulfinyl)benzamide using cellulose tris (3,5-dichlorophenylcarbamate) as a chiral stationary phase. Chem Lett A 2000;29:1176–7. [112] Cirilli R, Alcaro S, Fioravanti R, Ferretti R, Bolasco A, Gallinella B, Faggi C. A chromatographic study on the exceptional enantioselectivity of cellulose tris(4methylbenzoate) towards C5-chiral 4,5-dihydro-(1H)-pyrazole derivatives. J Chromatogr A 2011;1218:5653–7. [113] Khatiashvili T, Kakava R, Matarashvili I, Tabani H, Fanali C, Volonterio A, Farkas T, Chankvetadze B. Separation of enantiomers of selected chiral sulfoxides with cellulose

429

430

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

[114] [115]

[116]

[117]

[118]

[119]

[120] [121]

[122]

[123]

[124]

[125]

[126]

[127]

[128]

tris(4-chloro-3-methylphenylcarbamate)-based chiral columns in high-performance liquid chromatography with very high separation factor. J Chromatogr A 2018;1545:59–66. Wulff G, Sarhan A, Zabrocki K. Enzyme-analogue built polymers and their use for the resolution of racemates. Tetrahedron Lett 1973;14(44):4329–32. Sellergren B, Ekberg B, Mosbach K. Molecular imprinting of amino acid derivatives in macroporous polymers. Demonstration of substrate- and enantio-selectivity by chromatographic resolution of racemic mixtures of amino acid derivatives. J Chromatogr A 1985;347:1–10. Hosoya K, Yoshizako K, Shirasu Y, Kimata K, Araki T, Tanaka N, Haginaka J. Molecularly imprinted uniform-size polymer-based stationary phase for highperformance liquid chromatography structural contribution of cross-linked polymer network on specific molecular recognition. J Chromatogr A 1996;728(1–2):139–47. Haginaka J, Kagawa C. Chiral resolution of derivatized amino acids using uniformly sized molecularly imprinted polymers in hydro-organic mobile phases. Anal Bioanal Chem 2004;378(8):1907–12. Fu Q, Sanbe H, Kagawa C, Kunimoto K-K, Haginaka J. Uniformly sized molecularly imprinted polymer for (S)-Nilvadipine. Comparison of chiral recognition ability with HPLC chiral stationary phases based on a protein. Anal Chem 2003;75(2):191–8. Haginaka J, Sanbe H. Uniformly sized molecularly imprinted polymer for (S)naproxen: retention and molecular recognition properties in aqueous mobile phase. J Chromatogr A 2001;913(1–2):141–6. Haginaka J. Monodispersed, molecularly imprinted polymers as affinity-based chromatography media. J Chromatogr B 2008;866(1–2):3–13. Wang Y, Chen J-K, Xiong L-X, Wang B-J, Xie S-M, Zhang J-H, Yuan L-M. Preparation of novel chiral stationary phases based on the chiral porous organic cage by thiolene click chemistry for enantioseparation in HPLC. Anal Chem 2022;94(12):4961–9. Yuan C, Jia W, Yu Z, Li Y, Zi M, Yuan L-M, Cui Y. Are highly stable covalent organic frameworks the key to universal chiral stationary phases for liquid and gas chromatographic separations? J Am Chem Soc 2022;144(2):891–900. Xu N-Y, Guo P, Chen J-K, Zhang J-H, Wang B-J, Xie S-M, Yuan L-M. Chiral coreshell microspheres β-CD-COF@SiO2used for HPLC enantioseparation. Talanta 2021;235, 122754. Peng B, Fu S, Li Y, Zhang J, Xie S, Li L, Lyu Y, Duan A, Chen X, Yuan L. Enantioseparation by HPLC using an inorganic chiral mesoporous silica with highly-ordered structure. Chem Res Chin Univ 2019;35(6):978–82. Castells CB, Carr PW. Fast enantioseparations of basic analytes by high-performance liquid chromatography using cellulose tris(3,5-dimethylphenylcarbamate)-coated zirconia stationary phases. J Chromatogr A 2000;904(1):17–33. Park JH, Whang YC, Jung YJ, Okamoto Y, Yamamoto C, Carr PW, McNeff CV. Separation of racemic compounds on amylose and cellulose dimethylphenylcarbamatecoated zirconia in HPLC. J Sep Sci 2003;26(15–16):1331–6. Ling F, Brahmachary E, Xu M, Svec F, Frechet JMJ. Polymer-bound cellulose phenylcarbamate derivatives as chiral stationary phases for enantioselective HPLC. J Sep Sci 2003;26(15–16):1337–46. Tanaka N, Nagayama H, Kobayashi H, Ikegami T, Hosoya K, Ishizuka N, Minakuchi H, Nakanishi K, Cabrera K, Lubda D. Monolithic silica columns for HPLC, micro-HPLC, and CEC. J High Resol Chromatogr 2000;23:111–6.

References

[129] Chankvetadze B. Monolithic chiral stationary phases for liquid-phase enantioseparation techniques. J Sep Sci 2010;33:305–14. [130] Guo J, Wang Q, Xu D, Crommen J, Jiang Z. Recent advances in preparation and applications of monolithic chiral stationary phases. TrAC Trends Anal Chem 2020;123, 115774. [131] Chankvetadze B, Yamamoto C, Okamoto Y. Very fast enantioseparations in HPLC using cellulose tris(3,5-dimethylphenylcarbamate) as chiral stationary phase. Chem Lett 2003;32:850–1. [132] Peters EC, Lewandowski K, Petro M, Svec F, Frechet JM. Chiral electrochromatography with a ‘moulded’ rigid monolithic capillary column. J Anal Commun 1998;35:83–6. [133] L€ammerhofer M, Svec F, Frechet JMJ, Lindner W. Chiral monolithic columns for enantioselective capillary electrochromatography prepared by copolymerization of a monomer with quinidine functionality. 2. Effect of chromatographic conditions on the chiral separations. Anal Chem 2000;72:4623–8. [134] Echevarria RN, Keuchkarian S, Villarrochel-Rocha J, Sapag K, Reta M. Organic monolithic capillary columns coated with cellulose tris(3,5-dimethylphenyl carbamate) for enantioseparations by capillary HPLC. Microchem J 2019;149, 104011. [135] Chankvetadze B, Ikai T, Yamamoto C, Okamoto Y. High-performance liquid chromatographic enantioseparations on monolithic silica column containing covalently attached 3,5-dimethylphenylcarbamate derivative of cellulose. J Chromatogr A 2004;1042(1–2):55–60. [136] Lubda D, Cabrera K, Nakanishi K, Lindner W. Monolithic silica columns with chemically bonded β-cyclodextrin as a stationary phase for enantiomer separations of chiral pharmaceuticals. Anal Bioanal Chem 2003;377:892–901. [137] Lubda D, Lindner W. Monolithic silica columns with chemically bonded tertbutylcarbamoylquinine chiral anion-exchanger selector as a stationary phase for enantiomer separations. J Chromatogr A 2004;1036:135–43. [138] Calleri E, Massolini G, Lubda D, Temporini C, Loiodice F, Caccialanza G. Evaluation of a monolithic epoxy silica support for penicillin G acylase immobilization. J Chromatogr A 2004;1031:93–100. [139] Gumustas M, Zalewski P, Ozkan SA, Uslu B. The history of the core–shell particles and applications in active pharmaceutical ingredients via liquid chromatography. Chromatographia 2019;82:17–48. [140] Reischl RJ, Hartmanova L, Carrozzo M, Huszar M, Fruhauf P, Lindner W. Chemoselective and enantioselective analysis of proteinogenic amino acids utilizing N-derivatization and 1-D enantioselective anion-exchange chromatography in combination with tandem mass spectrometric detection. J Chromatogr A 2011;1218:8379–87. [141] Lomsadze K, Jibuti G, Farkas T, Chankvetadze B. Comparative high-performance liquid chromatography enantioseparations on polysaccharide based chiral stationary phases prepared by coating totally porous and core-shell silica particles. J Chromatogr A 2012;1234:50–5. [142] Kharaishvili Q, Jibuti G, Farkas T, Chankvetadze B. Further proof to the utility of polysaccharide-based chiral selectors in combination with superficially porous silica particles as effective chiral stationary phases for separation of enantiomers in highperformance liquid chromatography. J Chromatogr A 2016;1467:163–8. [143] Bezhitashvili L, Bardavelidze A, Ordjonikidze T, Farkas T, Chitty M, Chankvetadze B. Effect of pore-size optimization on the performance of polysaccharide-based

431

432

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

[144]

[145]

[146]

[147]

[148]

[149]

[150] [151]

[152]

[153] [154]

[155]

[156]

[157]

superficially porous chiral stationary phases for separation of enantiomers in highperformance liquid chromatography. J Chromatogr A 2017;1482:32–8. Bezhitashvili L, Bardavelidze A, Mskhiladze A, Volonterio A, Gumustas M, Ozkan S, Farkas T, B. Chankvetadze B. Application of cellulose 3,5-dichlorophenylcarbamate covalently immobilized on superficially porous silica for separation of enantiomers in ultra high-performance liquid chromatography. J Chromatogr A 2018;1571:132–9. Khundadze N, Pantsulaia S, Fanali C, Farkas T, Chankvetadze B. On our way to subsecond separations of enantiomers in high-performance liquid chromatography. J Chromatogr A 2018;1572:37–43. Pantsulaia S, Targamadze K, Khundadze N, Kharaishvili Q, Volonterio A, Chity M, Farkas T, Chankvetadze B. Advantages and current limitations of superficially porous silica materials for preparation of highly efficient polysaccharide-based chiral stationary phases for liquid-phase separation of enantiomers. J Chromatogr A 2020;1625, 461297. Dolzan MD, Spudeit DA, Breitbach ZS, Barber WE, Micke GA, Armstrong DW. Comparison of superficially porous and fully porous silica supports used for a cyclofructan 6 hydrophilic interaction liquid chromatographic stationary phase. J Chromatogr A 2014;1365:124–30. Patel DC, Breitbach ZS, Wahab MF, Barhate CL, Armstrong DW. Gone in seconds: praxis, performance, and peculiarities of ultrafast chiral liquid chromatography with superficially porous particles. Anal Chem 2015;87(18):9137–48. Barhate CL, Breitbach ZS, Pinto EC, Regalado EL, Welch CJ, Armstrong DW. Ultrafast separation of fluorinated and desfluorinated pharmaceuticals using highly efficient and selective chiral selectors bonded to superficially porous particles. J Chromatogr A 2015;1426:241–7. Wahab MF, Wimalasinghe RM, Wang Y, Barhate CL, Patel DC, Armstrong DW. Salient sub-second separations. Anal Chem 2016;88(17):8821–6. Patel DC, Wahab MF, Armstrong DW, Breitbach ZS. Advances in high-throughput and high-efficiency chiral liquid chromatographic separations. J Chromatogr A 2016;1467:2–18. Barhate CL, Regalado EL, Contrella ND, Lee J, Jo J, Makarov AA, Armstrong DW, Welch CJ. Ultrafast chiral chromatography as the second dimension in two-dimensional liquid chromatography experiments. Anal Chem 2017;89(6):3545–53. Patel DC, Wahab MF, O’Haver TC, Armstrong DW. Separations at the speed of sensors. Anal Chem 2018;90(5):3349–56. Roy D, Armstrong DW. Fast super/subcritical fluid chromatographic enantioseparations on superficially porous particles bonded with broad selectivity Chiral Selectors relative to fully porous particles. J Chromatogr A 2019;1605, 360339. Ismail OH, Pasti L, Ciogli A, Villani C, Kocergina J, Anderson S, Gasparrini F, Cavazzini A, Catani M. Pirkle-type chiral stationary phase on core–shell and fully porous particles: are superficially porous particles always the better choice toward ultrafast high-performance enantioseparations? J Chromatogr A 2016;1466:96–104. Catani M, Ismail OH, Gasparrini F, Antonelli M, Pasti L, Marchetti N, Felletti S, Cavazzini A. Recent advancements and future directions of superficially porous chiral stationary phases for ultrafast high-performance enantioseparations. Analyst 2017;142 (4):555–66. Ismail OH, Antonelli M, Ciogli A, Villani C, Cavazzini A, Catani M, Felletti S, Bell DS, Gasparrini F. Future perspectives in high efficient and ultrafast chiral liquid

References

[158]

[159]

[160]

[161]

[162]

[163]

[164]

[165]

[166]

[167]

[168]

[169]

chromatography through zwitterionic teicoplanin-based 2-μm superficially porous particles. J Chromatogr A 2017;1520:91–102. Catani M, Felletti S, Ismail OH, Gasparrini F, Pasti L, Marchetti N, De Luca C, Costa V, Cavazzini A. New frontiers and cutting edge applications in ultra high performance liquid chromatography through latest generation superficially porous particles with particular emphasis to the field of chiral separations. Anal Bioanal Chem 2018;410 (10):2457–65. Felletti S, De Luca C, Ismail OH, Pasti L, Costa V, Gasparrini F, Cavazzini A, Catani M. On the effect of chiral selector loading and mobile phase composition on adsorption properties of latest generation fully- and superficially-porous whelk-O1 particles for high-efficient ultrafast enantioseparations. J Chromatogr A 2018;1579:41–8. Felletti S, De Luca C, Lievore G, Pasti L, Chenet T, Mazzoccanti G, Gasparrini F, Cavazzini A, Catani M. Investigation of mass transfer properties and kinetic performance of high-efficiency columns packed with C18 Sub-2 μm fully and superficially porous particles. J Sep Sci 2020;43(9–10):1737–45. Felletti S, Catani M, Mazzoccanti G, De Luca C, Lievore G, Buratti A, Pasti L, Gasparrini F, Cavazzini A. Mass transfer kinetics on modern whelk-O1 chiral stationary phases made on fully- and superficially-porous particles. J Chromatogr A 2021;1637, 461854. Ismail OH, Catani M, Mazzoccanti G, Felletti S, Manetto S, De Luca C, Ye M, Cavazzini A, Gasparrini F. Boosting the enantioresolution of zwitterionic-teicoplanin chiral stationary phases by moving to wide-pore core-shell particles. J Chromatogr A 2022;1676:463190. Woiwode U, Reischl RJ, Buckenmaier S, Lindner W, L€ammerhofer M. Imaging peptide and protein chirality via amino acid analysis by chiral  chiral two-dimensional correlation liquid chromatography. Anal Chem 2018;90(13):7963–71. Woiwode U, Neubauer S, Lindner W, Buckenmaier S, L€ammerhofer M. Enantioselective multiple heartcut two-dimensional ultra-high-performance liquid chromatography method with a coreshell chiral stationary phase in the second dimension for analysis of all proteinogenic amino acids in a single run. J Chromatogr A 2018;1562:69–77. Schmitt K, Woiwode U, Kohout M, Zhang T, Lindner W, L€ammerhofer M. Comparison of small size fully porous particles and superficially porous particles of chiral anionexchange type stationary phases in ultra-high performance liquid chromatography: effect of particle and pore size on chromatographic efficiency and kinetic performance. J Chromatogr A 2018;1569:149–59. Geibel C, Dittrich K, Woiwode U, Kohout M, Zhang T, Lindner W, L€ammerhofer M. Evaluation of superficially porous particle based zwitterionic chiral ion exchangers against fully porous particle benchmarks for enantioselective ultra-high performance liquid chromatography. J Chromatogr A 2019;1603:130–40. Chankvetadze B. Application of enantioselective separation techniques to bioanalysis. TrAC Trends Anal Chem 2021;143, 116332 [Special issue dedicated to the memory of Prof Roman Kaliszan]. Perry JA, Rateike JD, Szczerba TJ. Eluting trace components before major constituents: I. Sensitivity enhancement in analytical determinations of optical purity. J Chromatogr A 1987;389:57–64. Dossou KS, Edorh PA, Chiap P, Chankvetadze B, Servais AC, Fillet M, Crommen J. Determination of enantiomeric purity of S-amlodipine by chiral LC with emphasis on reversal of enantiomer elution order. J Sep Sci 2011;34:1772–80.

433

434

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

[170] Tok KC, G€um€u¸stas¸ M, S€uzen HS, Ozkan SA, Chankvetadze B. The effect of enantiomer elution order on the determination of minor enantiomeric impurity in ketoprofen and enantiomeric purity evaluation of some commercially available dexketoprofen formulations. Molecules 2020;25:5865. [171] Pirkle WH, Hyun MH, Bank B. Rational approach to the design of highly-effective chiral stationary phases. J Chromatogr A 1984;316:585–604. [172] Matarashvili I, Ghughunishvili D, Chankvetadze L, Takaishvili N, Tsintsadze M, Khatiashvili T, Farkas T, Chankvetadze B. Separation of enantiomers of chiral weak acids with polysaccharide-based chiral columns and aqueous mobile phases in highperformance liquid chromatography: typical reversed-phase behavior? J Chromatogr A 2017;1483:86–92. [173] Ravisankar P, Anusha S, Supriya K, Kumar UA. Fundamental chromatographic parameters. Int J Pharm Sci Rev Res 2019;55(2):46–50. [174] Sepsey A, Horvath E, Catani M, Felinger A. The correctness of van ‘t Hoff plots in chiral and achiral chromatography. J Chromatogr A 2020;1611, 460594. [175] Asnin LD, Stepanova MA. van’t Hoff analysis in chiral chromatography. J Sep Sci 2018;41:1319–37. [176] Hearn W, Zhao G. Investigations into the thermodynamics of polypeptide interaction with nonpolar ligands. Anal Chem 1999;71:4874–85. [177] Maisuradze M, Sheklashvili G, Chokheli A, Matarashvili I, Gogatishvili T, Farkas T, Chankvetadze B. Chromatographic and thermodynamic comparison of amylose tris(3chloro-5-methylphenylcarbamate) coated or covalently immobilized on silica in highperformance liquid chromatographic separation of the enantiomers of select chiral weak acids. J Chromatogr A 2019;1602:228–36. [178] Felletti S, De Luca C, Lievore G, Chenet T, Chankvetadze B, Farkas T, Cavazzini A, Catani M. Shedding light on mechanisms leading to convex-upward van Deemter curves on a cellulose tris(4-chloro-3-methylphenylcarbamate)-based chiral stationary phase. J Chromatogr A 2020;1630, 461532. [179] Peluso P, Chankvetadze B. The molecular bases of chiral recognition in 2-(benzylsulfinyl)benzamide enantioseparation. Anal Chim Acta 2021;1141:194–205. [180] Asnin LD, Boteva AA, Krasnykh OP, Stepanova MV, Ali I. Unusual van Deemter plots of optical isomers on a chiral brush-type liquid chromatography column. J Chromatogr A 2019;1592:112–21. [181] West C, Konjaria ML, Shashviashvili N, Lemasson E, Bonnet P, Kakava R, Volonterio A, Chankvetadze B. Enantioseparation of novel chiral sulfoxides on chlorinated polysaccharide stationary phases in supercritical fluid chromatography. J Chromatogr A 2017;1499:174–82. [182] Kasat RB, Zvinevich Y, Hillhouse HW, Thomson KT, Linda Wang N-H, Franses EI. Direct probing of sorbent-solvent interactions for amylose tris (3,5-methylphenylcarbamate) using infrared spectroscopy, X-ray diffraction, solid state NMR, and DFT modeling. J Phys Chem B 2006;110:14114–22. [183] Kasat RB, Chin CY, Thomson KT, Franses EI, Linda Wang N-H. Interpretation of chromatographic retentions of simple solutes with an amylose-based sorbent using infrared spectroscopy and DFT modeling. Adsorption 2006;12:405–16. [184] Manolescu C, Grinberg M, Field C, Ma S, Shwn S, Lee H, Wang Y, Granger A, Chen Q, McCaffrey J, et al. Studies of the interactions of amino alcohols using high performance

References

[185]

[186]

[187]

[188]

[189]

[190]

[191]

[192]

[193]

[194]

[195]

[196] [197] [198]

liquid chromatography with crown ether stationary phases. J Liq Chromatogr Rel Techn 2008;31:2219–34. Tsui H-W, Willing JN, Kasat RB, Wang N-HL, Franses EI. Infrared spectroscopy and molecular simulations of a polymeric sorbent and its enantioselective interactions with benzoin enantiomers. J Phys Chem B 2011;115:12785–800. Yashima E, Yamamoto C, Okamoto Y. NMR studies of chiral discrimination relevant to the liquid chromatographic enantioseparation by a cellulose phenylcarbamate derivative. J Am Chem Soc 1996;118:4036–48. Yamamoto C, Yashima E, Okamoto Y. Structural analysis of amylose tris (3,5-dimethylphenylcarbamate) by NMR relevant to its chiral recognition mechanism in HPLC. J Am Chem Soc 2002;124:12583–9. Wenslow Jr RM, Wang T. Solid-state NMR characterization of amylose tris (3,5-dimethylphenylcarbamate) chiral stationary-phase structure as a function of Mobile-phase composition. Anal Chem 2001;73:4190–41951. Wang F, Wenslow Jr RM, Dowling TM, Mueller KT, Santos I, Wyvratt JM. Characterization of a thermally induced irreversible conformational transition of amylose tris(3,5-dimethylphenylcarbamate) chiral stationary phase in enantioseparation of dihydropyrimidinone acid by quasi-equilibrated liquid chromatography and solid-state NMR. Anal Chem 2003;75:5877–85. Wang T, Wenslow Jr RM. Effects of alcohol mobile-phase modifiers on the structure and chiral selectivity of amylose tris(3,5-dimethylphenylcarbamate) chiral stationary phase. J Chromatogr A 2003;1015:99–110. Ma S, Tsui H-W, Spinelli E, Busacca CA, Franses EI, Wang NHL, Wu L, Lee H, Senanayake C, Gonella N, et al. Insights into chromatographic enantiomeric separation of allenes on cellulose carbamate stationary phase. J Chromatogr A 2014;1362:119–28. Ma S, Shen S, Lee H, Yee N, Senanayake C, Nafie LA, Grinberg N. Vibrational circular dichroism of amylose carbamate: structure and solvent-induced conformational changes. Tetrahedron Asymmetry 2008;19:2111–4. Ma S, Shen S, Lee H, Eriksson M, Zeng X, Xu J, Fandrick K, Yee N, Senanayake C, Grinberg N. Mechanistic studies on the chiral recognition of polysaccharide-based chiral stationary phases using liquid chromatography and vibrational circular dichroism. Reversal of elution order of N-substituted alpha-methyl phenylalanine esters. J Chromatogr A 2009;1216:3784–93. Fornstedt T, Zhong G, Bensetiti Z, Guiochon G. Experimental and theoretical study of the adsorption behavior and mass transfer kinetics of propranolol enantiomers on cellulase protein as the selector. Anal Chem 1996;68:2370–8. Fornstedt T, Sajonz P, Guiochon G. Thermodynamic study of an unusual chiral separation. Propranolol enantiomers on an immobilized cellulase. J Am Chem Soc 1997;119:1254–64. Fornstedt T, Sajonz P, Guiochon G. A closer study of chiral retention mechanisms. Chirality 1998;10:375–81. Fornstedt T. Characterization of adsorption processes in analytical liquid–solid chromatography. J Chromatogr A 2010;1217:792–812. Samuelsson J, Arnell R, Fornstedt T. Potential of adsorption isotherm measurements for closer elucidating of binding in chiral liquid chromatographic phase systems. J Sep Sci 2009;32:1491–506.

435

436

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

[199] Mihlbachler K, Kaczmarski K, Seidel-Morgenstern A, Guiochon G. Measurement and modeling of the equilibrium behavior of the Tr€ oger’s base enantiomers on an amylosebased chiral stationary phase. J Chromatogr A 2002;955:35–52. [200] Cavazzini A, Felinger A, Kaczmarski K, Szabelski P, Guiochon G. Study of the adsorption equilibria of the enantiomers of 1-phenyl-1-propanol on cellulose tribenzoate using a microbore column. J Chromatogr A 2002;953:55–66. [201] Kaczmarski K, Cavazzini A, Szabelski P, Zhou D, Liu X, Guiochon G. Application of the general rate model and the generalized Maxwell–Stefan equation to the study of the mass transfer kinetics of a pair of enantiomers. J Chromatogr A 2002;962:57–67. [202] Felinger A, Cavazzini A, Guiochon G. Numerical determination of the competitive isotherm of enantiomers. J Chromatogr A 2003;986:207–25. [203] Felinger A, Zhou D, Guiochon G. Determination of the single component and competitive adsorption isotherms of the 1-indanol enantiomers by the inverse method. J Chromatogr A 2003;1005:35–49. [204] Kaczmarski K, Guiochon G. Comparison of the binary equilibrium isotherms of the 1indanol enantiomers on three high-performance liquid chromatography columns of different sizes. J Chromatogr A 2003;1015:73–87. [205] Zhou D, Kaczmarski K, Cavazzini A, Liu X, Guiochon G. Modeling of the separation of two enantiomers using a microbore column. J Chromatogr A 2003;1020:199–217. [206] Tsui H-W, Ye P-W, Huang SX. Effect of solvents on the chiral recognition mechanisms of immobilized cellulose-based chiral stationary phase. J Chromatogr A 2021;1637, 461796. [207] Tsui H-W, Zhang H-L, Hsie C-H. Effect of 2-propanol content on solute retention mechanisms determined using amylose tris(3,5-dimethylphenylcarbamate) chiral stationary phase under normal- and reversed-phase conditions. J Chromatogr A 2021;1650, 462226. [208] Matsumoto K, Yamamoto C, Yashima E, Okamoto Y. Chiral recognition of cellulose tris(5-fluoro-2-methylphenylcarbamate) toward (R)- and(S)-1,1’-bi-2-naphthol detected by electron ionization mass spectrometry. Anal Commun 1998;35:63–6. [209] Matsumoto K, Yamamoto C, Yashima E, Okamoto Y. Chiral recognition of cellulose tris(5-fluoro-2-methylpheylcarbamate) toward (R)- and(S)-1,10 -bi-2-naphthol detected by negative ion fast-atom bombardment mass spectrometry. Rapid Commun Mass Spectrom 1999;13:2011–3. [210] Matarashvili I, Kobidze G, Chelidze A, Dolidze G, Beridze N, Jibuti G, Farkas T, Chankvetadze B. The effect of temperature on the separation of enantiomers with coated and covalently immobilized polysaccharide-based chiral stationary phases. J Chromatogr A 2019;1599:172–9. [211] O’Brien T, Crocker L, Thompson R, Thompson K, Toma PH, Conlon DA, Feibush B, Moeder C, Bicker G, Grinberg N. Mechanistic aspects of chiral discrimination modified cellulose. Anal Chem 1997;69:1999–2007. [212] Kasat RB, Linda Wang N-H, Franses EI. Effects of backbone and side chain on the molecular environments of chiral cavities in polysaccharide-based biopolymers. Biomacromolecules 2007;8:1676–85. [213] Wang F, O’Brien T, Dowling T, Bicker G, Wyvratt J. Unusual effect of column temperature on chromatographic enantioseparation of dihydropyrimidinone acid and methyl ester on amylose chiral stationary phase. J Chromatogr A 2002;958:69–77.

References

[214] Pirkle WH, Pochapsky TC. Intermolecular 1H{1H} nuclear Overhauser effects in diastereomeric complexes: support for a chromatographically derived chiral recognition model. J Am Chem Soc 1986;108:5627–8. [215] Sardella R, Ianni F, Cossignani L, Aldini G, Carotti A. Binding modes identification through molecular dynamic simulations: a case study with carnosine enantiomers and the teicoplanin A2-2-based chiral stationary phase. J Sep Sci 2020;43:1728–36. [216] Ianni F, Cerra B, Shandiz ST, Di Michele A, Saluti G, Galarini R, Gioiello A, Sardella R, Carotti A. Integrating experimental and computational techniques to study chromatographic enantioresolutions of chiral tetrahydroindazole derivatives. J Chromatogr A 2020;1625, 461310. [217] Peluso P, Dessı` A, Dallocchio R, Sechi B, Gatti C, Chankvetadze B, Mamane V, Weiss R, Pale P, Aubert E, et al. Enantioseparation of 5,5’-dibromo-2,2’-dichloro-3-Selanyl4,4’-bipyridines on polysaccharide-based chiral stationary phases: exploring chalcogen bonds in liquid-phase chromatography. Molecules 2021;26:221. [218] Sardella R, Camaioni E, Macchiarulo A, Gioiello A, Marinozzi M, Carotti A. Computational studies in enantioselective liquid chromatography: forty years of evolution in docking- and molecular dynamics-based simulations. TrAC Trends Anal Chem 2020;122, 115703. [219] Peluso P, Mamane V, Dallocchio R, Dessı` A, Cossu S. Noncovalent interactions in high-performance liquid chromatography enantioseparations on polysaccharide-based chiral selectors. J Chromatogr A 2020;1623, 461202. [220] De Gauquier P, Vanommeslaeghe K, Vander Heyden Y, Mangelings D. Modelling approaches for chiral chromatography on polysaccharide-based and macrocyclic antibiotic chiral selectors: a review. Anal Chim Acta 2022;1198, 338861. [221] Scriba GKE. Chiral recognition in separation sciences. Part I: polysaccharide and cyclodextrin Selectors. TrAC Trends Anal Chem 2019;120, 115639. [222] Scriba GKE. Chiral recognition in separation sciences. Part II: macrocyclic glycopeptide, donor-acceptor, ion-exchange, ligand-exchange and micellar selectors. TrAC Trends Anal Chem 2019;119, 115628. [223] Yashima E, Yamada M, Kaida Y, Okamoto Y. Computational studies on chiral discrimination mechanism of cellulose trisphenylcarbamate. J Chromatogr A 1995;694: 347–54. [224] Booth TD, Wainer IW. Investigation of the enantioselective separations of αalkylarylcarboxylic acids on an amylose tris(3,5-dimethylphenylcarbamate) Chiral stationary phase using quantitative structure-enantioselective retention relationships. Identification of a conformationally driven chiral recognition mechanism. J Chromatogr A 1996;737:157–69. [225] Yamamoto C, Yashima E, Okamoto Y. Computational studies on chiral discrimination mechanism of phenylcarbamate derivatives of cellulose. Bull Chem Soc Jpn 1999;72:1815–25. [226] Ye YK, Bai S, Vyas S, Wirth MJ. NMR and computational studies of chiral discrimination by amylose tris(3,5-dimethylphenylcarbamate). J Phys Chem B 2007;111:1189–98. [227] Li Y, Liu D, Wang P, Zhou Z. Computational study of enantioseparation by amylose tris(3,5-dimethylphenyl-carbamate)-based chiral stationary phase. J Sep Sci 2010;33:3245–55.

437

438

CHAPTER 16 Fundamentals of enantioselective liquid chromatography

[228] Kim B-H, Lee SU, Moon DC. Chiral recognition of nphthaloyl, N-tetrachlorophthaloyl, and N-naphthaloyl α-amino acids and their esters on polysaccharide-derived chiral stationary phases. Chirality 2012;24:1037–46. [229] Alcaro S, Bolasco A, Cirilli R, Ferretti R, Fioravanti R, Ortuso F. Computer-aided molecular design of asymmetric pyrazole derivatives with exceptional enantioselective recognition toward the chiralcel OJ-H stationary phase. J Chem Inf Model 2012;52:649–54. [230] Ortuso F, Alcaro S, Menta S, Fioravanti R, Cirilli R. A chromatographic and computational study on the driving force operating in the exceptionally large enantioseparation of nthiocarbamoyl-3-(40 -biphenyl)-5-phenyl-4,5-dihydro-(1H)Pyrazoleon a 4methylbenzoate cellulose-based chiral stationary phase. J Chromatogr A 2014;1324:71–7. [231] Hu G, Huang M, Luo C, Wang Q, Zou J-W. Interactions between pyrazole derived enantiomers and chiralcel OJ: prediction of enantiomer absolute configurations and elution order by molecular dynamics simulation. J Mol Graph Model 2016;66:123–32. [232] Zhao B, Oroskar PA, Wang X, House D, Oroskar A, Oroskar A, Jameson CJ, Murad S. The composition of the mobile phase affects the dynamic chiral recognition of drug molecules by the chiral stationary phase. Langmuir 2017;33:11246–56. [233] Dallocchio R, Dessı` A, Solinas M, Arras A, Cossu S, Aubert E, Mamane V, Peluso P. Halogen bond in high-performance liquid chromatography enantioseparations: description, features and modelling. J Chromatogr A 2018;1563:71–81. [234] Wang X, Jameson CJ, Murad S. Modeling enantiomeric separations as an interfacial process using amylose tris(3,5-dimethylphenylcarbamate) (ADMPC) polymers coated on amorphous silica. Langmuir 2020;36:1113–24. [235] Phyo Y, Cravo S, Palmeira A, Tiritan M, Kijjoa A, Pinto M, Fernandes C. Enantiomeric resolution and docking studies of chiral xanthonic derivatives on chirobiotic columns. Molecules 2018;23:142. [236] Ashtari M, Cann NM. The docking of chiral analytes on proline-based chiral stationary phases: a molecular dynamics study of selectivity. J Chromatogr A 2015;1409:89–107. [237] Franzini R, Pierini M, Mazzanti A, Iazzetti A, Ciogli A, Villani C. Molecular recognition of the HPLC whelk-O1 selector towards the conformational enantiomers of nevirapine and oxcarbazepine. Int J Mol Sci 2021;22:144. [238] Ianni F, Carotti A, Marinozzi M, Marcelli G, Di Michele A, Sardella R, Lindner W, Natalini B. Diastereo- and enantioseparation of a Nα-Boc amino acid with a zwitterionic quinine-based stationary phase: focus on the stereorecognition mechanism. Anal Chim Acta 2015;885:174–82. [239] Ianni F, Pucciarini L, Carotti A, Gioiello A, Galarini R, Natalini S, Sardella R, Lindner W, Natalini B. Improved chromatographic diastereoresolution of cyclopropyl dafachronic acid derivatives using chiral anion exchangers. J Chromatogr A 2018;1557:20–7. [240] Jameson CJ, Wang X, Murad S. Molecular dynamics simulations of enantiomeric separations as an interfacial process in HPLC. AIChE J 2021;67, e17143. [241] Silva CF, Guimaraes L, Borges KB, Nascimento Jr CS. Development and validation of an experimental and theoretical method for the chiral discrimination of dinotefuran. Chirality 2020;32:53–63. [242] Silva CF, do Nascimento TA, Guimaraes L, Borges KB, Nascimento Jr CS. Elucidation of the chromatographic enantiomer elution order for praziquantel: an experimental and theoretical assessment. Chirality 2020;32:353–8.

References

[243] Li M, Zhang B, Yu J, Wang J, Guo X. Enantiomeric separation and simulation study of eight anticholinergic drugs on an immobilized polysaccharide-based chiral stationary phase by HPLC. New J Chem 2018;42:11724–31. [244] Barfeii H, Garkani-Nejad Z, Saheb V. Investigation of the mechanism of enantioseparation of some drug compounds by considering the Mobile phase in HPLC by molecular dynamic simulation. J Mol Mod 2019;25:297. [245] Ali I, Sekkoum K, Belboukhari N, Rebizi MN, Zaid MEA, Yusuf K, Alothman AA, AlJumahd BA, Ouladsmane M. Determination of enantio-separation, absolute configuration and chiral recognition mechanism of ofloxacin and flumequine by HPLC and modeling studies. J Chem Technol Biotechnol 2021;96:2901–8. [246] Liu Y, Wang X, Yu J, Guo X. Chiral separation and molecular simulation study of six antihistamine agents on a coated cellulose tri-(3,5-dimethylphenycarbamate) column (chiralcel OD-RH)and its recognition mechanisms. Electrophoresis 2021;42:1461–72. [247] West C, Guenegou G, Zhang Y, Morin-Allory L. Insights into chiral recognition mechanisms in supercritical fluid chromatography. II. Factors contributing to enantiomer separation on tris-(3,5-dimethylphenylcarbamate) of amylose and cellulose stationary phases. J Chromatogr A 2011;1218:2033–57. [248] West C, Zhang Y, Morin-Allory L. Insights into chiral recognition mechanisms in supercritical fluid chromatography. I. Non-enantiospecific interactions contributing to the retention on tris-(3,5-dimethylphenylcarbamate) amylose and cellulose stationary phases. J Chromatogr A 2011;1218:2019–32. [249] Khater S, Zhang Y, West C. Insights into chiral recognition mechanism in supercritical fluid chromatography III. Non-halogenated polysaccharide stationary phases. J Chromatogr A 2014;1363:278–93. [250] Khater S, Zhang Y, West C. Insights into chiral recognition mechanism in supercritical fluid chromatography IV. Chlorinated polysaccharide stationary phases. J Chromatogr A 2014;1363:294–310. [251] Khater S, West C. Characterization of three macrocyclic glycopeptide stationary phases in supercritical fluid chromatography. J Chromatogr A 2019;1604, 460485. [252] Raimbault A, Ma CMA, Ferri M, B€aurer S, BonnetP BS, L€ammerhofer M, West C. Cinchona-based zwitterionic stationary phases: exploring retention and enantioseparation mechanisms in supercritical fluid chromatography with a fragmentation approach. J Chromatogr A 2020;1612, 460689. [253] West C, Khater S. Characterization of retention and separation mechanisms with Pirkle-type enantioselective stationary phases in supercritical fluid chromatography. J Chromatogr A 2020;1626, 461352.

439

CHAPTER

Hydrophobic interaction chromatography

17

Deepika Sarin, Srishti Joshi, and Anurag S. Rathore Department of Chemical Engineering, IIT Delhi, New Delhi, India

Abbreviations 2D-LC ADC AEX AF APCI BisAb BPC CE CEX DAR ELSD FLD GC HCA HCIC HIC HILIC HPLC IEC LALLS LC LDA LSC mAb MS PDA RID RP-HPLC scFv SEC

two-dimensional liquid chromatography antibody drug conjugates anion-exchange chromatography affinity chromatography atmospheric pressure chemical ionization bispecific antibody bonded-phase chromatography capillary electrophoresis cation-exchange chromatography drug antibody ratio evaporative light scattering detector fluorescence detector gas chromatography hydrophobic contact area hydrophobic charge induction chromatography hydrophobic interaction chromatography hydrophilic interaction chromatography high-performance liquid chromatography ion-exchange chromatography low-angle laser light scattering detection liquid chromatography linear photodiode array liquid–solid chromatography monoclonal antibody mass spectrometry photodiode array detector refractive index detector reverse-phase high-performance liquid chromatography single-chain variable fragments size-exclusion chromatography

Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00026-6 Copyright # 2023 Elsevier Inc. All rights reserved.

441

442

CHAPTER 17 Hydrophobic interaction chromatography

17.1 Introduction Though it has been more than a century since the first use of chromatography, it continues to be the primary workhorse of analytical chemistry, particularly when it comes to the separation of proteins. The principle of chromatography involves separating components of a mixture by passing it through a stable solid or fluid phase and subsequently detecting the resolved components with the help of a detector in the form of a chromatogram. The stable solid or fluid phase where solute separation occurs is known as the “stationary phase.” The movement of the mixture and its components in a chromatographic system is aided by a “mobile phase”. Eventually, the solute components are detected, identified, and/or quantified by a detector. Depending upon the mobile phase, chromatography is segregated into gas chromatography (GC) and liquid chromatography (LC) [1]. Column chromatography and planar chromatography are two of the major formats of LC. Column chromatography is relatively more popular wherein the stationary phase of desired resin chemistry is tightly packed inside a column ensuring adequate ligand density and pore size for specific types of separations. The mobile phase is then passed through this packed column to achieve the required separation, either through interaction with the stationary phase or through differences in migration time due to size. Traditionally, columns with large particle sizes (up to 100 μm) were operated at slow flow rates (0.02 cm/s). However, modern formats such as high-performance liquid chromatography (HPLC) facilitate rapid separation, identification, and quantification of different components in a mixture by utilizing small packing particles and columns that can operate under relatively high pressure [2]. Aided by the advancement in column resin and ligand chemistry along with new column packing procedures, most contemporary LC separations have now been adapted to the HPLC format. This includes liquid–solid chromatography (LSC), bonded-phase chromatography (BPC), ionexchange chromatography (IEC), size-exclusion chromatography (SEC), and affinity chromatography (AF) [3]. Hydrophobic interaction chromatography (HIC) is an adsorption-based chromatography under BPC that separates solutes in a mixture based on hydrophobic interactions between solutes in the mobile phase and the hydrophobic surface of the stationary phase [4]. As illustrated in Figure 17.1, the operating principle of HIC is based upon the binding of the hydrophobic surface of a solute (protein in this case) to the hydrophobic ligands of the stationary phase at high salt concentrations of the mobile phase. The elution of protein molecules occurs with decreasing salt concentration, due to weakening hydrophobic interactions between the protein surface and the stationary phase. The more hydrophobic the protein surface, the stronger it binds to the ligands, requiring progressively low salt concentrations for elution [5]. In HIC, the binding and elution profile of a molecule is dictated by the hydrophobicity of the solute and the sorbent [6]. Hence, for different types of solutes and sorbents, the hydrophobicity scale is used to determine the elution order. The peak

17.1 Introduction

FIGURE 17.1 Illustration of the principle of hydrophobic interaction chromatography (HIC). Brown circle indicates the ligand, pink circle indicates the less hydrophobic region of the protein (pink + green), red circle indicates the more hydrophobic region of the protein, and blue circles indicate water molecules.

capacity and efficiency of a HIC process are also primarily determined by the hydrophobicity of a solute molecule and the degree to which it interacts with the non-polar stationary phase [7]. Due to its non-denaturing format, HIC is used both as a preparative and an analytical technique. Common HIC applications include the separation of a variety of solutes such as protein mixtures with varying hydrophobicity, DNA mixtures, enzymes, and small molecules. Due to its non-denaturing mode of operation, HIC also finds application in purification-related bioprocess operations.

17.1.1 Historical perspective HIC was first demonstrated by Arne Tiselius as “salting-out chromatography” [5] in the year 1948. After its initial demonstration, HIC resurfaced after almost two decades when the adsorption of lipophilic proteins was performed using hydrophobic-charged resins. Hjerten demonstrated the salt-mediated separation of proteins using the HIC principle in 1973. He also coined the term “hydrophobic interaction chromatography” [8]. Subsequently, throughout the 1970s and 1980s, a variety of salt and mobile phase compositions and resins (both charged/uncharged) were continuously developed to be used for different applications in HIC. In the 1990s, volatile buffers compatible with the HIC technique were explored [9] that eventually

443

444

CHAPTER 17 Hydrophobic interaction chromatography

FIGURE 17.2 Hydrophobic interaction chromatography (HIC) methods used for identification/ characterization. LC denotes liquid chromatography, 2DLC-MS denotes two-dimensional liquid chromatography-mass spectrometry, HIC-CE denotes hydrophobic interaction chromatography-capillary electrophoresis, and HIC-MS denotes hydrophobic interaction chromatography-mass spectrometry.

expanded the applicability of HIC LC coupled with mass spectrometry (MS). The use of volatile buffers also enabled the coupling of orthogonal techniques for hyphenated separations such as HIC-capillary electrophoresis (CE) [10] and HIC in twodimensional liquid chromatography (2D-LC) with MS [11,12]. Today, HIC methods are used to serve a variety of objectives including characterization, identification, and quantification of solutes in chemical and biomedical fields (Figure 17.2).

17.2 Operating principles of HIC Chromatographic techniques involve a specific type of solute-sorbent interactions, usually referred to as retention mechanisms of the solute in a column. In HIC, solute retention in a column is based on the hydrophobic environment under high and low salt concentrations. Though hydrophobicity is widely accepted as a strong noncovalent force for solute-sorbent interaction, the rationale for such interactions is still not well understood. Hence, multiple theories explain the retention of solutes in HIC. The most popular ones are described later. One of the first attempts describing the binding mechanisms of solute molecules in HIC comprised coalition of the condensation theory based on electrostatic interactions [13] and the solvophobic theory for salting-out of proteins [14,15]. The solvophobic theory explains a two-step process for solute adsorption on a surface. The first step includes cavity formation by water for the solute on the surface, followed by a second step where the solute fills the cavity for adsorption. The corresponding total free energy change is the sum of individual contributions to free energy from cavity formation, solute-solvent interactions, solvent interactions, and electrostatic

17.2 Operating principles of HIC

interactions [7]. All individual free energy terms in solvophobic theory are defined except the electrostatic free energy. Manning [13] explained the electrostatic free energy term for a charged protein (solute) as a simple ion at low salt concentrations. In contrast, at high concentrations, the spacing between charged species decreases, leading to an increase in ionic bonding and hence more retention. The effect of salt composition on protein retention has also been explained by a similar theory that puts forward the salting-out effect as preferential hydration of protein solutes in HIC [16]. The theory suggests an increase in the hydrophobic surface area of proteins in the presence of salts due to the rise in free energy. The free energy increase is limited by intermolecular association, lowering the contact area between the hydrophobic surface and polar solvent. This explains the binding of proteins to the hydrophobic stationary phase in a salt solvent media (especially high salt concentration) to build an environment of minimum free energy change. This phenomenon can also result from a corresponding increase in the surface tension of water by sodium and phosphate salts due to the preferential hydration of proteins. The theory is widely accepted and valid for a range of solutions except for divalent cations with high protein solubility, despite the increment in proposed surface tension [17]. The interaction between two hydrophobic molecules in an aqueous environment has been explained thermodynamically as well as through an entropy-driven process [18–20]. The change in entropy (S), enthalpy (H), and free energy (G) for a hydrophobic interaction follows the following equation: ΔG ¼ ΔH  TΔS

(17.1)

A decrease in entropy occurs when a hydrophobic solute enters an aqueous solution due to an increase in the degree of order of water molecules surrounding the hydrophobic group. This process is thermodynamically unfavorable (Eq. 17.1) and does not occur spontaneously because the enthalpy change is negligible compared to entropy change producing a positive free energy change. Hence, if hydrophobic compounds like that from a stationary phase are inserted into the aqueous environment, there is a spontaneous association between the hydrophobic solute and the non-polar stationary phase. This results in a shift of ordered water molecules around the hydrophobic solute, increasing the entropy and implying a negative free energy change, thereby making the process thermodynamically favorable [21]. Besides the proposed retention mechanisms, HIC includes several process parameters too that define the solute-solvent interactions. Such parameters are important to consider as they not only affect the retention of the solute but also its separation and selectivity in the aqueous hydrophobic environment.

17.2.1 Operation of HIC HIC can be performed both as a preparative and an analytical separation technique. The general process flow for HIC involves the basic pattern in any chromatography of bind and elute using different buffers and sorbent properties. Here, we explain the HIC analytical process flow performed using an HPLC (Figure 17.3). An HPLC

445

446

CHAPTER 17 Hydrophobic interaction chromatography

FIGURE 17.3 Process parameters for hydrophobic interaction chromatography (HIC) in order relating to different HPLC modules where 1 refers to buffer reservoir, 2 to pump module, 3 to sampler, 4 to column compartment, and 5 to detector. UV denotes ultraviolet detection, FLD denotes florescence detection, and ADC denotes antibody-drug conjugates.

consists of five significant modules and the parameters for each depend on the type of separation taking place.

17.2.1.1 Mobile phase HIC process starts from module 1 of an HPLC known as the buffer reservoir (Figure 17.3). As the name suggests, a buffer reservoir is a mobile phase required for binding and elution in the chromatography process. The choice of mobile phase for HIC depends on two important parameters: salt composition and the required pH of the buffer for proper binding and elution to occur. The effect of salt ions on solutes was first noticed by Hofmeister [22]. He divided the salt ions into two types: saltingout or kosmotropic and salting-in or chaotropic. The salting-out ions are known to cause conformational changes in a biomolecule (solute) leading to enhanced hydrophobic interactions by stabilizing the hydrophobic core of the biomolecule [23]. At a high concentration of the kosmotropic salts, protein solubility in water decreases due to fewer water molecules being available in bulk. The decrease in solubility leads to increased surface tension and stability of proteins, favoring hydrophobic interactions. Examples of kosmotropic salts include sulfate, acetate, and citrate ion salts. In contrast, the chaotropic salts are structure-breaking salts that cause increased protein solubility, decreased surface tension, and unfavorable hydrophobic interactions [24]. Solubilization of proteins takes place under the influence of chaotropic salts containing magnesium, calcium, and barium. Table 17.1 summarizes the most commonly used non-volatile salts in HIC [25]. It is evident from the table that as per the

17.2 Operating principles of HIC

Table 17.1 Salt characteristics of the mobile phase used in hydrophobic interaction chromatography (HIC). Salt

Characteristics

Ammonium sulfate Sodium acetate

Most commonly used in HIC

Sodium citrate Sodium sulfate

Has selective effects on proteins. High retention for hydrophobic proteins but decreased retention for hydrophilic proteins Can be used but has preparation issues in multiple salt solutions Exhibits effects that deviate from the solvophobic theory

Hofmeister series, sulfate, acetate, and citrate-based salts are essential for salting-out of proteins in the presence of a hydrophobic stationary phase. Sodium sulfate has a high salting-out capability but has low solubility in water which depends on the temperature; hence, when high salt concentrations are required, sodium sulfate can be replaced with ammonium sulfate [26]. The pH of the buffer also plays a vital role in the adsorption of the solute to the sorbent in HIC. Changes in pH can have a significant impact on chromatographic separation, such as early elution, improper binding, or solute denaturation. In the case of proteins, it is observed that maximum hydrophobic interactions occur at the isoelectric point of the protein [27]. In general, an increase in pH leads to increased titration of charged groups, increasing hydrophilicity, and decreasing hydrophobic interactions [8]. In HIC, elution usually occurs by decreasing salt concentration, using chaotropic salts, or increasing the buffer pH. Alternatively, additives may also be used for modulating elution in HIC [20]. Detergents and organic modifiers decrease the surface tension and polarity of the eluent while maintaining solute stability. Organic solvents such as isopropanol and ethylene glycol are modifiers that compete with bound proteins for hydrophobic ligands on sorbents leading to displacement [28].

17.2.1.2 Elution gradient Following module 1, the buffer from the reservoir passes to the pump module of an HPLC system (Figure 17.3). It is through the pump system that proper flow of the mobile phase through the stationary phase is maintained and the required gradient is generated for solute elution. HIC elution can be performed by a linear gradient or step elution. The general rule for method development is to first apply the linear gradient to study the elution profile of a solute followed by optimization by step elution to improve resolution and reduce the time of the analysis [29]. Gradient elution in HIC involves a gradual, linear decrease of salt concentration over a defined volume. In case the solute does not elute after a linear gradient, alternative approaches like using organic solvents or alcohols, weaker lyotropic salts, or weaker

447

448

CHAPTER 17 Hydrophobic interaction chromatography

hydrophobic adsorbents need to be explored [30]. Therefore, the aim of a linear gradient method is to identify the correct salt concentration and stationary phase parameters to be then controlled using step elution to improve equipment requirements.

17.2.1.3 Solute properties Module 3 of an HPLC consists of the autosampler where the solute samples to be analyzed are placed as microliter solutions in vials (Figure 17.3). As mentioned earlier, HIC is used for the separation of several kinds of solutes such as proteins, nucleic acids, and enzymes. Proteins are extensively studied biomolecules that exhibit structural segregation at different stages of folding, starting from amino acids as a primary structure and through the protein structural hierarchy. The physicochemical property of proteins responsible for the chromatographic behavior in HIC is hydrophobicity. Theoretical protein hydrophobicity can be determined through consolidated measurements of individually exposed amino acid hydrophobicity in a protein primary sequence [31]. This weighted average of exposed hydrophobic amino acid residues in a protein is known as “average surface hydrophobicity” [32]. This property is used to calculate the retention time of protein solutes in HIC salt-based elution. Another model explaining the measurement of protein hydrophobicity is called “hydrophobic contact area (HCA)” that includes heterogeneous distribution of hydrophobic patches on the protein surface [31]. During interaction with a ligand, the solvent-exposed area of a protein occupied by hydrophobic residues is known as the hydrophobic accessible area. When HCA and hydrophobic accessible area exhibit similar values, it is assumed that the protein has a homogeneous surface hydrophobicity distribution whereas differing values of HCA and hydrophobic accessible area are due to a heterogeneous surface hydrophobicity distribution in protein, such that only hydrophobic patches consist of extensive hydrophobic residues that are linked to hydrophobic ligand interactions of the stationary phase [33].

17.2.1.4 Stationary phase Moving toward, the column compartment (Module 4) or the stationary phase of an HPLC, it is imperative to note that the choice of compatible stationary phase in HIC depends not only on the chromatographic conditions required but also on the hydrophobicity of the solute so as to avoid its precipitation while maintaining proper separation environment [34] (Figure 17.3). Base matrix and ligand characteristics are the two main requirements to provide different degrees of hydrophobicity and solutebinding capacity in HIC. The base matrix can either be based on natural polymers like cellulose and agarose or can be derived from synthetic polymers or inorganic materials like silica. Since matrices do not contribute to the hydrophobic nature of an HIC adsorbent, the hydrophilic base matrices that yield higher efficiency have always been preferred [35]. Today, agarose is the most commonly used base matrix in HIC due to the range of hydrophobic ligands that can be attached to it [20]. The hydrophobicity of an HIC stationary phase depends mostly on the ligand attached to the base matrix and the degree of substitution. Since the birth of HIC

17.2 Operating principles of HIC

as a salting-out chromatography through non-biospecific adsorption, many spacer arms have been developed. Spacer arms are carbon chains covalently bonded to the solid matrix and sites for functional group attachment. In affinity chromatography, spacer arms orient the functional group so that it is easily available to the ligand and not sterically hindered by the matrix. Spacer arms can be both charged and uncharged in nature, and when the former is used, additional electrostatic interactions are observed [36]. In general, linear chain alkenes (butyl and octyl) or aromatic groups such as phenyl are used as ligands in HIC. It is important to note here that as the n-alkyl chain length increases at a constant degree of substitution, the hydrophobicity of the adsorbent increases, though there might be a decrease in the adsorption specificity [37]. Hence, to balance out hydrophobicity and selectivity, for strong hydrophobic solute molecules, small alkyl chain lengths are used for adsorption, and higher alkyl chain lengths for weak hydrophobic solutes. Figure 17.4 summarizes different HIC ligands commonly used with increasing hydrophobicity (bottom to top). As the hydrophobicity increases, the resolution and selectivity of proteins increase but only up to a limit (phenyl) after which a decrease in selectivity is observed (in the case of butyl and hexyl). It should be noted here that certain ligands in HIC are too hydrophobic and may lead to irreversible adsorption and subsequent denaturation of solute molecules. Such sorbents are now replaced by intermediate hydrophobic polymeric ligands [38]. Substitution of strong hydrophobic ligands with intermediate ones is an important aspect of HIC that prevents the denaturation of solutes, a distinguishing characteristic between HIC and reverse-phase chromatography (RP-HPLC). RP-HPLC is a type of column liquid chromatography that is similar to HIC as it too exploits the hydrophobic interaction of solute and sorbent to achieve separation. However, RP-HPLC suffers from a major drawback of

FIGURE 17.4 Ligand characteristics of the stationary phases used in hydrophobic interaction chromatography (HIC) and their effect on the elution profile of certain common proteins. 1 refers to ribonuclease A, 2 to lysozyme, and 3 to α-chymotrypsinogen.

449

450

CHAPTER 17 Hydrophobic interaction chromatography

Table 17.2 A comparison between reverse-phase liquid chromatography (RP-HPLC) and hydrophobic interaction chromatography (HIC). S. no. 1 2

3

4

5

6

7

RP-HPLC

HIC

Stationary phase is strongly hydrophobic Risk of protein denaturation due to a highly hydrophobic environment

Stationary phase is comparatively less hydrophobic Less risk of protein denaturation in the optimum hydrophobic environment of HIC Stationary phases may show different effects depending on the type of support and ligand

Stationary phases show only minor differences as regards the retention and selectivity of separation for various solutes The selection of stationary and mobile phases is standardized with the ones frequently used Several hundred mmol/ml gel of C4–C18 alkyl ligands usually used for RPC adsorbents Protein binding to RPC adsorbents is usually very strong, which requires the use of non-polar solvents for their elution Generally, retention decreases with increasing temperature so that the retention enthalpy is negative

The selection of stationary and mobile phases depends on a number of factors and is not standardized The degree of substitution of HIC adsorbents is usually in the range of 10–50 mmoL/mL gel of C2–C8 alkyl ligands The polarity of the complete system of HIC is increased by decreased ligand density on the stationary phase and by adding salt to the mobile phase The temperature dependence of retention observed in RPC is the opposite of that in HIC where the retention usually increases with temperature

denaturing the solute molecule due to the nature of strong stationary phases generally employed. Hence, HIC is preferred over RP-HPLC in utilizing moderately hydrophobic stationary phases and salt compositions as mobile phases to prevent solute denaturation. Table 17.2 lists the general differences between RP-HPLC and HIC. Lastly, critical hydrophobicity for adsorption of the solute on a stationary phase also depends on the degree of substitution. An increase in the degree of substitution leads to an increase in hydrophobicity and higher binding capacities but consequently requires harsher elution conditions that might result in solute denaturation [20]. The degree of substitution for charged hydrophobic groups can be determined using a dye carrying hydrophobic groups with a negative charge bound irreversibly to a ligand. On saturation of the dye, under certain conditions, there will always be a certain amount of dye bound even after extensive washing giving an estimate of the degree of substitution [39]. Sorbents with the highest degree of substitution are known to be the least stable. In a study describing HIC as a polishing step for downstream purification of monoclonal antibodies, standard HIC resins, new generation HIC resins, and hydrophobic charge induction chromatographic (HCIC) resins were tested for separation,

17.2 Operating principles of HIC

selectivity, and pore size efficiency [40]. HCIC exhibited a reverse-salt effect where the resin-binding capacity was observed to decrease with increasing salt concentration due to pH and non-specific electrostatic interactions. Resin pore size optimization leads to a better rate of mass transfer in the case of HCIC resin. Further, in the case of new-generation HIC resins, the pore size was increased leading to a significant increase in dynamic binding capacity. In general, it was observed that HCIC resins show more selection at the binding stage and HIC resins provide more resolution during the elution stage. Hence, while choosing an appropriate stationary phase, it is imperative to consider resin properties, pore size, and solute characteristics. Apart from the base matrix, the degree of substitution, and ligand density of a stationary phase, the operational temperature of a stationary phase is also a crucial factor for HIC. It is evident from Eq. (17.1) that temperature has a positive effect on the adsorption of the solute on the hydrophobic sorbent. However, the temperature can also cause effects like a decrease in surface tension and an increase in the solvation of certain molecules leading to resistance in hydrophobic interactions [41]. Such instances have been observed for protein molecules that denature and lose their biological activity at high temperatures. In general, hydrophobic interactions increase with an increase in temperature only up to a limit of around 60°C; above which stability due to hydrogen-bonding electrostatic interactions decreases and disulfide bridges decrease leading to a drop in hydrophobic interactions [42]. Temperature effects of the HIC stationary phase on retention and selectivity have been extensively researched in parallel to thermodynamic models. One such theory includes a description of retention factor (k) under isocratic conditions in terms of reference temperatures TH and TS when standard enthalpy change (ΔH°) and standard entropy change (ΔS°) for solute transfer from the mobile phase to stationary phase are zero, respectively [43]: ln k ¼

  ΔCp ° T H T s   1 + ln∅ T T R

(17.2)

In Eq. (17.2), ΔCp° is the invariant heat-capacity change and ϕ is the phase-ratio (volume ratio of stationary phase to mobile phase in a column). To eliminate the dependence of temperature on ΔCp° at high temperatures [44], a simpler form of Eq. (17.2) was adopted for chromatographic retentions dependent on temperature. This simpler Eq. (17.3) is termed as the “quadratic equation”: ln k ¼ a +

b c + 2 + ln∅ T T

(17.3)

17.2.1.5 Detector Finally, the eluate from the column compartment reaches the final module of the HIC-HPLC process which is the detector (Figure 17.3). HPLC can be equipped with many different types of detectors such as photodiode array detector (PDA) for ultraviolet (UV) detection, refractive index detector (RID), fluorescence detector (FLD)

451

452

CHAPTER 17 Hydrophobic interaction chromatography

for the detection of emission and excitation in fluorescent or fluorescently labeled molecules, evaporative light-scattering detector (ELSD), and conductivity detector for cation/anion analysis. The choice of the detector to be used depends solely on the solute characteristics. For example, in the case of proteins, UV detection at 220 nm for peptides and 280 nm for aromatic proteins is preferred. Proteins rich in fluorescent aromatic amino acids like tryptophan and phenylalanine can also be detected using an FLD detector. Apart from this, certain solutes can be tagged fluorescently to be used with an FLD [45]. Small carbohydrate separation using HIC has been known to be compatible with ELSD and atmospheric pressure chemical ionization mass spectrometry (APCI-MS) yielding both quantitative and spectral results of analytes [46]. In another example, alkaline phosphatase was detected using a linear photodiode array (LDA) in combination with gradient elution HIC using low-angle laser light-scattering detection (HIC-LALLS) [47]. In certain cases, HIC coupled with MS can be a good form of detection where the elute from HPLC is directly injected into MS for a mass-to-charge ratio analysis.

17.3 Applications of HIC Different modes of liquid column chromatography exploit unique separation techniques and hence each has promising applications in chemical and biomedical fields. HIC too is a successful chromatographic technique with a wide range of applications including purification, protein folding-refolding, and solute characterization, and other prominent HIC applications are mentioned later.

17.3.1 Solute purification HIC is most commonly used as a part of downstream operations for the purification of solutes like monoclonal antibodies (mAbs), recombinant proteins, nucleic acids, enzymes, viruses, plant proteins, and PEGylated proteins. The first attempt to provide a method for the purification of human IgA by salt-mediated hydrophobic interaction chromatography was performed by Doellgast and Plaut [48]. This was followed by purification of IgGs in the presence of ammonium sulfate in 1977 [49]. Purification of monoclonal antibodies was initially done using ammonium sulfate precipitation, but with the development of different modes of column chromatography, the purification strategies started shifting toward modern tools such as AEX and HIC [50]. Today, preparative HIC is quite popular as a purification step in the downstream processing of mAbs, especially in the clearance of aggregates [51]. Earlier, both CEX and HIC steps were carried out sequentially for aggregate removal in mAbs as polishing steps [52]. Recently, a novel technique was applied wherein elution from the CEX step was directly loaded into an HIC column operating in a flow-through mode, hence forming a two-stage chromatography process [53]. This type of purification proved to be faster and highly efficient with complete removal of mAb aggregates. Additionally, optimization of purification processes

17.3 Applications of HIC

under process analytical technology leads to the production of therapeutics with improved quality standards [54]. In a similar approach, mechanistic modeling for HIC as a part of process analytical technology has been applied to effectively predict and clear aggregates in mAbs to limit batch-to-batch variations [55]. The proposed pooling strategy proved to be successful in delivering consistent results compared to the traditional techniques and has further applications in scale-up and online process chromatography [56]. The non-denaturing and mild hydrophobic environment in HIC has increased its popularity to be used in the purification of recombinant protein molecules such as human growth hormone [57] and recombinant streptokinase [58]. One of the examples exemplifying the prevalence of HIC as a large-scale downstream purification strategy is in the purification of recombinant granulocyte colony-stimulating factor (GCSF) hormone that forms a high number of insoluble aggregates. HIC has been employed in the refolding step to obtain a purity of close to 95% for the drug substance [59]. It has also been prolifically used for nucleic acid purification. The first attempt at using HIC as a plasmid purification step was done by Diogo and Queiroz [60] for the treatment of cystic fibrosis. The separation occurs due to the more hydrophobic character of single-stranded DNA and denatured plasmid from the native plasmid DNA [61] and in the presence of lipid A that promotes interaction with an HIC stationary phase [62]. A similar principle was applied for the scale-up production of plasmids used in vaccines against rabies [63]. Comparative interactions of supercoiled plasmid DNA and bacterial genomic DNA with hydrophobic supports have been explored by many researchers [64]. Enzymes are proteins with catalytic activity, and therefore, enzymes too have hydrophobic patches that can serve for hydrophobicity-based purifications. For example, in the case of labile rat liver mitochondrial enzymes, HIC purification proves to be an easier method when compared to longer and laborios alternatives that often lead to loss of enzyme activity [65]. Restriction endonucleases are enzymes that are employed in recombinant technology but are difficult to purify. HIC in combination with precipitation has been proven to ease endonuclease purification, enabling the isolation of novel restriction enzymes with superior activity [66]. HIC also aids in plant enzyme purification which is usually a daunting task as it involves the extraction of such enzymes from small fractions of total protein extracts. This in turn has benefited the manufacturing of pharmaceutical secondary products from plant cell cultures [67]. PEGylation is an important modification (drug conjugate) necessary for drug stability and enhanced activity of certain biologic products. PEGylation involves a bond formation between the drug of interest and polyethylene glycol [68] and changes the chemical and physical properties of a drug including its hydrophobicity. Therefore, PEGylated protein purification can be carried out through HIC depending on the increase or decrease of hydrophobicity of the native protein [69]. HIC has also been employed as a quantitative method for the analysis of small carbohydrates using parallel detectors [46]. Lastly, a major breakthrough in HIC application was achieved when it was used for the purification of influenza A and B viruses [70]. The influenza

453

454

CHAPTER 17 Hydrophobic interaction chromatography

virus is an enveloped virus with membrane proteins responsible for its high diversity and difficulty in purification [71]. Thus, HIC utilizes an interaction between membrane proteins and the hydrophobic sorbent in combination with the orthogonal AEX step to achieve successful downstream virus purification.

17.3.2 Protein refolding HIC has been recognized as a powerful tool to monitor and provide stable conditions for protein refolding as in the case of recombinant human interferon [72]. The monitoring of protein refolding is usually done through peak height and retention time that are considered characteristic features of conformational changes in a protein [73]. To determine the various factors affecting refolding, HIC can also be used in combination with SEC. Xindu and Xiaoqing [72] used HIC to refold proteins by removing the denaturing agent and maintaining a stable refolding environment of hydrophobicity, ionic strength, and viscosity by gradient elution. Conversely, the effects of protein unfolding in HIC depending on salt composition and temperature for stability and adsorption have been successfully established using a twoconformation adsorption model [74]. The model enables predicting retention trends to provide applications in HIC process development.

17.3.3 Solute characterization Aside from many purification-based applications, HIC has a multitude of applications in solute characterization which makes it a powerful chromatographic technique. One such application is in the characterization of biopharmaceuticals. HIC can be used for the characterization of small molecules [75], mAbs, antibody-drug conjugates (ADCs), and bispecific antibodies. Characterization of variations like oxidation and deamidation in biotherapeutics is of utmost importance as it directly impacts product quality [76] In an interesting study, HCIC has been employed to successfully remove closely related product variants such as methionine oxidation in GCSF based on multimodal chromatography using pH and salt-based elution [77]. Such multimodal chromatography methods lead to increased product recovery and faster protein purification [78]. HIC has been extensively used in the characterization of mAbs specifically for the characterization of Fc and Fab fragments of mAb after papain enzyme digestion [79]. Fab and Fc fragmentation of mAb leads to the formation of small peptides with hydrophobic properties that can be exploited for separation using a hydrophobic stationary phase. Moreover, analysis of intact mAb to enlist N-glycan heterogeneity can be done using HIC [45]. It has also been used for the separation and characterization of the mAb population after stressinduced terminal processing to form conformationally altered populations in a drug product [80]. Oxidation of methionine and unexposed tryptophan residues in mAbs is difficult to characterize, but with the help of HIC, many new methods have been explored that have eased the characterization of complex biopharmaceuticals such as mAbs [81].

17.3 Applications of HIC

ADCs are a form of conjugated biopharmaceutical drugs that target tumor cells and treat cancer. These are complex drugs (bioconjugates) that consist of an antibody linked to an anti-cancer cytotoxic payload. Drug-to-antibody ratio (DAR) is a measure of the number of drugs per molecule which is different for every ADC. HIC is considered a reference technique for estimating and segregating ADCs based on DAR [82]. The major reason for HIC to be preferred for ADC analysis is its nondenaturing nature and ease of fraction collection [83]. High-resolution HIC has been known to separate DAR positional isomers [84] since the number of payloads attached is directly proportional to hydrophobicity and hence retention time. Increasing hydrophobic payloads have a greater retention time, and if a payload is exposed toward amino acids on the antibody, its interaction with the sorbent matrix increases with longer retention times. Lastly, in terms of ADC analysis, HIC is a quicker technique with no dependency on organic solvents [85]. Bispecific antibodies (BisAb) and single-chain variable fragments (scFv) are another class of biotherapeutics currently being studied with the help of HIC. This is carried out by analyzing the number of peaks eluting with and without the presence of scFv in an IgG [86]. HIC-generated profile is an important indication of protein solubility and therefore can be used to give a qualitative estimate of the highest relative solubility of proteins in reference to the Hofmeister series. It is an index of optimum concentrations of stabilizing additives and parameters like temperature and pH. The tendency of a protein to self-associate depends on the minimum chemical potential and maximum protein solubility [87]. In simple terms, the measure of the height of an unretained peak in HIC is a measure of protein solubility. The taller the unretained peak, the lower the interaction between protein and column indicating protein solubility as the tendency of the protein to self-associate is lower [88].

17.3.4 HIC as a part of multi-dimensional separation platforms In its simplest form, HIC follows a one-dimensional process flow using aqueous saltbased buffers to achieve resolution. When used in a two-dimensional separation setup, both aqueous buffers and organic solvents/volatile salts can be used to attain either higher resolution or better characterization of the solute because of enhanced orthogonality and easy coupling with MS. Coupling of HIC with CE gives a remarkable resolution, generally not observed in the simple one-dimensional analysis. Lastly, HIC-MS is known to implement a top-down characterization for the identification of solutes and utilizes organic solvents or volatile salts. Two-dimensional liquid chromatography (2D-LC) utilizing HIC is emerging as an essential tool in characterizing lignosulfonates. Lignosulfonates are amphiphilic molecules and have great surface-active properties. They are by-products of the sulfite pulping industry, hence used as stabilizers, detergents, and surfactants [89]. The chemical nature and structure of the hydrophilic part of lignosulfonates have mostly been studied by SEC and ion pair RP-HPLC, but their hydrophobic nature has also been studied through 2D-LC analysis using HIC and SEC [90]. Finally, 2D-LC with

455

456

CHAPTER 17 Hydrophobic interaction chromatography

HIC and weak cation exchange chromatography have also been applied for online plasma fractionation of intact proteins to reach excellent retention, selectivity, and resolution [91].

17.4 Challenges and limitations Though a widely useful separation technique, HIC also suffers from some challenges. First, the separation efficiency in HIC suffers due to poor diffusion of large-molecular-weight proteins in highly viscous buffer solutions. Researchers have attempted to compensate for the loss in peak capacity by decreasing the particle size but that can only be done to a limit, and therefore, only some proteins can be separated in a single HIC analysis. Also, selectivity in terms of achieving different separations under different elution conditions is not easy in HIC [92]. Second, in the case of native cysteine conjugate ADCs, HIC is limited due to low kinetic efficiency in comparison to RP-HPLC, including in the determination of average DAR [93]. Third, the requirement of high salt concentrations as a mobile phase for binding is a major challenge in HIC, especially when coupling with MS. Though many volatile buffer systems compatible with MS have been developed over the years, the separation efficiency is still proven to be better for organic solvent-dependent RP-HPLC. Fourth, in the case of downstream purification of mAbs where HIC is used as a polishing step, there exist certain challenges such as the lower binding capacity of HIC resins, lower yield, and high amount of salt in HIC eluate, leading to difficulties in large-scale process transitions during purification [40]. In recent years, novel pore size-optimized resins for HIC have been developed to facilitate easy mass transfer and overcome downstream purification limitations using HIC. Finally, method development of HIC requires resolving and optimization of many parameters like salt concentration adjustment relative to the sorbent being used, pH, and solute characteristics which under certain circumstances can be timeconsuming.

17.5 Conclusions HIC is a reliable separation technique with diverse chromatographic applications that still need exploration. The last century has witnessed significant advancements in the development of novel stationary phase matrices for HIC with varying resin properties and pore size. Figure 17.5 depicts the timeline of HIC evolution from a simple salting-out process to a major chromatographic technique with a diverse set of applications. HIC utilizes a high amount of salts for binding and elution which often have storage and handling problems in addition to cost like in the case of ammonium sulfate [94]. In addition, the destruction of high salt solutions during waste disposal is a

17.5 Conclusions

FIGURE 17.5 Timeline of development and application of hydrophobic interaction chromatography (HIC).

major operation and cost limitation in HIC. Therefore, the use of HIC at times may be hindered by such process economic concerns. To alleviate this, researchers have attempted to optimize HIC methods with low-salt elution conditions. Often, researchers prefer RP-HPLC over HIC due to the ease of method development in the former and improved resolution. The beginning of the 21st century witnessed great orthogonal analytical developments such as two-dimensional chromatography and multi-dimensional/multiattribute analysis. HIC too has been optimized to enable coupling with other chromatographic techniques such as in 2D-LC and with CE and MS. However, a continuous coupling of HIC with LC-CE-MS is yet to be developed and will extend the utilization of HIC as an analytical characterization tool. Further, continuous bioprocessing has also recently been gaining much attention, and with the development of cost-efficient tools, HIC is likely to find further application as a part of the continuous bioprocessing platform.

Acknowledgments This work was funded by the Center of Excellence for Biopharmaceutical Technology grant under the Department of Biotechnology, Government of India (BT/COE/34/SP15097/2015). The authors would also like to thank Agilent Technologies India Private Limited for funding this research as part of its corporate social responsibility initiative.

457

458

CHAPTER 17 Hydrophobic interaction chromatography

Conflict of interest The authors declare that they have no financial or personal relationships with other people or organizations that can inappropriately influence our work and there is no conflict of interest regarding the publication of this paper.

References [1] Ettre LS, Sakodynskii KI. M.S. Tswett and the discovery of chromatography II: Completion of the development of chromatography (1903–1910). Chromatographia 1993;35(5):329–38. https://doi.org/10.1007/BF02277520. [2] Rathore AS, Joshi S. Liquid chromatography jhistorical development. In: Encyclopedia of analytical science; 2019. p. 86–92. https://doi.org/10.1016/B978-0-12-4095472.14535-4. [3] Currie BL. Chromatography. Concepts and contrasts. 2nd edition by James M. Miller. Wiley-Interscience, Hoboken, NJ. 2005. Xiv + 490 pp. 16  24 cm. ISBN 047147207-7. $94.50. J Med Chem 2005;48(10):3660. https://doi.org/10.1021/JM0582120. [4] Tomaz CT, Queiroz JA. Hydrophobic interaction chromatography. Liq Chromatogr Fundam Instrum 2013;121–41. https://doi.org/10.1016/B978-0-12-415807-8.00006-7. [5] Tiselius A. Adsorption separation by salting out. Arkiv for Kemi, Mineralogi Och Geologi, B 1948;26(1). [6] Wingfield P. Protein precipitation using ammonium sulfate. Curr Protoc Protein Sci 1998;13(1):A.3F.1–8. https://doi.org/10.1002/0471140864.PSA03FS13. [7] Perkins TW, Mak DS, Root TW, Lightfoot EN. Protein retention in hydrophobic interaction chromatography: modeling variation with buffer ionic strength and column hydrophobicity. J Chromatogr A 1997;766(1–2):1–14. https://doi.org/10.1016/S00219673(96)00978-8. [8] Hjerten S. Some general aspects of hydrophobic interaction chromatography. J Chromatogr A 1973;87(2):325–31. https://doi.org/10.1016/S0021-9673(01)91733-9. [9] Nakamura H, Konishi T, Kamada M. Use of volatile buffers in high performance hydrophobic interaction chromatography of proteins. Anal Sci 1990;6(1):137–8. https://doi. org/10.2116/analsci.6.137. [10] Liang H, Wang Y, Wang Q, et al. Hydrophobic interaction chromatography and capillary zone electrophoresis to explore the correlation between the isoenzymes of salivary α-amylase and dental caries. J Chromatogr B: Biomedical Sciences and Applications 1999;724(2):381–8. https://doi.org/10.1016/S0378-4347(98)00522-2. [11] Sarrut M, Corgier A, Fekete S, Guillarme D, Lascoux D, Janin-Bussat MC, et al. Analysis of antibody-drug conjugates by comprehensive on-line two-dimensional hydrophobic interaction chromatography x reversed phase liquid chromatography hyphenated to high resolution mass spectrometry. I  Optimization of separation conditions. J Chromatogr B 2016;1032:103–11. https://doi.org/10.1016/J.JCHROMB.2016.06.048. [12] Valeja SG, Xiu L, Gregorich ZR, Guner H, Jin S, Ge Y. Three dimensional liquid chromatography coupling ion exchange chromatography/hydrophobic interaction chromatography/reverse phase chromatography for effective protein separation in top-down proteomics. Anal Chem 2015;87(10):5363–71. https://doi.org/10.1021/ACS.ANALCHEM.5B00657/SUPPL_FILE/AC5B00657_SI_001.PDF.

References

[13] Manning GS. The molecular theory of polyelectrolyte solutions with applications to the electrostatic properties of polynucleotides. Q Rev Biophys 1978;11(2):179–246. https:// doi.org/10.1017/S0033583500002031. [14] Lienqueo ME, Mahn A, Salgado JC, Asenjo JA. Current insights on protein behaviour in hydrophobic interaction chromatography. J Chromatogr B Analyt Technol Biomed Life Sci 2007;849(1–2):53–68. https://doi.org/10.1016/J.JCHROMB.2006.11.019. [15] Sinanoglu O. The solvophobic theory for the prediction of molecular conformations and biopolymer bindings in solutions with recent direct experimental tests. Int J Quantum Chem 1980;18(2):381–92. https://doi.org/10.1002/QUA.560180207. [16] Lee JC, Lee LL. Preferential solvent interactions between proteins and polyethylene glycols. J Biol Chem 1981;256(2):625–31. https://doi.org/10.1016/S0021-9258(19) 70019-2. [17] Arakawa T, Timasheff SN. Mechanism of protein salting in and salting out by divalent cation salts: balance between hydration and salt binding. Biochemistry 2002;23 (25):5912–23. https://doi.org/10.1021/BI00320A004. [18] H€ubner G. Sh. Lewin, Displacement of Water and its Control of Biochemical Reactions. XIII, 367 S., 75 Abb. (z. T. farbig), 10 Tab. London-New-York 1974: Academic Press, £ 9.30. Z Allg Mikrobiol 1975;15(8):655. https://doi.org/10.1002/JOBM.19750150818. [19] Hjerten S. Hydrophobic interaction chromatography of proteins on neutral adsorbents. Methods Protein Sep 1976;233–43. https://doi.org/10.1007/978-1-4684-9984-1_6. [20] Ochoa JL. Hydrophobic (interaction) chromatography. Biochimie 1978;60(1):1–15. https://doi.org/10.1016/S0300-9084(78)80193-X. [21] Queiroz JA, Tomaz CT, Cabral JMS. Hydrophobic interaction chromatography of proteins. J Biotechnol 2001;87(2):143–59. https://doi.org/10.1016/S0168-1656(01) 00237-1. [22] Hofmeister F. Zur Lehre von der Wirkung der Salze. Archiv F€ ur Experimentelle Pathologie Und Pharmakologie 1888;24(4):247–60. https://doi.org/10.1007/BF01918191. [23] Hofstee BHJ, Otillio NF. Non-ionic adsorption chromatography of proteins. J Chromatogr A 1978;159(1):57–69. https://doi.org/10.1016/S0021-9673(00)98546-7. [24] Hamaguchi K, Geiduschek EP. The effect of electrolytes on the stability of the deoxyribonucleate helix. J Am Chem Soc 2002;84(8):1329–38. https://doi.org/10.1021/ JA00867A001. [25] Rippel G, Bede A, Szepesy L. Systematic method development in hydrophobic interaction chromatography I. Characterization of the phase system and modelling retention. J Chromatogr A 1995;697(1–2):17–29. https://doi.org/10.1016/0021-9673(94)01031-9. [26] Kato Y. High-performance hydrophobic interaction chromatography of proteins. In: Giddings JC, editor. Advances in chromatography, vol. 26. CRC Press; 1987. [27] Yon RJ. Chromatography of lipophilic proteins on adsorbents containing mixed hydrophobic and ionic groups. Biochem J 1972;126(3):765–7. https://doi.org/10.1042/ BJ1260765. [28] Amersham Pharmacia Biotech. Hydrophobic interaction chromatography principles and methods 18-1020-90; 2000. [29] Ofarrell PA. Hydrophobic interaction chromatography. In: Molecular biomethods handbook. 2nd edn; 2008. p. 731–9. https://doi.org/10.1007/978-1-60327-375-6_42. [30] McCue JT. Chapter 25 theory and use of hydrophobic interaction chromatography in protein purification applications. Methods Enzymol 2009;463(C):405–14. https://doi.org/ 10.1016/S0076-6879(09)63025-1.

459

460

CHAPTER 17 Hydrophobic interaction chromatography

[31] Jennissen HP, Heilmeyer LMG. General aspects of hydrophobic chromatography. Adsorption and elution characteristics of some skeletal muscle enzymes. Biochemistry 2002;14(4):754–60. https://doi.org/10.1021/BI00675A017. [32] Lienqueo ME, Mahn A, Asenjo JA. Mathematical correlations for predicting protein retention times in hydrophobic interaction chromatography. J Chromatogr A 2002;978 (1–2):71–9. https://doi.org/10.1016/S0021-9673(02)01358-4. [33] Mahn A, Lienqueo ME, Salgado JC. Methods of calculating protein hydrophobicity and their application in developing correlations to predict hydrophobic interaction chromatography retention. J Chromatogr A 2009;1216(10):1838–44. https://doi.org/10.1016/J. CHROMA.2008.11.089. [34] Tomaz CT. Hydrophobic interaction chromatography. Liq Chromatogr Fundam Instrum 2017;1:171–90. https://doi.org/10.1016/B978-0-12-805393-5.00007-5. second edn. [35] Eriksson KO, Belew M. Hydrophobic interaction chromatography. Protein purification: principles, high resolution methods, and applications. 3rd. Wiley; 2011. p. 165–82. [36] Nishikawa AH, Bailon P. Lyotropic salt effects in hydrophobic chromatography. Anal Biochem 1975;68(1):274–80. https://doi.org/10.1016/0003-2697(75)90705-8. [37] Machold C, Deinhofer K, Hahn R, Jungbauer A. Hydrophobic interaction chromatography of proteins: I. Comparison of selectivity. J Chromatogr A 2002;972(1):3–19. https:// doi.org/10.1016/S0021-9673(02)01077-4. [38] Hubert P, Mathis R, Dellacherie E. Polymer ligands for mild hydrophobic interaction chromatography—principles, achievements and future trends. J Chromatogr A 1991;539(2):297–306. https://doi.org/10.1016/S0021-9673(01)83938-8. [39] Hofstee BH. Non-specific binding of proteins by substituted agaroses. Adv Exp Med Biol 1974;42:43–59. https://doi.org/10.1007/978-1-4684-6982-0_4. [40] Chen J, Tetrault J, Ley A. Comparison of standard and new generation hydrophobic interaction chromatography resins in the monoclonal antibody purification process. J Chromatogr A 2008;1177(2):272–81. https://doi.org/10.1016/J.CHROMA.2007.07.083. [41] Jennissen HP. Evidence for negative cooperativity in the adsorption of phosphorylase b on hydrophobic agaroses. Biochemistry 2002;15(26):5683–92. https://doi.org/10.1021/ BI00671A001. [42] Pace CN, Fu H, Fryar KL, Landua J, Trevino SR, Schell D, Thurlkill RL, Imura S, Scholtz JM, Gajiwala K, Sevcik J, Urbanikova L, Myers JK, Takano K, Hebert EJ, Shirley BA, Grimsley GR. Contribution of hydrogen bonds to protein stability. Protein Sci 2014;23(5):652–61. https://doi.org/10.1002/PRO.2449. [43] Haidacher D, Vailaya A, Horva´th C. Temperature effects in hydrophobic interaction chromatography. Proc Natl Acad Sci 1996;93(6):2290–5. https://doi.org/10.1073/ PNAS.93.6.2290. [44] Privalov PL, Makhatadze GI. Heat capacity of proteins: II. Partial molar heat capacity of the unfolded polypeptide chain of proteins: protein unfolding effects. J Mol Biol 1990;213(2):385–91. https://doi.org/10.1016/S0022-2836(05)80198-6. [45] Rustandi RR. Hydrophobic interaction chromatography to analyze glycoproteins. Meth Mol Biol (Clifton, NJ) 2013;988:211–9. https://doi.org/10.1007/978-1-62703-327-5_13. [46] Kimball BA, Arjo WM, Johnston JJ. Single-point calibration with a non-linear detector: carbohydrate analysis of conifer needles by hydrophobic interaction chromatography– evaporative light-scattering detection (HIC–ELSD). J Liq Chromatogr Relat Technol 2006;27(12):1835–48. https://doi.org/10.1081/JLC-120038771. [47] Krull IS, Stuting HH, Krzysko SC. Conformational studies of bovine alkaline phosphatase in hydrophobic interaction and size-exclusion chromatography with linear diode

References

[48]

[49]

[50]

[51]

[52]

[53]

[54] [55]

[56]

[57]

[58]

[59]

[60]

[61]

array and low-angle laser light scattering detection. J Chromatogr A 1988;442(C): 29–52. https://doi.org/10.1016/S0021-9673(00)94455-8. Doellgast GJ, Plaut AG. Purification of human IgA by salt-mediated hydrophobic chromatography. Immunochemistry 1976;13(2):135–9. https://doi.org/10.1016/0019-2791 (76)90281-0. Goudswaard J, Virella G, Noordzij A, Pol J. Isolation of equine IgG(T) by hydrophobic interaction chromatography. Immunochemistry 1977;14(9–10):717–9. https://doi.org/ 10.1016/0019-2791(77)90147-1. Pavlu B, Johansson U, Nyhlen C, Wichman A. Rapid purification of monoclonal antibodies by high-performance liquid chromatography. J Chromatogr A 1986;359(C): 449–60. https://doi.org/10.1016/0021-9673(86)80098-X. Rathore AS, Kumar D, Kateja N. Recent developments in chromatographic purification of biopharmaceuticals. Biotechnol Lett 2018;40(6):895–905. https://doi.org/10.1007/ s10529-018-2552-1. Bansode V, Gupta P, Kateja N, Rathore AS. Contribution of protein A step towards cost of goods for continuous production of monoclonal antibody therapeutics. J Chem Technol Biotechnol 2021;97(9):2420–33. https://doi.org/10.1002/jctb.6686. Kumar V, Rathore AS. Two-stage chromatographic separation of aggregates for monoclonal antibody therapeutics. J Chromatogr A 2014;1368:155–62. https://doi.org/ 10.1016/j.chroma.2014.09.077. Rathore AS, Singh SK. Production of protein therapeutics in the quality by design (QbD) paradigm. Topics Med Chem 2016;21:41–67. https://doi.org/10.1007/7355_2015_5004. Shekhawat LK, Rathore AS. Mechanistic modeling based process analytical technology implementation for pooling in hydrophobic interaction chromatography. Biotechnol Prog 2019;35(2). https://doi.org/10.1002/btpr.2758. Rathore AS, Parr L, Dermawan S, Lawson K, Lu Y. Large scale demonstration of a process analytical technology application in bioprocessing: use of on-line high performance liquid chromatography for making real time pooling decisions for process chromatography. Biotechnol Prog 2010;26(2):448–57. https://doi.org/10.1002/btpr.320. Wu SL, Hancock WS, Pavlu B, Gellerfors P. Application of high-performance hydrophobic-interaction chromatography to the characterization of recombinant DNAderived human growth hormone. J Chromatogr A 1990;500(C):595–606. https://doi. org/10.1016/S0021-9673(00)96094-1. Gueorguieva L, Palani S, Rinas U, Jayaraman G, Seidel-Morgenstern A. Recombinant protein purification using gradient assisted simulated moving bed hydrophobic interaction chromatography. Part II: process design and experimental validation. J Chromatogr A 2011;1218(37):6402–11. https://doi.org/10.1016/J.CHROMA. 2011.07.008. Wang C, Geng X. Refolding and purification of recombinant human granulocyte colonystimulating factor using hydrophobic interaction chromatography at a large scale. Process Biochem 2012;47(12):2262–6. https://doi.org/10.1016/J.PROCBIO.2012.09.002. Diogo MM, Queiroz JA, Monteiro GA, Martins SAM, Ferreira GNM, Prazeres DMF. Purification of a cystic fibrosis plasmid vector for gene therapy using hydrophobic interaction chromatography. Biotechnol Bioeng 2000;68:576–83. https://doi.org/10.1002/ (SICI)1097-0290(20000605)68:5. Diogo MM, Queiroz JA, Monteiro GA, Prazeres DMF. Separation and analysis of plasmid denatured forms using hydrophobic interaction chromatography. Anal Biochem 1999;275(1):122–4. https://doi.org/10.1006/ABIO.1999.4297.

461

462

CHAPTER 17 Hydrophobic interaction chromatography

[62] Seltmann G, Holst O. The outer membrane of the gram-negative bacteria and their components. Bact Cell Wall 2002;9–102. https://doi.org/10.1007/978-3-662-04878-8_2. [63] Diogo MM, Ribeiro S, Queiroz JA, Monteiro GA, Perrin P, Tordo N, Prazeres DMF. Scale-up of hydrophobic interaction chromatography for the purification of a DNA vaccine against rabies. Biotechnol Lett 2000;22(17):1397–400. https://doi.org/10.1023/ A:1005665003375. [64] Diogo MM, Queiroz JA, Prazeres DMF. Studies on the retention of plasmid DNA and Escherichia coli nucleic acids by hydrophobic interaction chromatography. Bioseparation 2001;10(4):211–20. https://doi.org/10.1023/A:1016361721316. [65] Miller NT, Feibush B, Corina K, Powers-Lee S, Karger BL. High-performance hydrophobic interaction chromatography: purification of rat liver carbamoylphosphate synthetase I and ornithine transcarbamoylase. Anal Biochem 1985;148(2):510– 7. https://doi.org/10.1016/0003-2697(85)90260-X. [66] Pirrotta V, Bickle TA. [11] General purification schemes for restriction endonucleases. Methods Enzymol 1980;65(C):89–95. https://doi.org/10.1016/S0076-6879(80)65013-7. [67] Pennings EJM, Meijer AH, Stevens LH, Verpoorte R. The applications of hydrophobic interaction chromatography to the purification of plant proteins. Phytochem Anal 1991;2 (5):191–8. https://doi.org/10.1002/PCA.2800020502. [68] Pisal DS, Kosloski MP, Balu-Iyer SV. Delivery of therapeutic proteins. J Pharm Sci 2010;99(6):2557–75. https://doi.org/10.1002/JPS.22054. [69] Mayolo-Deloisa K, Lienqueo ME, Andrews B, Rito-Palomares M, Asenjo JA. Hydrophobic interaction chromatography for purification of monoPEGylated RNase A. J Chromatogr A 2012;1242:11–6. https://doi.org/10.1016/J.CHROMA.2012.03.079. [70] Weigel T, Soliman R, Wolff MW, Reichl U. Hydrophobic-interaction chromatography for purification of influenza A and B virus. J Chromatogr B 2019;1117:103–17. https:// doi.org/10.1016/J.JCHROMB.2019.03.037. [71] Modrow S, Falke D, Truyen U, Sch€atzl H. Molecular virology. Mol Virol 2013;1–1016. https://doi.org/10.1007/978-3-642-20718-1. [72] Geng X, Chang X. High-performance hydrophobic interaction chromatography as a tool for protein refolding. J Chromatogr A 1992;599(1–2):185–94. https://doi.org/10.1016/ 0021-9673(92)85472-6. [73] Benedek K, Dong S, Karger BL. Kinetics of unfolding of proteins on hydrophobic surfaces in reversed-phase liquid chromatography. J Chromatogr A 1984;317(C):227–43. https://doi.org/10.1016/S0021-9673(01)91662-0. [74] Xiao Y, Rathore A, O’Connell JP, Fernandez EJ. Generalizing a two-conformation model for describing salt and temperature effects on protein retention and stability in hydrophobic interaction chromatography. J Chromatogr A 2007;1157(1–2):197–206. https://doi.org/10.1016/j.chroma.2007.05.009. [75] Gehas J, Wetlaufer DB. Hydrophobic interaction chromatography of small molecules: characterization of the retention of acyl coenzyme a homologues. J Chromatogr A 1989;477(2):249–58. https://doi.org/10.1016/S0021-9673(01)89634-5. [76] Gupta S, Jiskoot W, Sch€oneich C, Rathore AS. Oxidation and deamidation of monoclonal antibody products: potential impact on stability, biological activity, and efficacy. J Pharm Sci 2021;111(4):903–18. https://doi.org/10.1016/j.xphs.2021.11.024. [77] Bhambure R, Gupta D, Rathore AS. A novel multimodal chromatography based single step purification process for efficient manufacturing of an E. coli based biotherapeutic

References

[78]

[79]

[80]

[81]

[82]

[83]

[84]

[85] [86]

[87] [88]

[89] [90]

[91]

[92]

protein product. J Chromatogr A 2013;1314:188–98. https://doi.org/10.1016/j. chroma.2013.09.026. Halan V, Maity S, Bhambure R, Rathore AS. Multimodal chromatography for purification of biotherapeutics—a review. Curr Protein Pept Sci 2018;20(1):4–13. https://doi. org/10.2174/1389203718666171020103559. Harris RJ, Kabakoff B, Macchi FD, Shen FJ, Kwong M, Andya JD, Shire SJ, Bjork N, Totpal K, Chen AB. Identification of multiple sources of charge heterogeneity in a recombinant antibody. J Chromatogr B Biomed Sci Appl 2001;752(2):233–45. https:// doi.org/10.1016/S0378-4347(00)00548-X. Valliere-Douglass J, Wallace A, Balland A. Separation of populations of antibody variants by fine tuning of hydrophobic-interaction chromatography operating conditions. J Chromatogr A 2008;1214(1–2):81–9. https://doi.org/10.1016/J.CHROMA.2008.10.078. Boyd D, Kaschak T, Yan B. HIC resolution of an IgG1 with an oxidized Trp in a complementarity determining region. J Chromatogr B 2011;879(13–14):955–60. https://doi. org/10.1016/J.JCHROMB.2011.03.006. Haverick M, Mengisen S, Shameem M, Ambrogelly A. Separation of mAbs molecular variants by analytical hydrophobic interaction chromatography HPLC: overview and applications. MAbs 2014;6(4):852–8. https://doi.org/10.4161/Mabs.28693. Rodriguez-Aller M, Guillarme D, Beck A, Fekete S. Practical method development for the separation of monoclonal antibodies and antibody-drug-conjugate species in hydrophobic interaction chromatography, part 1: optimization of the mobile phase. J Pharm Biomed Anal 2016;118:393–403. https://doi.org/10.1016/J.JPBA.2015.11.011. Fekete S, Veuthey JL, Beck A, Guillarme D. Hydrophobic interaction chromatography for the characterization of monoclonal antibodies and related products. J Pharm Biomed Anal 2016;130:3–18. https://doi.org/10.1016/J.JPBA.2016.04.004. Fleming R. ADC analysis by hydrophobic interaction chromatography. Methods Mol Biol 2020;2078:147–61. https://doi.org/10.1007/978-1-4939-9929-3_10. Kimerer LK, Pabst TM, Hunter AK, Carta G. Chromatographic behavior of bivalent bispecific antibodies on hydrophobic interaction chromatography columns. J Chromatogr A 2020;1617, 460836. https://doi.org/10.1016/J.CHROMA.2019.460836. Arakawa T, Timasheff SN. Preferential interactions of proteins with salts in concentrated solutions. Biochemistry 2002;21(25):6545–52. https://doi.org/10.1021/BI00268A034. ˚ . An adaptation of hydrophobic interaction chromaGagnon P, Mayes T, Danielsson A tography for estimation of protein solubility optima. J Pharm Biomed Anal 1997;16 (4):587–92. https://doi.org/10.1016/S0731-7085(97)00153-2. Aro T, Fatehi P. Production and application of lignosulfonates and sulfonated lignin. ChemSusChem 2017;10(9):1861–77. https://doi.org/10.1002/CSSC.201700082. Musl O, Sulaeva I, Bacher M, Mahler AK, Rosenau T, Potthast A. Hydrophobic interaction chromatography in 2 D liquid chromatography characterization of lignosulfonates. ChemSusChem 2020;13(17):4595–604. https://doi.org/10.1002/CSSC.202000849. Geng X, Ke C, Chen G, Liu P, Wang F, Zhang H, Sun X. On-line separation of native proteins by two-dimensional liquid chromatography using a single column. J Chromatogr A 2009;1216(16):3553–62. https://doi.org/10.1016/J.CHROMA.2009. 01.085. Reh E, Birkner C, Weigert MP. Possibilities and limitations of hydrophobic interaction chromatography evaluation with strepavidin. Chromatographia 1986;22(7):391–8. https://doi.org/10.1007/BF02268797.

463

464

CHAPTER 17 Hydrophobic interaction chromatography

[93] D’Atri V, Pell R, Clarke A, Guillarme D, Fekete S. Is hydrophobic interaction chromatography the most suitable technique to characterize site-specific antibody-drug conjugates? J Chromatogr A 2019;1586:149–53. https://doi.org/10.1016/J.CHROMA. 2018.12.020. [94] Kato Y, Nakamura K, Kitamura T, Moriyama H, Hasegawa M, Sasaki H. Separation of proteins by hydrophobic interaction chromatography at low salt concentration. J Chromatogr A 2002;971(1–2):143–9. https://doi.org/10.1016/S0021-9673(02)01039-7.

CHAPTER

Ion chromatography

18 P.N. Nesterenko

Lomonosov Moscow State University, Moscow, Russia

18.1 Introduction 18.1.1 Definitions As a rule, the terminology used for the classification of various chromatographic techniques is based on either key separation/retention mechanism utilized. Ion chromatography is an exception to this general rule, as the name was derived from the class of separated analytes, and not the type of chromatographic interaction. This then produces some significant difficulties when attempting to correctly define what chromatographic separations can be collectively termed ion chromatography (IC), as the potential to utilize many varied chromatographic methods for the “high-performance” separation of ions exists. However, for the majority of cases, it is appropriate to define modern ion chromatography as the “high-performance liquid chromatography (HPLC) of small charged solutes, predominantly based upon electrostatic interactions with groups/ions from the ion-exchange stationary phase followed by conductivity detection after effective decrease in concentration of electrolytes composing eluent.” In the following Chapter, those chromatographic methods which closely fit this definition will be presented. Methods for ion separations based upon mainly mobile phase interactions, such as ion-pairing and ion-interaction reversedphase HPLC, are not considered here.

18.1.2 History Staying true to the above definition, the liquid chromatographic separation of 13 lanthanides on a cation-exchange column, as a classified process within the very famous Manhattan Project during the Second World War [1], and the ion exchange-based chromatography of 22 natural amino acids as a part of the amino-acid analysis project established in 1958 by Moore and Stein, together with Spackman [2], are excellent early examples of what can be considered the origins of today’s modern ion chromatography, both of which are examples of remarkably efficient ion-exchange separations for the period. However, the term ion chromatography as we define it Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00021-7 Copyright # 2023 Elsevier Inc. All rights reserved.

465

466

CHAPTER 18 Ion chromatography

today only appeared in 1975 as a result of the work of Hamish Small and colleagues [3], who provided an efficient solution to the ultra-sensitive on-line detection of small inorganic ions separated on an ion-exchange chromatographic column. Small and co-workers were able to improve the sensitivity of conductimetric detection through post-column suppression of background eluent conductivity (see Section 18.3.3), at once establishing a new chromatographic technique for the rapid and quantitative determination of inorganic anions and cations, the fundamentals principles of which have to this day changed only in terms of extended instrumental capabilities.

18.2 Basic principles and separation modes 18.2.1 Ion-exchange chromatography An ion-exchange separation mechanism is mainly based on electrostatic interactions between hydrated ions from a sample (A1, A2 … An), and oppositely charged functional groups (F) of an microparticulate ion-exchanger housed within a chromatographic column. Solute ions are delivered to the column via a flow of an eluent, possessing an appropriate elution strength arising from the presence of competing eluent ions, E. In the process of the separation, the ion-exchanger is equilibrated with eluent ions, and in the case of anion-exchange chromatography, the following basic interaction to describe the exchange process can be written: F+ E2 + Aj 2 ¼ F+ Aj 2 + E2

The strength of interaction between any particularly anion, A j , and the functional group F+ of anion exchanger for the eluent E is evaluated by retention time tR or by retention factor k and the relative retention of anions defines the ion-exchange selectivity of the system. The affinity of an anion-exchanger for a specific anion, A j , can be expressed by either a distribution coefficient: h i A2 j D ¼ h is 2 Aj

m

or by a selectivity coefficient: A

K Ej

h i A ½E  m j s ¼h i A ½E s j m

  where [A j ]s and [Aj ]m are equilibrium concentrations of the anion Aj in the stationary and mobile phases, respectively. The selectivity coefficient KAEj takes into account the affinity of an anion-exchanger toward both analyte anion A j and eluent competing anions, E2. Similar, but precisely opposite expressions are of course true for a cation-exchange process.

18.2 Basic principles and separation modes

The retention factor, k, for an ion retained within an ion-exchange column is proportional to the distribution coefficient, D, and to the phase ratio, φ, which is accepted as constant for each chromatographic column, containing ms grams of stationary phase and Vm milliliters of mobile phase: k¼

tR  to ms ¼ Dφ ¼ D to Vm

Correspondingly, the separation selectivity α ¼ kj/ki is equal to the ratio of distribution coefficients Di/Dj for the pair of separated ions. In practice, for each of the methods presented within this chapter, selectivity depends on the following four parameters: 1. 2. 3. 4.

Properties of the stationary phase (see Section 18.3.1); Properties of the mobile phase (see Sections 18.2.5 and 18.3.3); Type and nature of the solute ion; A combination of secondary interactions and equilibria, such as solvate effects, hydrophobic interactions for organic ions, hydrogen bonding, and phase complexation. If these interactions impacted significantly the retention of target analytes, the retention mechanism is considered as mixed mode.

In the absence of any such secondary interactions, the retention of an ion in ionexchange chromatography is simply proportional to its charge, polarizability and ionic radius. Theoretically, among ions of a similar charge, stronger interaction and therefore chromatographic retention should be expected for smaller ions. However, due to their interaction with water molecules, the actual size of hydrated ions changes drastically and in the opposite order to the size of the original ion, as demonstrated by Figure 18.1. The smaller ions also have a lower polarizability [5]. This results in a reversal of selectivity to that predicted, following the approximate orders shown graphically in Figure 18.1, where affinity toward standard ion-exchange phases increases from left to right.

18.2.2 Ion-exclusion chromatography Ion-exclusion chromatography (IEC), like most modes of liquid chromatography, takes its name from the dominant retention or separation mechanism being exploited, in this case, “ion exclusion.” The technique is mainly applied to the separation of weak acids, particularly carboxylic acids, but has also been applied to the separation of carbohydrates, phenols, and amino acids and can also be used for the separation of weak bases. The commonly accepted retention mechanism at play within IEC involves the formation of a pseudo semi-permeable membrane around the ionexchange resin. This is often referred to as a Donnan membrane equilibrium, through the accumulation of water molecules within a dense hydrated layer upon the surface and within the pore structure of a high capacity ion-exchange resin, forming what is often termed the occluded phase. Ionic solutes of similar charge to the stationary phase (generally sulfonated cation exchangers are used for IEC of weak acids),

467

CHAPTER 18 Ion chromatography

0.4

0.6

in r adi us w

h hyd

ration

hydrati on

in rad ius wit

Na+ ion Incr ic eas po in te g nt ial

F– Ag+ K+

3.2 0.2

ith hyd rati on

Ba2+ Pb2+

4X inc rease

rease

in radiu

s with

Li+

6X inc

with hydrat ion

8X increa

3.4

se in radius

3.6

Ca2+

i nc rea se

ration radius with hyd 10X increase

3.8

Cd2+ Sr2+

hydration

14X increase in

4.0

Zn2+

Mg2+

in radius with

4.2

c ni io ng l si tia ea en cr pot

4.4

In

Be2+

0.8

1.0

2X

4.6

Radius of hydrated ion (Å)

468

1.2

Tl+ Rb+ 1.4

Cl–

NH4+

Br–

I–

Cs+

1.6

1.8

2.0

2.2

Ionic radius with no hydration (Å)

FIGURE 18.1 Correlation between ionic and hydrated radius for the selected groups of inorganic ions. Source: Reproduced with permission from Ref. [4].

experience repulsion from the resin surface, whereas neutral species can penetrate the resins pores and stationary occluded phase, thus experiencing retention. Separation selectivity in such a system is primarily solute dependent, as IEC generally uses very simple eluent systems, such as dilute sulfuric acid (for weak acid solutes), and the exchange groups upon the resin surface do not interact with the similarly charged solutes. Therefore, the potential of solutes to partition from the mobile phase into the stationary phase is governed by solute charge (excluding secondary hydrophobic interactions, which in practice can be significant). For weak acids (HR), this of course is dependent upon acid dissociation constants (pKa values) according to: HRm ¼ H+ m + R m and K a ¼

½H+ m ½R m ½HRm:

Therefore, retention, in the form of elution volume, within IEC can be directly correlated to pKa values for a range of weak acid and basic solutes. Excluding secondary interactions, the solute retention factor in IEC can be described by the following general equation [6]: k¼

Cs,HR V s Cm,HR V m + CR V m

where Vm and Vs are the volumes of the mobile phase (eluent) and occluded liquid stationary phase, respectively, Cm,HR is the concentration of acid HR within the

18.2 Basic principles and separation modes

mobile phase, Cs,HR is the concentration of acid in the stationary phase, and C R is the concentration of anion R in the mobile phase. In practice, IEC is accomplished over a limited working range scaling between totally excluded fully ionized species, and the retention volume of a completely neutral marker. As mentioned above, hydrophobic adsorption of larger organic acids is a significant contributory factor to selectivity in IEC, and often 10–30% of organic solvent is required within the mobile phase to reduce these interactions. Stationary phase capacity and degree of cross-linking (for poly(styrene-divinylbenzene) (PS-DVB) based and similar resins) also play an important role in the resultant selectivity [7].

18.2.3 Chelation ion chromatography Chelation ion chromatography (CIC) is the name given to a mode of IC, where the stationary phase used is one which not only interacts with the solute ions through a simple ion-exchange mechanism, but has the potential to form coordinate bonds based upon multipoint interactions (chelate formation) with the solute ions, typically inorganic cations. This coordination capacity comes from the stationary phase bound chelating functional groups used in CIC, typical of which are iminodiacetic acid (IDA) or aminophosphonic acid (APA) [8]. The strength of chelation is described by the stability constants for each metal cation and the specific immobilized ligand, and as the ligands applied in CIC are typically weak acid based, retention is heavily dependent upon eluent pH. In CIC, retention can results from simultaneous ion exchange and coordination type interactions. The contribution of the former can be minimized through the use of a relatively high ionic strength eluent, typically 0.5 to 1.0 M of an inorganic salt, formed by an alkali metal and non-complexing anion e.g., KNO3. The following expression describes the distribution ratio, DM, of a metal cation, My+, between the negatively charged chelating ion exchanger and the eluent [8]: DM ¼

h i h i ðynÞ+ ðynÞ+ MR + MR e

½My+ 

c

h i h i ðynÞ+ ðynÞ+ where MR and MR are equilibrium concentrations of the metal e

c

cation retained through electrostatic-based ion exchange and stationary phase complexation (chelation), respectively. The free metal ion concentration in the eluent is given by [My+]. Inclusion of a phase ratio constant to describe column-specific variables, φ ¼ Vs/Vm, relates DM to retention factor, k: k ¼ DM :φ

Assuming electrostatic-based ion-exchange interactions can be reduced to negligible levels through the use of high ionic strength eluents and that surface complexation occurs with a 1:1 stoichiometry only, the retention of cations in CIC can be directly

469

470

CHAPTER 18 Ion chromatography

related to the formation constant of the metal–ligand complex formed on the stationary phase, β1:  n  φ k ¼ β1 R

 n  is the concentration of the immobilized chelating functional groups. In where R practice, although eluent pH and ionic strength each exert a major influence on solute retention, the addition of a secondary complexing ligand to the eluent is often utilized to adjust separation selectivity (α) [8,9]. Clearly, in the absence of secondary interactions in the eluent the separation selectivity in CIC is defined by the ratio of the stability constants of the surface complexes formed by separated metals: α¼

k2 βMe2 ¼ 1 k1 βMe1 1

18.2.4 Zwitterionic ion chromatography Zwitterionic ion-exchangers, which can be defined as those within which both positive and negative charges (ion-exchange sites), are located in close proximity, often exhibit alternative ion selectivity to the standard anion or cation exchangers used in IC [10]. With a large variety of possible chemistries, including both strong and weak anion and cation exchange groups in many structural arrangements and combinations, these versatile phases allow greater control of selectivity through manipulation of the simultaneous electrostatic attraction and repulsion forces acting between the solute ions and the stationary phase. The repulsion effect is more profound for multiply charged and bulky ions, which are less retained on zwitterionic ion-exchangers [11]. Also, a decrease in the effective ion-exchange capacity due to the formation of internal salt structures between oppositely charged functional groups within stationary phases results in weaker retention of separated ions in diluted eluents, as shown in Figure 18.2. For some zwitterionic ion-exchangers containing equal numbers of strong anion and cation exchange sites in close proximity to each other (e.g., in a sulfobetaine structure), pure water can be used as an eluent, in what was initially termed electrostatic ion chromatography—EIC’ [12]. An example of EIC separation of the sample containing two salts composed of different cation and anions is shown in Figure 18.3, where four eluted peaks represent all possible ionic associates formed in pure water. Thus, in so-called zwitterionic ion chromatography (ZIC), researchers have exploited these dual functionality phases for the simultaneous separation of both anionic and cationic solutes, often using relatively dilute eluents compared to traditional IC, which additionally acts to increase detector sensitivity, particularly when using conductivity detection. Both covalently bound and dynamically coated phases have been applied to ZIC [9] and utilized with a large variety of eluent systems for the separation of inorganic and organic anions and cations.

18.2 Basic principles and separation modes

FIGURE 18.2 The effect of decreasing retention of inorganic anions on zwitterionic ion-exchanger (silica bound proline) with dilution of the eluent (citric acid). Reproduced with permission from Ref. [11].

a

d

b

0

c

5

10

Retention time (min)

FIGURE 18.3 Chromatogram of an aqueous sample containing 2 mM NaSCN and 1 mM CaCl2. Peak identities: a: Na+-Cl; b: Ca2+-2Cl; c: Na+-SCN; and d: Ca2+-2SCN. Column, ODS-packed column (250  4.6 mm I.D.) coated with Zwittergent-3-14 micelles; mobile phase, water with added Zwittergent-3-14; flow rate, 1.0 mL/min, conductivity detection. Source: Reproduced with permission from Ref. [12].

471

472

CHAPTER 18 Ion chromatography

The use of high concentrations of organic solvent in the eluent under HILIC conditions [13,14] can improve separation selectivity of zwitterionic exchangers. Figure 18.4 illustrates the simultaneous separation of inorganic anions and cations on a bonded zwitterionic phase under HILIC conditions [13]. All common anion-exchangers (more are in Section 18.3.1) are divided into two groups. Type I anion-exchangers containing trialkylammonium functional groups (-N+R3) are normally used with carbonate eluents. Type II anion-exchangers having quaternary ammonium functional groups (-N+R(3-n)R0 n) with both alkyl (R)- and alkanol (R0 )-substituents are considered to be hydroxide-selective. In this case, the retention of separated anions, especially bulky anions, decreases with an increase in hydroxide concentration cOH in the eluent as compared with the corresponding decrease of retention with an increase of carbonate/hydrocarbonate concentration cHCO3/CO3 in the eluent for Type I resin. This difference in chromatographic behavior of Type I and Type II resins is connected with the partial dissociation of hydroxyl groups in alkanol-substituents R0 with increased pH value of the eluent at elevated concentrations of hydroxide [5]. This results in formation of zwitterionic functional groups, which are less selective for bulky anions such as perchlorate, iodate, thiocyanate due to repulsion from dissociated hydroxyl groups in R0 . Therefore, Type II anion-exchange resins should be considered as pH-sensitive zwitterionic rather than hydroxide-selective.

FIGURE 18.4 Simultaneous separation of the 25 common pharmaceutical anions and cations using Acclaim Trinity P1 column (Thermo Fisher, 50  3.0 mm ID, 3 μm) with charged aerosol detection (CAD and five steps gradient elution, flow rate 0.5 mL/min. Peaks: 1—lactate, 2—procaine, 3—choline, 4—tromethamine, 5—Na+, 6—K+, 7—meglumine, 8—mesylate,   9—gluconate, 10—maleate, 11—NO 3 , 12—Cl , 13—Br , 14—besylate, 15—succinate, 2+  16—tosylate, 17—H2PO4 , 18—malate, 19—Zn , 20—Mg2+, 21—fumarate, 22—tartrate, 23—citrate, 24—Ca2+, 25—SO2 4 . Source: Reproduced with permission from Ref. [13].

18.2 Basic principles and separation modes

18.2.5 Eluents for ion chromatography 18.2.5.1 Typical eluents for anion exchange The vast majority of eluents used with IC are water based. In some specific applications, a small percentage of organic solvent may be required to minimize unwanted hydrophobic interactions. The eluents applied to anion exchange chromatography range from dilute electrolyte solutions to complex multi-component buffer solutions and can be inorganic or organic in nature. Elution strength of an eluent for anion exchange chromatography, and therefore solute retention, depends upon the concentration of the eluent competing anion and its selectivity for the particular anion exchange stationary phase. Choice of eluent electrolytes and/or buffer solutions often depends primarily on the mode of detection being used with the particular variety of IC and type of anion-exchange resin. pH of the eluent can also influence retention and separation selectivity of anions representing dissociated weak acids. For the majority of IC applications, conductivity detection is used either directly, indirectly, or following post-column eluent conductivity suppression (see Section 18.3.3), and in each instance, an eluent is chosen to match the requirements for solute detection, namely either a low conductivity eluent, a high conductivity eluent, or an eluent with which conductivity can be suppressed. Table 18.1 shows typical eluents used in anion exchange chromatography, together with the typical mode of detection associated with each eluent. The advanced technology of Reagent-Free™ IC with eluent generation (see Section 18.3.2) is a routine method for in-line preparation of hydroxide, carbonate, and bicarbonate eluents for anion-exchange chromatography. Figure 18.5 shows a typical anion exchange separation of common inorganic and organic anions using a hydroxide eluent, delivered as a gradient, with suppressed conductivity detection.

18.2.5.2 Typical eluents for cation exchange Two main types of eluent systems are commonly applied for the IC separation of inorganic and organic cations, these being either dilute inorganic and organic acids, or in certain applications, protonated amino acids [16] and organic bases. The most popular eluents for standard IC separations of inorganic monovalent and divalent alkali and alkaline earth metal cations, together with small organic amines, are dilute nitric or methanesulfonic acid (MSA). The MSA-based eluent can be electrolytically generated (see Section 18.3.2). In both cases, hydronium (H3O+) is the active eluent competing ion; therefore, solute retention is directly related to eluent pH. As hydronium ions have a very high equivalent ionic conductance, detection is generally in the form of suppressed conductivity (see Section 18.3.3), although sensitive indirect conductivity is also possible. For separations of transition and heavy metal cations, selectivity can often be manipulated through the inclusion of a complexing ligand to the eluent, often an organic acid (citric, oxalic, tartaric, dipicolinic) and ethylenediamine, with detection achieved by post-column reaction (PCR) with a suitable colorforming ligand prior to visible absorbance detection (see Section 18.3.3).

473

474

CHAPTER 18 Ion chromatography

Table 18.1 Common eluents for anion exchange chromatography. Eluent system

Comments

Hydroxide (sodium or potassium)

The hydroxide ion is the weakest eluent competing anion but has a very high equivalent ionic conductance. It is suitable for weakly retained anions and is used with hydroxide selective (Type II) anion exchange columns and can be produced by using eluent generators A stronger eluent as compared with hydroxide, allows greater control of selectivity through control of carbonate/ bicarbonate ratio. This eluent is used with Type I columns and can be produced by using eluent generators Both aromatic and aliphatic SAs are strong eluents, with relatively low to moderate conductivity. Aromatic SAs are highly UV absorbing, and aliphatic SAs are weakly UV absorbing They have low ionic conductances and are strongly UV absorbing. Eluents can provide buffering action over a relatively wide range of pH’s These are moderate strength eluents, and relatively highly conducting, with weak to moderate UV absorption. Elution strength depends on pH. They are useful the elution of weakly retained anions Strong eluents with low UV absorbance. They are incompatible with conductimetric detection

Carbonate/ bicarbonate (sodium or potassium)

Aromatic and aliphatic sulfonic acids (SAs)

Aromatic carboxylic acids (benzoic, phthalic, p-hydroxybenzoic)

Aliphatic carboxylic acids (e.g., citric, tartaric, acetic, and formic acids)

Inorganic salts (chloride, sulfate, phosphate)

Common detection mode Suppressed conductivity or indirect conductivity

Suppressed conductivity

Aromatic SAs—direct and indirect conductivity, indirect UV absorbance. Aliphatic SAs—direct UV absorbance

Direct conductivity or indirect UV absorbance

Direct and indirect conductivity, volatile formic and acetic acids—MS detection

Direct UV absorbance, electrochemical detection

18.3 Instrumentation

F– NO2–

SO4

NO3–

K+

Conductance

Conductance

Cl–

Na+

2–

Br–

Rb+

Li+ NH4+

Cs+

HPO42–

0

(A)

2

4 Minutes

6

0

(B)

3

6 9 Minutes

12

FIGURE 18.5 Typical IC separations of inorganic anions using anion exchange chromatography and a NaHCO3/Na2CO3-based eluent (A) inorganic cations using cation exchange chromatography with an HCl-based eluent (B), both with suppressed conductivity detection. Source: Reproduced with permission from Ref. [15].

18.3 Instrumentation 18.3.1 Ion chromatography columns In the last decade, remarkable progress in the development and production of new IC columns has been achieved [17]. There are a few general parameters for ionexchangers which should be considered when attempting column selection. Ionexchange selectivity depends upon the type of ion-exchange groups, ion-exchange capacity (or more correctly charged group distribution on the surface), accessibility for the interaction with solute ions and, to some degree, the hydrophobicity of the matrix of the stationary phase [5]. Particle size and structure of the bonded layer define separation efficiency.

475

476

CHAPTER 18 Ion chromatography

1. Ion-exchange functionality. Functional groups differ by charge, bulkiness, polarizability, hydrophobicity, and ability for multipoint solute interactions [18]. The type of functional group defines not only the retention mechanism, but also separation selectivity and type of eluent suitable for the separation of a specific group of ions. 2. Ion-exchange capacity expressed as meq/g or meq/mL or meq/column is responsible for the strength of interaction between solute ions and the ionexchanger as well as for the loading capacity of a chromatographic column. Thus, ion-exchange capacity predefines the concentration of the eluent required to elute ions in the shortest possible time without loss of resolution. The first ionexchangers applied in IC had very low ion-exchange capacities, approximately 0.01–0.05 meq/column and were of size 250  4.0 mm ID, due to coarse particles used for the column packing and the limited capability of packed suppressors to suppress background conductivity of eluents (see Section 18.3.3) and hence, provide sensitive conductimetric detection of inorganic anions and cations. Modern suppressors can effectively minimize conductivity of eluents at concentrations up to 200 mM NaOH and up to 100 mM MSA [19], so the current trend in IC is the introduction of ion-exchangers with significantly higher ionexchange capacity (Figure 18.6, bottom). It should be noted that the high ionexchange capacity of IC columns (up to 0.5–0.7 meq per standard size anionexchange column and up to 2.8–8.4 meq per cation-exchange column) extends the limits of IC to the analysis of more concentrated ionic samples and the handling of larger sample volumes. If the ion-exchanger contains weak acidic (carboxylic, phosphonic) or weak basic (primary or secondary amines) functional groups, its capacity also depends on the eluent pH. 3. Ion-exchanger matrix. There are relatively few inorganic (silica, alumina, titania, zirconia, and porous graphitic carbon) materials and organic polymers suitable for the preparation of ion-exchangers. Chemical inertness and hydrolytic stability of the matrix are crucial requirements, especially, for anion-exchangers to be used in IC. Inorganic materials have limited hydrolytic stability in alkaline eluents (silica) or increased reactivity toward phosphates and carboxylates (titania, zirconia), so organic polymer-based ion-exchangers, including highly cross-linked poly(styrene-divinylbenzene) (PS-DVB), poly(methacrylate) (PMA) and poly(vinyl alcohol) (PVA) functionalized materials, are normally used for the separation of anions with alkaline eluents. More efficient columns packed with silica-based ion-exchangers are often preferable for the separation of cations under acid conditions due to the higher separation efficiency (see Section 18.2.5). Other important characteristics of the matrix are porous structure, specific surface area, and hydrophobicity in case of organic polymers (see Table 18.2). Obviously, the latter property should be taken into consideration when optimizing the eluent system for the separation of hydrophobic ions. At present, chemically modified hydrophilic polymers (PVA or PMA) or PS-DVBbased ion-exchangers with immobilized layers of highly hydrophilic ionexchange polymers are developed to minimize unwanted hydrophobic interactions with separated analytes.

18.3 Instrumentation

FIGURE 18.6 Trends in particle size (top) and ion-exchange capacity (bottom) of anion-exchangers produced by Dionex and Thermo Fisher Scientific companies in the last three decades.

477

478

CHAPTER 18 Ion chromatography

Table 18.2 Properties of common substrate matrices used for the preparation of ion-exchangers [5].

Matrix

Hydrolytic stability

Mechanical stability

Residual ionexchange activity

Hydrophobicity

Silica PS-DVB PMA PVA

1–7 0–14 2–12 3–12

Excellent Good Moderate Moderate

Significant No Some Low

Low Substantial Low-moderate Low

4. Particle size. Both column efficiency and peak resolution are dependent on particle size of the adsorbent, so there is a clear current trend to progressively decrease particle sizes for anion-exchange columns, again shown in Figure 18.6, bottom. Modern, fully plastic IC instruments can operate at pressures up to 500 bars allowing the use of IC columns packed with 3–4 μm ion-exchangers. The other path to improve column efficiency is to further develop monolithic porous ion-exchange columns for IC; however, only silica monolithic columns have demonstrated impressive efficiencies for small ions to date, but these are not stable in alkaline eluents [20,21]. 5. Column diameter. In the past decade, great attention has been paid to capillary IC with a series of commercially available columns [22]. This has required a significant modification of eluents generators, suppressors, detectors, and other components of the IC instrumentation. Special interest has been directed to open tubular capillary ion-exchange columns, which can operate at very low pressures [23]. Two very important characteristics of ion-exchangers are their porous structure and the method of immobilization of the functional groups. These characteristics are responsible for the kinetics of mass-transfer of ions within the chromatographic column. According to the structure and method of preparation, the following types of ion-exchangers can be collectively classified: 1. Ion-exchangers with isolated grafted functional groups. These are the most common type of ion-exchangers, in which the surface of a suitable material (inorganic oxides, organic polymers) is modified via one or more surface reactions providing a network of functional groups attached directly to the matrix. Being charged, these groups are uniformly distributed on the outer surface of the substrate due to repulsion from each other. Typical examples include, strong cation-exchange resin Hamilton PRPx200 (see Table 18.4) prepared by sulfonation of porous PS-DVB matrix, or strong anion-exchange resin Hamilton PRPx100 (Table 18.3) with immobilized N,N,Ntrimethylammonium groups prepared by chloromethylation of PS-DVB surface followed by reaction with trimethylamine.

Table 18.3 Properties and common applications of commercially available anion exchangers for IC. Column properties Bonded groups

dp, μm

Column size, mm

Capacity, μeq/col

IonPac AS9HC

-CH(OH)CH2N+R3

9

250  4.0

190

IonPac AS11-HC

-CH(OH) CH2N+CH3(CH2CH2OH)2

9

250  4.0

290

IonPac AS14A

-N+R2R0 OH

9

250  4.0

65

IonPac AS15

-N+R2R0 OH

8.5

250  0.4

2.25

EVB-DVB, 55%; 10 nm pores

IonPac AS15

-N+R2R0 OH

8.5

250  2.0

56

IonPac AS16

-CH(OH) CH2N+CH3(CH2CH2OH)2

9

250  4.0

170

-N+R2R0 OH

10.5

EVB-DVB, 55%; 10 nm pores Macroporous EVB-DVB, 55%; latex beads with cross-linking Microporous EVB-DVB, 55%; latex beads with cross-linking Macroporous EVB-DVB, 55%

Stationary phase

IonPac AS17

IonPac AS19

-N+(CH3)(CH(CH2OH)O)3

7.5

250  4.0

250  4.0

30

160

Matrix

Applications

References

Macroporous 200 nm pores; EVB-DVB, 55%; 90 nm latex beads with 15% cross-linking Macroporous 200 nm pores; EVB-DVB, 55%; 70 nm latex beads with 6% cross-linking EVB-DVB, 55%; 10 nm pores

[41]

80 nm 1%

Determination of trace anions in concentrated hydrofluoric and glycolic acids with ionexclusion pretreatment Determination of trace anions in methanesulfonic and phosphoric acids with ionexclusion pretreatment Determination of hydrolysis products in hexafluorophosphate salts Determination of methanesulfonic acid in Antarctic snow and ice Determination of free cyanide in drinking water Determination of oxyhalides and haloacetic acids in drinking water The determination of trace level phosphorus in purified quartz, trace anions in boric acid

[49]

75 nm 6%

The determination of polyphosphate detergents, trace bromate in drinking water

[50,51]

[42]

[43]

[44,45]

[46] [47,48]

Continued

Table 18.3 Properties and common applications of commercially available anion exchangers for IC—cont’d Column properties Stationary phase

Bonded groups

dp, μm

Column size, mm

Capacity, μeq/col

Matrix

Applications

References

Metrosep A Supp 1 Metrosep A Supp 4 Metrosep A Supp 5, Metrosep Anion Dual 1

-N R3

7.0

250  4.6

64

PS-DVB

Speciation analysis of selenium

[52]

-N+R3

9

250  4.0

46

PVA

[53]

-N+R3

5

100  4.0

39

PVA

-N+R3

10

150  3.1

9

Metrosep Anion Dual 2 IonPac Cryptand A1

-N+R3

6

75  4.6

34

Macroporous poly(hydroxymethacrylate), 20–60 m2/g PMA

2,2,1 cryptand

5

150  3.0

73*

Macroporous, 100 nm pores, EVB-DVB 55%

IonPac AS18

-N+R2R0 OH

7.5

250  4.0

285

IonPac AS20

-N+R2R0 OH

7.5

250  4.0

310

Macroporous, 200 nm pores, EVB-DVB 55%, 65 nm latex beads with 8% cross-linking Macroporous, 200 nm pores, EVB-DVB 55%

Suitable for all routine tasks in water analysis Determination of bromide in canine plasma Determination of inorganic anions in commercial seed oils and in virgin olive oils Determination of chloride in magnesium metal Determination of polyphosphates and polysulfonates; alkanesulfonic acids in a chromic acid plating bath Determination of anions in toothpaste

[60,61]

IonPac AS21

-N+R2R0 OH

7

250  2.0

45

IonPac AS22

-CH(OH)CH2N+R2R0 OH

6.5

250  4.0

210

IonPac AS23

-CH(OH)CH2N+R2R0 OH

6

250  4.0

320

Determination of iodide and iodate in seawater, perchlorate in complex samples Determination of perchlorate in fertilizer Determination of anions in ionic liquids Determination of trifluoroacetic acid in drugs

+

Macroporous, 200 nm pores, EVB-DVB 55% Macroporous, 200 nm pores, EVB-DVB 55% Macroporous, 200 nm pores, EVB-DVB 55%

[54] [55]

[56] [57,58]

[59]

[62] [63] [64]

IonPac AS24

-CH(OH)CH2N+R2R0 OH

7

250  2.0

140

Macroporous, 200 nm pores, EVB-DVB 55%

Metrosep A Supp7

-NR3

5

150  4.0

76

PVA based

IonPac AS26

-CH(OH)CH2N+R2R0 OH

7.5

250  4.0

250

IonPac AS27

-N+R2R0 OH

6.5

250  4.0

220

Hamilton PRPx100 Metrosep A Supp16

-N+(CH3)3

5

250  4.0

190

100  4.0

80

Sulfonated PEVB-DVB (55%, dpore 200 nm, 20 m2/g) layer of hyperbranched polymer Sulfonated PEVB-DVB (55%, dpore 200 nm, 20 m2/g) layer of hyperbranched polymer Mesoporous PS-DVB, 10 nm, 415 m2/g PS-DVB based

Quaternary ammonium

Two-dimensional IC determination of haloacetic acids Simultaneous quantification of chromate, arsenate, selenate, perchlorate, and other Inorganic anions in environmental media Two-dimensional IC determination of haloacetic acids ds

[65]

[66]

[65]

Determination of chlorite and chlorate in fresh-cut vegetables

[67]

Analysis of inorganic arsenic species Determination of trace bromate in various waters

[68] [69]

482

CHAPTER 18 Ion chromatography

2. Ion-exchangers with grafted layers of ionogenic polymers or polymer coated ion-exchangers. This type of surface modification is used when extra ionexchange capacity is required, or when the central core or matrix displays unsatisfactory properties such as a lack of hydrolytic stability or the presence of strong binding sites which must be shielded [70,71]. An example of this type of ion-exchanger is one commonly used for the simultaneous separation of alkali and alkaline earth metal cations, which is based upon a silica particles coated with a layer of poly(butadiene-maleic) acid (PBDMA) [72]. Anion-exchangers with thin polymer ionogenic layers grafted to the surface of substrates can also be prepared by a series of repeating reactions with diepoxide and amines [18,70]. 3. Agglomerated ion-exchangers. These are prepared using composite bead technology, which generally involves the coating, either electrostatically or covalently, of a surface layer of charged nanoparticles upon the surface of a larger supporting substrate oppositely charged microparticle [73]. The presence of the layer of nanoparticles provides favorable mass-transfer kinetics and simultaneously delivers an ion-exchanger with increased surface area and higher ion-exchange capacity. For many years, this type of adsorbents was produced by Dionex, Thermo Fisher Scientific, and Phenomenex [74]. 4. Dynamically modified ion-exchangers. These are usually prepared by saturation of a column packed with a hydrophobic adsorbent with hydrophobic ionic molecules. These ionic molecules are strongly retained by hydrophobic interactions on the column and form a stable coating with aqueous eluents. This method gained popularity because of the ability to prepare various types of ionexchanger, based on the same commercially available efficient reversed-phase column. With this approach, ion-exchangers with multiple functional groups, e.g., zwitterionic groups, can be easily obtained, (see Section 18.2.4) [75].

18.3.1.1 Anion-exchange columns The effective separation of anions formed through dissociation of weak acids is possible only in alkaline conditions, so hydrolytic stability of modern anion-exchangers is a crucial condition. PMA, PVA, PS-DVB and poly(styrene-ethylvinylbenzene) (PS-EVB)-based anion-exchangers form the most common types of commercially available columns. All organic polymers exhibit reasonable stability in alkaline eluents (see Table 18.2), but the low mechanical stability of PVA and PMA limits column length and particle size, due to limited column backpressure tolerance. In the case of PS-DVB-based ion-exchangers, hydrophobic interactions between bulky polarizable anions and substrate contribute significantly to retention and can cause peak distortion. The hydrophobicity of the anion-exchanger can be evaluated by measurement of the methylene group increment α(CH2) in retention of homologues of alkanesulfonic or alkanoic acids [5]. To avoid issues associated with elevated hydrophobicity and to combine mechanical stability of hydrophobic PS-DVB polymer matrix with hydrophilicity of the outer ion-exchange layer, a new generation of, so-called hyperbranched, anion-exchangers was recently developed [70]. For example, the IonPac AS24 anion-exchanger (see characteristics in the Table 18.3) is

18.3 Instrumentation

composed of a core macroporous PS-EVB (55% cross-linking) microparticle, coated with a hydrophilic multi-layer, obtained by consecutive treatment of the surface with diepoxy- and amino-substrates [18,70]. The resulting material has excellent mechanical stability with minimal swelling/shrinking in the presence of organic solvents and concentrated eluents and excellent hydrolytic stability. Anion-exchangers bearing quaternary ammonium groups cover over 95% of all anion-exchangers produced for IC. Among these, two types of anion-exchange functionalities can be highlighted. Anion-exchange “Type I” groups consist of a trialkylammonium head connected through an alkyl spacer to the surface. Anion-exchangers of this type are also known as providing “carbonate type selectivity.” The effect of the structure of the quaternary alkylammonium sites (type of substituent, length and bulkiness of substitutes, length of spacer, etc.) on the separation selectivity is well studied and described in a monograph by Fritz and Gjerde [76]. Anion-exchange “Type II” groups consist of two alkyl substituents and one hydroxyalkyl substituent (usually hydroxyethyl-) on nitrogen and provides so-called “hydroxyl type selectivity,” which is due to partial dissociation of hydroxyls in functional groups and their transformation into zwitterionic structures, as described in Section 18.2.4. Weak acid anionexchangers bearing primary, secondary, and tertiary amino groups may be also used in IC [77]. The properties of some anion-exchangers and anion-exchange columns for IC are presented in Table 18.3. Typical anion-exchange selectivity for Type I anionexchanger is shown in Figure 18.7.

Cations Pu4+ >> La3+ > Ce3+ > Pr3+ > Nd3+ > Sm3+ > Eu3+ > Y3+ > Sc3+ > Al3+ >> Ba2+ > Pb2+ > Sr2+ > Ca2+ > Ni2+ > Cd2+ > Cu2+ > Co2+ > Zn2+ > Mg2+ > Be2+ >> Tl+ > Ag+ > Cs+ > Rb+ > K+ > NH4+ > Na+ > H+ > Li+ Anions PF6– > ClO4– > SCN– > I– > S2O32– > WO42– > MoO42– > CrO42– > C2O42– > SO42– > SO32– > HPO42– > NO3– > Br– > NO2– > CN– > Cl– > HCO3– > H2PO4– > CH3COO– > IO3– > HCOO– > BrO3– > ClO3– > F– > OH–

FIGURE 18.7 Typical ion-exchange selectivity for PS-DVB-based cation-exchangers with sulfonic acid groups and anion-exchangers with quaternary ammonium groups. The selectivity can vary depending on pH (ions associated with weak acids and bases), eluent concentration (anions with different effective charges), temperature, and other factors.

483

484

CHAPTER 18 Ion chromatography

18.3.1.2 Cation exchange columns Commercially available cation-exchangers for ion chromatography can be divided into three groups: 1. Strong acid cation-exchangers bearing sulfonic acid functional groups; 2. Weak acid cation exchangers, which includes carboxylic, phosphonic and phosphoric acid functional groups or their combination; 3. Complexation type ion-exchangers, for example, ion-exchangers with immobilized crown ether groups, capable of forming inclusion complexes with specific cations and thus selectively retain them [17,78]. Typical ion-exchange selectivity for cation-exchanger with sulfonic acid functional groups is shown in Figure 18.7. Sulfonic acid type ion-exchangers have a high affinity for alkaline earth metal cations, which increases the separation time and restricts their practical application for the simultaneous determination of alkali- and alkaline earth metals. However, they can be used for the efficient and selective separation of transition metals by using ethylenediamine—organic acid (tartaric, citric)-based eluents [79], or for the separation of lanthanides with α-hydroxyisobutyric acid containing eluents [80]. Optimum selectivity for alkali- and alkaline earth metal cations and their simultaneous separation in acceptably short runtimes can be achieved with carboxylic type cation-exchangers. This type of cation-exchangers would usually contain an immobilized layer of PBDMA [72], or structurally similar poly(itaconic acid) layer [40]. In some ion-exchangers, e.g., IonPac CS-12A, the addition of a bonded phosphonic acid functionality is used to improve the separation selectivity for manganese in relation to the group of alkaline earth metals [81]. Crown-ethers or macrocycles containing cation-exchangers provides unique selectivity for the separation of cations with similar charge and ionic radius, e.g., potassium and ammonium. The cation-exchanger IonPac CS15 (see Table 18.4) combines three types of functional groups, namely carboxylic, phosphonic acid, and 18-crown-6-ether, displaying a perfectly attuned selectivity for the separation of alkali, alkaline earth, and ammonium cations [27]. The use of acidic eluents for the separation of metal cations is favorable as this prevents unwanted secondary equilibria in chromatographic systems, such as hydrolysis, formation of aqua-complexes and other possible complexation with the active constituents of the eluent. Being hydrolytically stable in acidic eluents, silica provides a good base substrate for the preparation of efficient cation-exchange columns. Therefore, unlike anion exchangers, silica-based cation exchangers for IC are produced by several companies. Table 18.4 shows a summary of the common cation exchangers available for use in IC, while Table 18.5 shows the range of cation exchange columns applied in IEC.

18.3.2 Eluent generators One of the more significant advances in IC over the past decade has been the development of eluent generators. These devices require a concentrated source of electrolyte solution and act to generate the IC eluent (isocratic or gradient) using a flow of

Table 18.4 Properties and common applications of cation exchangers for IC. Column properties Stationary phase

Bonded groups

d p, μm

Column size, mm

Capacity, μeq/col

IonPac CS5A IonPac CS12A

-SO3H

9

250  4.0

-COOH -PO3H2

8.5

250  4.0

2800

IonPac CS15

-COOH -PO3H2 18-crown6 ether

8.5

250  4.0

2800

IonPac CS16

-COOH

5.5

250  5.0

8400

EVB-DVB, 55%; 15 nm pores; 450 m2/g

IonPac CS17

-COOH

7

250  4.0

1450

IonPac CS18

-COOH

6

250  4.0

1160

EVB-DVB, 55%; 15 nm pores; 450 m2/g EVB-DVB, 55%; 15 nm pores; 450 m2/g

Matrix EVB-DVB, 55%; EVB-DVB, 55%; 15 nm pores; 450 m2/g EVB-DVB, 55%; 15 nm pores; 450 m2/g

Applications to analysis of complex matrices

References

Speciation of aluminum complexes Determination of Mg and Ca in 30% NaCl brine

[24]

Determination of trace level Na+ in cooling waters, trace level NH+4 in environmental waste water containing a high Na+ concentration, trace level NH+4 in a KCl soil extract Determination of trace level NH+4 in high concentrations of Na+; trace level Na+ in high concentrations of NH+4 or amines, alkali, and alkaline earth metal ions Simultaneous separation of alkali, alkaline earth cations, and boiler water amine additives Determination of biogenic amines in alcoholic beverages

[27]

[25,26]

[28]

[29]

[30]

Continued

Table 18.4 Properties and common applications of cation exchangers for IC—cont’d Column properties Stationary phase

Bonded groups

d p, μm

Column size, mm

Capacity, μeq/col

IonPac CS19

-COOH

5.5

250  0.4

24

Universal cation

-COOH

7

100  4.6



Metrosep Cation 1–2

-COOH

7

125  4.0

122

Silica based, pores 10 nm, 350 m2/g

Metrosep C2

-COOH

5

250  4.0

194

Silica based

Metrosep C4

-COOH

5

100  4.0

6

Silica based

Metrosep C6

-COOH

5

250  4.0

50

Silica based

IonPac SCS 1

-COOH

4.5

250  4.0

318

Hamilton PRPx200 Hamilton PRPx800

-SO3H

10

250  4.1

35 μeq/g

-COOH

5

150  4.0

3700 μeq/ g

Silica based, 12 nm pores; 300 m2/g PS-DVB, 10 nm pores PS-DVB, macroporous

Matrix EVB-DVB, 55%; macroporous; 60–80 m2/g Silica based

Applications to analysis of complex matrices

References

Separation of biogenic amines

[31]

Determination of cations at trace levels in ice core samples Simultaneous determination of alkali, alkaline earth, and transition metal elements in uranium and thorium-based nuclear fuel materials Determination of methylamines and trimethylamine-N-oxide in particulate matter air samples Determination of adulterants in dietary supplements Transition metal determination in lithium ion battery electrolytes Simultaneous determination of alkali, alkaline earth, and transition metal cations Determination of mercury and methylmercury in seafood Trace alkaline earth and transition metals in brines

[32]

[33]

[34]

[35] [36]

[37,38]

[39] [40]

Table 18.5 Properties and common applications of ion exchangers for ion exclusion IC. Column properties Stationary phase

Bonded groups

d p, μm

Column size, mm

Capacity, μeq/col

IonPac ICE-AS1

-SO3H

7.5

250  9

27,000

IonPac ICE-AS6

Mixed SO 3 and -COOH

8

250  9

27,000

IonPac ICE-Borate

-SO3H

7.5

250  9

27,000

Hamilton PRPx300 Aminex HPX-87H

-SO3H

7

250  4.1

170 μeq/g

-SO3H

9

300  7.8

1700 μeq/g

PS-DVB, 8% cross-linking

TSKgel OApak A

-COOH

5

150  6.0

100 μeq/mL

PMA

TSKgel Super IC-A/C

-COOH

3

150  6.0

200 μeq/mL

PMA

TSKgel SCX

-SO3H

5

150  6.0

>1500 μeq/mL

PMA, 6 nm pores

Matrix

Applications

References

Microporous, PS-DVB, 8% cross-linking, hydrophilic surface Microporous PSacrylate -DVB, 8% cross-linking

Trace analysis of anions in weak acids including hydrofluoric acid

[82,83]

Application to environmental analysis; determination of trace anions in concentrated weak acids Trace borate analysis in deionized water

[41,84]

Determination of arsenate and arsenite Analysis of carbohydrates and organic acids Simultaneous determination of inorganic anions and cations in acid rain waters; organic acids Simultaneous determination of inorganic anions and cations Organic acids

[86]

Microporous, hydrophilic surface, 8% cross-linking PS-DVB

[85]

[87,88]

[89]

[89,90]

[91]

488

CHAPTER 18 Ion chromatography

deionized water from the IC pump module, prior to sample injection. Commercial devices are based either upon electrolysis together with an ion exchange membrane (e.g., Thermo Fisher EG50 Eluent Generator), or a controlled dosing arrangement (e.g., Metrohm 845 Eluent Synthesizer). Additional eluent conditioning devices can be included within each system to remove ionic contaminants or adjust eluent pH, prior to the eluent entering the sample injection module. In the case of electrolytically generated eluents, an additional degassing unit is also necessary. The advantages of the above systems are the production of purer eluents (lower detection background and noise), precise computer control of eluent ion concentration (the ability to produce complex linear and step gradients with an isocratic pump module), and the overall savings in separation time such automation can provide. Figure 18.8 shows the schematic representation of the Dionex EG50 eluent generation device. In this example, the device is configured to generate a potassium hydroxide (KOH) eluent, although similar devices are also available to produce  potassium carbonate/bicarbonate eluents (CO2 3 /HCO3 ), also for anion exchange, or indeed MSA-based eluents for cation exchange-based separations. The most important operational parameters of eluent generators including eluent concentration range, maximum operating pressure, and range of operational flow rates for analytical scale IC are summarized in Table 18.6. It should be noted that capillary modifications of EGC III KOH and EGC III MSA cartridges can generate eluents in the concentration range from 0.1 to 200 mM at flow rates from 0.001 to 0.03 mL/min, which fits the requirements for capillary IC.

Pt anode (H2O

2H+ +

Pump

½O2 +

2e–)

Hydroxide generation chamber

Vent K+ Electrolyte reservoir K+

Cation exchange connector

H2 Degas unit

H2O

(2H2O + 2e–

Pt cathode 2OH– + H2)

KOH

CR-ATC anion trap

FIGURE 18.8 Schematic representation of an eluent generator module for production of potassium hydroxide eluents. Source: Reproduced with permission of Ref. [92].

Table 18.6 Eluent generators produced by Thermo Fisher Scientific. EGC III KOH, NaOH

EGC III LiOH

EGC III MSA

EGC III K2CO3

EGC500-KOH

EGC500-MSA

Flow rate, mL/min

0.01–3.0

Maximum operating pressure, MPa

20.7

Eluent concentration range, mM

0.1–100

Solvent compatibility

Up to 25% CH3OH

Non-compatible with any solvents

Up to 25% CH3OH

Non-compatible with any solvents

IC columns

Type II anion-exchange columns

Cationexchange columns

Type II anionexchange columns

Cation-exchange columns

Operational conditions and limitations.

34.5

0.1–80

0.1–100

Type I anionexchange columns

490

CHAPTER 18 Ion chromatography

18.3.3 Detection in ion chromatography 18.3.3.1 Conductimetric detection Non-suppressed conductivity By far the most common mode of detection in IC is conductivity detection. Conductivity detectors are universal (non-selective) bulk property detectors for ionic solutes, which measure the overall conductivity of the eluent and the transient bands of eluting analyte ions. As the majority of common anions and cations have no, or only weak UV chromophores, conductivity is the detector of choice for the majority of IC applications. Conductivity detection is based upon measurement of the resistance (or strictly the impedance) between two electrodes within the flow cell. The response of a standard conductivity cell (here for anion exchange chromatography) can be described by the following equation: △G ¼

ðλS  λE ÞCS 103 K

where ΔG is the conductance signal, λS- and λE- are the limiting equivalent ionic conductances of the solute (analyte) and eluent anions, respectively, and CS is the concentration of the solute (analyte) anion. K is a constant (called the cell constant) taking into account the physical dimensions of the cell. Conductivity can be measured in two ways. Firstly, in the “direct” mode, an eluent with an overall low ionic conductance is used, commonly a dilute solution of a weak organic acid for anion exchange, under which conditions eluting analyte ions are detected as positive peaks, due to their greater limiting equivalent ionic conductance, compared to the eluent ion. In this configuration, a relatively low capacity ion exchange analytical column (0.01–0.05 meq/column) is typically required. Alternatively, a highly conducting eluent, such as a dilute solution of a strong acid solution or alkali hydroxide can be used, whereby analyte ions exhibit a lower equivalent ionic conductance in relation to the eluent ion, and thus appear as negative peaks within the resultant chromatogram. This mode of conductivity detection is termed indirect conductivity.

Suppressed conductivity The breakthrough in the development of what we today regard as modern IC was achieved in the mid-1970’s by Small, Stevens, and Bauman [3], who introduced what was initially called the stripper column, and which we now know as the first example of an eluent suppressor. The first commercial suppressors were developed by Dionex Corp., in 1975 and the company as a part of Thermo Fisher has remained at the forefront of eluent suppressor technology to this day. An eluent suppressor acts to reduce or “suppress” the background conductivity of the IC eluent through the exchange of ions across an ion exchange membrane (membrane suppressor) or through the use of a high capacity ion exchange resin (packed bed suppressor). In IC methods where the eluent is composed of a simple dilute solution of either a strong acid (e.g., HNO3) or base (e.g., KOH) (see Section 18.2.5), the

18.3 Instrumentation

high background conductance generated by the eluent can essentially be eliminated through the exchange of eluent counter ions, namely NO3 and K+ in the above examples, with either hydroxide (OH) or hydronium ions (H3O+), forming simply H2O in both instances. The provision of these suppressing ions can occur, as mentioned above, from either a packed bed suppressor containing a high capacity anion or cation exchange resin in either basic or acid form, respectively, or delivered across an anion or cation exchange membrane, from a ion source, such as a stream of strong acid or base, or generated electrolytically. In a similar fashion, other common eluent systems for anion exchange chromatography, such as sodium or potassium carbon ate/bicarbonate (CO2 3 /HCO3 )-based eluents, can be converted to a less conducting solution of carbonic acid (H2CO3) through exchange of the alkali cation counter ions with hydronium ions. Figure 18.9 shows a schematic representation of the exchange of ions within a membrane-based suppressor module for a NaOH eluent used for anion exchange-based IC. As can be seen, the analyte counter ion within the sample (being Na+ in the diagram shown) is simultaneously exchanged for H3O+ (H+) during passage through the suppressor module or column, acting to additionally increase the Waste

Waste Analyte in Na+ and OH–

Na+HSO4–

HSO4– + H+

Na+HSO4– Na+

Na+

H+

H+

Analyte in H2O

H+HSO4–

HSO4– + H+

H+HSO4– To detector

Cation exchange membranes

FIGURE 18.9 Schematic representation of a membrane-based suppressor showing chemical suppression of a sodium hydroxide eluent using sulfuric acid.

491

492

CHAPTER 18 Ion chromatography

relative conductance of the analyte band as it passes through the subsequent conductivity detector cell. Recent developments in IC, such as the generation of capillary IC columns, have necessitated the development of new suppressor technologies capable of eluent suppression at capillary scale without introducing significant post-column band broadening. Capillary suppressor modules, such as the ACES300 system, use an ionexchange membrane capillary, coiled within an ion-exchange resin filled chamber, through which a continuous flow of regenerant solution flows. Suppression of eluent ions occurs across the capillary membrane in a similar fashion to the much larger standard membrane suppressor; however, in the capillary format, the internal capillary void volume is as little as 1.5 μL, which provides negligible peak broadening of chromatographic peaks. Two electrode chambers separated from the resin filled chamber by ion exchange membranes are used to electrolytically supply the regenerant ions for continuous eluent suppression.

18.3.3.2 Electrochemical detection Charge detector Charge detection is a type of detection where signal is proportional to the charge for a selected analyte ion and its concentration. The measured total amount of charge, more correctly, an equivalent amount of charges M+ and E ions diffused through anionexchange and cation-exchange membranes located in a flow cell at the anode and cathode (see Figure 18.10), respectively. Then, the appearance of these ions enhances behind membranes and electrolytic dissociation of water causes the final detector response measured as an electric current at a fixed potential. The signal of the charge detector depends on the residence time of the analyte ions in the central channel, which is regulated by the effluent flow rate, and applied voltage effecting diffusion to the electrodes and dissociation of water. The calibration graph is non-linear at low –

Cathode 2H2O + 2e– = 2OH– + H2

H2O

Anion-exchange membrane M+ DC, µA

E–

Suppressed effluent flow

Cation-exchange membrane H2O = 2H+ + 0.5 O2 + 2e– +

FIGURE 18.10 Charge detector operational principle for IC.

Anode

H2O

18.3 Instrumentation

concentration, and the level of background noise is typically higher than for a conductivity detector. The key advantages of charge detection include a stronger signal for multiply charged ions, possibility of sensitive detection of weak acids and bases and additional options for identification of separated ions according to their peak height/charge ratio. For example, the charge detector response for the cations M3+ is three times higher than for cation M+; therefore, comparison of chromatograms obtained with conductimetric and charge detectors connected in series provides valuable information for the charge of separated analytes [93].

Amperometry Amperometric detection can be used for the detection of electro-active solutes, specifically those readily amenable to reduction or oxidation. A certain potential is applied between a working and reference electrode, and where solutes passing over the working electrode are reduced or oxidized, current will flow, this being the principle of the detection method. Amperometric detection requires close control of eluent temperature, pH, and eluent flow (eluent flow through the reactor should be continuous and pulse free) in order to obtain stable and reproducible results. Pulsed amperometric detection (PAD) is the most common mode for IC applications, which utilizes a measuring potential and two cleaning potentials to provide constant electrochemical regeneration of the electrode surface. In PAD, the analysis is performed by a series of cyclic potentials applied to the working electrode, with a measuring potential applied first, and the current measured after a after a suitable equilibration time. Following this, a large positive potential is applied to the electrode causing the oxidative removal of any reaction products, and a subsequent negative potential to return the working electrode to its original state. The whole process is constantly repeated, typically lasting Mw > Mn while, for monodisperse species, Mz ¼ Mw ¼ Mn. Discussion here will focus exclusively on the former scenario, in which case Mz is characteristic of the higher (larger molar mass) end of the MMD, Mn of the lower end, and Mw of an intermediate region near the mode. An example of this is seen in Figure 19.1 for the monomodal MMD of a disperse, linear polystyrene (PS) homopolymer. Because the z-average molar mass is usually located in a region of the MMD occupied by long chains, this average can inform knowledge of processing characteristics such as flex life and stiffness. Conversely, being in the small-molecule region of the MMD, the Mn of a polymer can provide some idea as to the brittleness and flow properties of the material [3]. Given the statistical nature of the various molar mass (M) averages, however, it is possible for polymers with very different MMDs to have identical Mn, Mw, Mz, and so on [4]. Additionally, certain processing and end-use properties such as elongation, hardness, and yield strength may increase with increasing M, but decrease with a narrowing of the MMD. As such, examination of the MMD should almost always be performed in conjunction with an examination of the M averages of a polymer. The most common method for obtaining the MMD and accompanying averages, along with a host of other physicochemical polymer properties, is, generally, size-exclusion chromatography (SEC). aFor

an excellent discussion of the concept of statistical moments as it applies to the MMD of polymers, the reader is referred to

Section 2.4 of reference [1].

Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00022-9 Copyright # 2023 Elsevier Inc. All rights reserved.

509

CHAPTER 19 Size-exclusion chromatography

Mw

1.25

Differential weight fraction

510

Mn

1.00

Mz

0.75

0.50

0.25

0.00

10

4

10

5

6

10

Molar mass M (g mol-1)

FIGURE 19.1 Differential MMD and M averages of broad dispersity, linear PS. Mn ¼ 2.77  105 g mol1, Mw ¼ 5.38  105 g mol1, Mz ¼ 8.29  105 g mol1, Mw/Mn ¼ 1.94. Determined by SEC/MALS/ DRI (MALS: multi-angle static light scattering; DRI: differential refractometry) employing a set of three PSS GRALlinear 10-μm particle size columns and one PSS GRAL10000 10-μm particle size column, preceded by a guard column. Solvent: N,N-dimethyl acetamide +0.5% LiCl; temperature: 35°C; flow rate: 1 mL min1. Detectors: DAWN E MALS and Optilab DSP DRI [2].

19.2 Historical background Many of the relationships between macromolecular properties and the various M averages and/or the MMD were well-recognized by the early 1960s. However, no convenient method existed to determine the MMD and accompanying M averages of a polymer in a single experiment. To address this shortcoming, John Moore at the Dow Chemical Company developed a technique he termed gel permeation chromatography or “GPC.” His paper “Gel permeation chromatography. I. A new method for molecular weight distribution of high polymers” was published in 1964 [5]. The work of Moore built upon earlier research by Wheaton and Bauman and by Porath and Flodin [6–8]. In 1953, the former researchers noted the fractionation of non-ionic substances during passage through an ion exchange column, which indicated that separation based on size should be possible in aqueous solution. This type of separation was demonstrated in 1959 by Porath and Flodin, who employed columns packed with crosslinked polydextran gel, swollen in aqueous media, for the size-based separation of various water-soluble macromolecules. This aqueous-based

19.2 Historical background

technique became known as gel filtration chromatography or “GFC.” While other hydrophobic gels were also developed for the separation of compounds of biological interest, the fact that the gels swelled only in aqueous media limited their application to water-soluble substances. In his pioneering work, Moore employed styrene/divinylbenzene gels crosslinked to a degree that balanced rigidity and permeability. Columns packed with these gels were connected to a differential refractometer, specially designed by James Waters with an optical cell smaller than what was commercially available at the time, with continuous flow in both the sample and reference sides of the cell, and capable of operating at temperatures up to 130°C [9]. Moore recognized that, with proper calibration, GPC could provide both the MMD and M averages of synthetic polymers, a capability which was quickly capitalized upon by many scientists in the polymer industry, who had been longing for just such a technique. In the words of A. C. Ouano, “With the introduction of gel permeation chromatography (GPC) by Moore, molecular weight distribution data for polymers took a sudden turn from near nonexistence to ready availability” [10]. The early history of the instrumental development and commercialization of GPC by Waters Corp. is elegantly recounted in the articles by McDonald [11,12]. We reconcile at this point in the chapter the terms gel permeation chromatography and gel filtration chromatography under the common term size-exclusion chromatography or “SEC.” There are several reasons for doing this: First, elution in both GPC and GFC proceeds by a common size-exclusion mechanism. Second, while many SEC columns are still packed with crosslinked gels, just as many are packed with “non-gel” materials such as porous silica and alumina and, more recently, monoliths. Lastly, because GPC was the term used when operating in organic solvents while GFC denoted experiments in aqueous media, it is difficult to avoid pointing out that a particular researcher might perform GPC experiments on a Monday and GFC experiments on a Friday of the same week, using the exact same hardware (and, perhaps, even the same columns) and separating analytes via the same chromatographic mechanism, only employing a different solvent. Because of these reasons, the all-inclusive and more aptly descriptive term size-exclusion chromatography is preferred and employed from here onward. The column packings employed both by Moore and by Porath and Flodin were lightly crosslinked, semi-rigid networks of large (75–150 μm) particles that could be used only at low flow rates and operating pressures ( 27 mL corresponds to molecules with insufficient angular dissymmetry, at experimental conditions, to allow for accurate measurement of RG; see Section 19.6 for details [2].

where K is the solute distribution coefficient, R the gas constant, T the absolute temperature, and ΔHo and ΔSo are, respectively, the standard enthalpy and entropy differences between the phases. In “traditional” LC (e.g., normal- and reversed-phase LC), retention is generally governed by solute-stationary phase interactions, sorptive or otherwise, and solute transfer between phases is associated with large enthalpy changes. In SEC, noninteracting (hopefully) column packing materials are employed, corresponding to ΔHo  0, and it is the change in entropy between phases that governs solute retention, as per: K SEC  eΔS

o

=R

(19.3)

where KSEC is the solute distribution coefficient in SEC, corresponding to the ratio of the average solute concentration inside the pores of the column packing material to the concentration outside the pores. Because solute mobility is more limited inside than outside the pores, solute permeation in SEC is associated with a decrease in solution conformational entropy, corresponding to negative values of ΔSo. Eq. (19.3) predicts that solute retention in SEC should be temperatureindependent. While it is realized that the size of polymers in solution and, hence, their SEC retention volume, has a modest temperature dependence (and, sometimes, a larger dependence, as when transitioning through the theta point of the solution),

513

514

CHAPTER 19 Size-exclusion chromatography

Table 19.1 Temperature independence of KSEC. KSEC

Methyl-α-D-mannopyranoside Methyl-α-D-galactopyranoside

25°C

50°C

j%Δj

0.768 0.709

0.776 0.706

1.03 0.423

this does not affect the mechanism by which these analytes elute. The entropic nature of SEC retention has been confirmed by noting the virtual lack of change in KSEC with changes in temperature for many mono-, di-, and oligosaccharides in both aqueous and non-aqueous systems [13–19]. Recent results, for a pair of non-mutarotating monosaccharides, are given in Table 19.1, where it is seen that, over a 25°C range, the change in KSEC is 1% or less for these analytes, attesting to the entropic nature of the SEC separation [20]. Solvent: H2O + 0.02% NaN3. Flow rate: 0.5 mL min1. All standard deviations 0 and at poor conditions A2 < 0. This “theta” should not be confused with the scattering angle θ, to which it is unrelated. A2 can usually be determined by employing the MALS photometer off-line, for so-called batch mode experiments; see Section 9.3.3 of Ref. [4] for details.

525

526

CHAPTER 19 Size-exclusion chromatography

Given the dilute nature of the solutions being analyzed, and the additional dilution occurring in a chromatographic experiment, the approximation A2cMw  0 can often be made so that: ΔRðθÞ  Mw K∗c

(19.20)

The consequence of the above is that an SLS measurement at θ ¼ 0o provides the absolute weight-average molar mass of the polymer without assumptions, models, or the need to construct a calibration curve.f While measurements at θ ¼ 0o are highly impractical, when θ ¼ 7o then P(θ) ¼ 0.98 for macromolecules with RG  150 nm, with measurements made at angles closer to zero (i.e., at lower angles) providing for increased accuracy. In 1974, Ouano and Kaye published “Gel-permeation chromatography: X. Molecular weight detection by low-angle laser light scattering,” in which they demonstrated the on-line coupling of SEC to a commercially available low-angle static light scattering (LALS) detector with a specially designed flow-through cell [52]. Measurement of scattered light was made at a narrow angular range of 4.1°–4.8°, and the system also employed a differential refractometer. As seen from Eq. (19.20): Mw ∝

ΔRðθÞ c

(19.21)

The SLS photometer measures ΔR(θ) and the DRI measures concentration c at each slice i eluting from the SEC columns, so that the combination of detectors allows the Mw of each slice, Mw,i, to be determined. Because of the narrowness of the slices, they can be regarded as virtually monodisperse, so that Mw,i  Mn,i  Mz,i…  Mi. Incorporating the Mi from SLS and the ci from DRI into the Meyerhof equation: X

c Mx i i i , when c Mx1 i i i

Mβ ¼ X

x ¼ 0, b ¼ n; when x ¼ 1,b ¼ w; when x ¼ 2, b ¼ z

(19.22)

it can be seen how an SEC/SLS/DRI experiment allows the determination of the various M averages of a polymer, as well as of the polymer MMD, on an absolute basis and without the need to construct a calibration curve. The above determinations of molar mass are performed most accurately by coupling SEC with LALS, the type of detector employed by Ouano and Kaye in their experiments, because in LALS the data do not need to be corrected for angular effects. On the other hand, LALS provides no information about the size of the molecule (as, at θ ¼ 0o, the RG,z term vanishes from Eq. (19.18c)). Also, LALS experiments are notoriously sensitive to dust and other particulate matter (e.g., from the shedding of fines from column packing material), which scatters preferentially in the forward direction, i.e., at low angles, rendering SEC/LALS data inherently noisy and plagued by spikes. It was these drawbacks of LALS (sensitivity to dust, no size f

At a fundamental level, the calculations rely on the relationship between index of refraction, dielectric constant, and polarizability, as given by the Clausius-Mosotti equation and Maxwell’s theory of radiation.

19.6 Size-exclusion chromatography enters the modern era

information) that prompted the introduction of multi-angle static light scattering (MALS) detection. In MALS, the scattered light is measured at a multiplicity of angles, simultaneously, using several individual photodiodes placed at discrete angular intervals around the MALS cell (an example of the placement of photodiodes around a commercial MALS cell is shown in Figure 19.8A, while an example of the cell itself is shown in Figure 19.8B). The combined results are then extrapolated to θ ¼ 0o to obtain Mw, while the angular dependence of the scattered light (angular dissymmetry, due to intramolecular interference effects) is used to provide a measure of the size of the molecule, by way of the z-average radius of gyration RG,z.g Figure 19.9 shows a plot of K*c/ΔR(θ) vs sin2(θ/2) for the slice eluting at the SEC peak apex, as measured by the 90° photodiode of the MALS detector, for the broad PS sample from Figure 19.1. Each data marker represents the measurement from each photodiode of the MALS (the angular placement of the photodiodes is given by the numbers next to the individual data markers), with concentration c provided by the DRI after correction for interdetector delay. Because each slice eluting from the SEC column(s) is considered to be virtually monodisperse, coupling MALS and a concentration-sensitive detector to SEC allows calculation of the statistical averages and distribution of not just M but also RG. Figure 19.10A shows the RG distribution of the PS sample with MMD shown in Figure 19.1. Given in Figure 19.10B is the dependence of RG on M. This so-called conformation plot can inform our knowledge of polymer architecture and/or dilute solution conformation. For example, in the present case, the slope of 0.53 is in the range expected for linear random coils at good solvent/ temperature conditions, 0.5–0.6. MALS detection for SEC was introduced in the mid-1980s [53]. While it took some time for this detection method to gain acceptance, it is now considered the benchmark to which other determinations are compared. Current commercial offerings include systems which measure scattered light at 2, 3, 7, 8, 18, and 21 angles, simultaneously (the two latter systems provide measurements of, at most, 17 and 20 angles when the MALS is connected to a SEC system or other type of separation device). As was the case with intrinsic viscosity, a (long-chain-) branched polymer will have a smaller RG than will its linear counterpart of the same M and composed of the same monomeric repeat unit. As such, a comparison of the RG of the linear and branched molecule, at the same M, can provide an indication of branching. With the MALS detector connected to a SEC system (or other suitable separation method) and assuming a suitable linear standard can be found,h

Molecules that are small compared to λ, the wavelength of radiation in the medium, are considered near-isotropic scatterers, e.g., they scatter almost equally in all directions and, consequently, their RG cannot be determined by MALS. A rule-of-thumb is that the cut-off for accurate determination of RG by MALS is when RG < λ/40, where λ ≡ λ0/n0. h The requirements for accurately performing long-chain branching calculations, including the suitability qualifications for a linear standard, are described in Section 11.2 of Ref. [4] and in Ref. [46]. g

527

528

CHAPTER 19 Size-exclusion chromatography

FIGURE 19.8 (A) Read head of a commercial 18-angle MALS unit. (B) Flow cell assembly of 3-angle and 18-angle commercial MALS units. Figures courtesy of Wyatt Technology Corp.

branching can be determined across the chromatogram and, hence, across the MMD of the polymer. This determination of branching can be made fully quantitative by applying the theory developed by Zimm and Stockmayer in their classic 1949 paper “The dimensions of chain molecules containing branches and rings” [54].

19.7 Size-exclusion chromatography today

1.7x10

-6

1.6x10

-6

1.6x10

-6

K*c/ΔR(θ)

145.2

152.3

137.2 128.4 118.7 109.0

1.5x10

-6

99.5 90.0

1.5x10

-6

1.5x10

-6

1.4x10

-6

1.4x10

-6

80.5 71.0 62.4

26.4

21.2

54.9 47.2 40.6 33.7

0.0

0.2

0.4

0.6

0.8

1.0

sin2(θ/2) FIGURE 19.9 Angular variation of scattered light intensity. Sample and experimental conditions as given in legend for Figure 19.1. Data are for SEC slice eluting at peak apex, as monitored by 90o MALS photodiode, for which Mw ¼ 7.33  105 g mol1 and RG,z ¼ 32 nm. Error bars, representing instrumental standard deviation, are substantially smaller than data markers and, therefore, not shown. Solid line represents non-weighted, first-order linear fit to the data, with r2 > 0.999. Numbers next to individual markers denote angle θ of measurement, in degrees (°), after correction for reflectance at the solvent-glass interface of the MALS cell [2].

19.7 Size-exclusion chromatography today: Multidetector separations, physicochemical characterization, twodimensional techniques While multidetector SEC combining DRI, VISC, and MALS has been in use for at least the last three decades, the combination of detectors being employed and the information obtained therefrom has certainly proliferated in this century [55,56]. The differential viscometer, for one, has grown beyond its role as a universal calibration tool [57] and, like MALS, both these detectors are now also appreciated for the valuable architectural and conformational information they can provide when used together. The addition of quasi-elastic light scattering (QELS, also known as dynamic light scattering) detection, with the detector usually housed in the same unit as the MALS, adds another tool to the arsenal of detectors employed for

529

CHAPTER 19 Size-exclusion chromatography

2.5

(a)

2.0 0.8 1.5 0.6

1.0

0.4

0.5

0.2

0.0 1

Cumulative weight fraction

Differential weight fraction

1.0

0.0 100

10

Radius of gyration RG (nm) 100

Radius of gyration RG (nm)

530

(b)

0.53

10

5

10

6

10

7

10

Molar mass M (g mol-1)

FIGURE 19.10 (A) Differential (solid blue line) and cumulative (open red circles) distributions of RG. Sample and experimental conditions as given in legend for Figure 19.1. Both distributions based on first-order polynomial fits to the experimental data. (B) Conformation plot for same sample as in (A), at same experimental conditions (for explanation of scatter at lower M values, see legend to Figure 19.2). Slope is for a non-weighted, first-order linear fit of M data between 2  105 g mol1 and 3  106 g mol1, for which r2 ¼ 0.999 [2].

characterizing physical aspects of macromolecules (consequently, these detectors are often referred to as “physical detectors”). In addition to long-chain branching information, these detectors can determine the persistence length, characteristic ratio, fractal dimension, etc. of macromolecules across the MMD. Details of how this is done are given in Chapter 11 of Ref. [4].

19.7 Size-exclusion chromatography today

Macromolecular behavior is often dictated not only by the topology but also by the chemistry of polymers, with observed conformational differences among physically similar species [58]. This is especially so in the case of copolymers, where the relative amounts of the different monomers, and the arrangement of these monomers both within the chain and across the MMD, influences processing, end-use, and dilute solution properties [59]. To better understand the underlying basis of this behavior, results from the above physical detectors have been augmented by the addition of mass spectrometry; ultraviolet, infrared, fluorescence, and nuclear magnetic resonance spectroscopy; conductivity; etc. detection. When used in conjunction with MALS and a suitable concentration-sensitive detector, the latter detectors (often referred to as “chemical detectors”) can measure how the average ratio of the various monomers in a copolymer changes as a function of molar mass, a datum known as chemical heterogeneity, or how tacticity or polyelectrolytic charge changes as a function of M. Ludlow et al. have combined various chemical detectors, namely UV, 1H NMR, ESI-MS, and off-line continuous FT-IR in the study of polymer additives [60], while Striegel and colleagues have applied quadruple- and quintupledetector SEC with a combination of physical and chemical detectors to the study of copolymers and blends [41,61]. For example, using SEC/MALS/QELS/VISC/ UV/DRI, with all detectors on-line, Rowland and Striegel determined the chemical heterogeneity in a poly(acrylamide-co-N,N-dimethyl acrylamide) copolymer, the chemical-heterogeneity-corrected molar mass averages and distribution of the copolymer, its dilute solution conformation and how this conformation changed across the MMD, as well as the physicochemical basis for the observed changes, all in a single analysis [61]. While many characteristics of homo- and copolymers can be characterized by multidetector SEC, a number of other separation techniques provide complementary information [3,62]. Examples are the chemical composition distribution (CCD) of copolymers, obtained via so-called “interactive” macromolecular separation methods such as gradient polymer elution chromatography (GPEC), thermal fieldflow fractionation (ThFFF), or liquid chromatography at the critical condition (LCCC) [3,63–69]. Alternatively, techniques such as hydrodynamic chromatography or field-flow fractionation may be employed to study molecules or particles that are either too large or too fragile to analyze successfully by SEC [25–27,30– 34,70,71]. A corollary of these needs is that, for a more complete understanding of complex polymers (understood as polymers with distributions in more than one property) and blends, it is necessary to couple SEC to other types of separation methods. In these two-dimensional (2D) separations, a polymer with distributions in both molar mass and chemical composition may be analyzed by e.g., GPEC  SEC, to determine the combined CCD  MMD of the macromolecule. Figure 19.11 shows results of the ThFFF  SEC analysis of a blend of two block copolymers of polystyrene and poly(methyl methacrylate), PS-b-PMMAs, with the evaporative light scattering detector (ELSD) fractogram at the top of the figure [72]. The molar masses of the two copolymers are nearly identical to one another, but the analytes differ with respect to the ratios of the two polymeric constituents, 83% PS in

531

532

CHAPTER 19 Size-exclusion chromatography

FIGURE 19.11 ThFFF  SEC contour diagram of blend of two PS-b-PMMA copolymers. ThFFF separates by chemical composition in the first dimension, SEC by size in the second dimension. Fractogram, as determined using ELSD detection, is shown at top. See Ref. [72] for details. Reprinted with permission from Ref. [72].

one copolymer, 44% PS in the other. This difference in chemical composition, through its influence on the thermal diffusion coefficient of the analytes, allowed for separation by ThFFF in the first dimension. Analysis by SEC in the second dimension affords, with proper calibration, the molar mass distribution of each constituent in the blend. Recent reviews of multidimensional analysis of polymers include Refs. [73–75], while the theory and applications of 2D-LC with SEC as one of the dimensions are treated in Chapter 14 of Ref. [4]. It can safely be said that multidimensional, multidetector macromolecular separations will be the growth area in polymer chromatography in upcoming years, due to the power of this group of techniques both with respect to peak capacity as well as to the wealth of information they can provide about the physicochemical phase space occupied by complex polymers and blends.

19.8 Conclusions Size-exclusion chromatography can be said to have “come of age” in the 1980s with the ability to determine absolute M, as imparted by on-line static light scattering and viscometric detection (the latter because it permitted construction of universal

References

calibration curves), using robust, commercially available detectors. The next decade was chiefly governed by triple-detector methods involving MALS, VISC, and DRI to determine several physical properties, with many studies involving the measurement (with varying levels of accuracy) of long-chain branching across the MMD of both natural and synthetic macromolecules. During this period, the coupling of SEC to chemical detectors also grew. Today, multidetector SEC experiments may involve three, four, and even five detectors, all on-line, to characterize a wide range of physicochemical properties such as M averages and distributions, chemical and sequencelength heterogeneity, long- and short-chain branching, fractal dimension and persistence length, and more. With a growing knowledge of the power of SEC has also come the realization of many of its limitations, especially when a more complete characterization of complex polymers and blends is desired [4,76,77]. To deconvolute from each other the multiple physical and chemical distributions that may be present in these types of materials, two-dimensional separations are usually necessary, and SEC has found a central role in these methods, as well. Indeed, it is in the acceptance and popularization of multidetector, multidimensional techniques that macromolecular separation science can be expected to grow and to demonstrate its full power in the upcoming years. This will hopefully be accompanied by advances in the synthesis of stationary phases specifically tailored for interactive macromolecular separations, and by continued computer modeling and simulations of the various distributions and heterogeneities present in copolymers and related materials [2,50,59,78].

Disclaimer The identification of certain commercial equipment, instruments, or materials does not imply recommendation or endorsement by the National Institute of Standards and Technology. These identifications are made only to specify the experimental procedures in adequate detail.

References [1] Rudin A. The elements of polymer science and engineering. New York: Academic Press; 1982. [2] Striegel AM. Size-exclusion chromatography. Chapter 10. In: Fanali S, Haddad PR, Riekkola M-L, editors. Liquid chromatography: fundamentals and instrumentation. 2nd ed. Amsterdam: The Netherlands; 2017. p. 245. [3] Striegel AM. Method development in interaction polymer chromatography. Trends Anal Chem 2020;130, 115990. [4] Striegel AM, Yau WW, Kirkland JJ, Bly DD. Modern size-exclusion liquid chromatography. 2nd ed. New York: Wiley; 2009. [5] Moore JC. Gel permeation chromatography. I. a new method for molecular weight distribution of high polymers. J Polym Sci A 1964;2:835–43. [6] Wheaton RM, Bauman WC. Non-ionic separations with ion exchange resins. Ann N Y Acad Sci 1953;57:159–76.

533

534

CHAPTER 19 Size-exclusion chromatography

[7] Porath J, Flodin P. Gel filtration: a method for desalting and group separation. Nature 1959;183:1657–9. [8] Striegel AM. Separation science of macromolecules: what is the role of multidetector size-exclusion chromatography? Chapter 1. In: Striegel AM, editor. Multiple detection in size-exclusion chromatography. Washington, DC: American Chemical Society; 2005. p. 2. ACS Symp Ser 893. [9] Moore JC. Gel permeation chromatography: its inception. J Polym Sci C 1968;21:1–3. [10] Ouano AC. Gel-permeation chromatography. VII. Molecular weight determination of GPC effluents. J Polym Sci A1 1972;10:2169–80. [11] McDonald PD. Waters corporation: fifty years of innovation in analysis and purification. Chem Heritage Summer 2008;33–7. [12] McDonald PD. James Waters and his liquid chromatography people: a personal perspective, www.waters.com. [13] Striegel AM. Anomeric configuration, glycosidic linkage, and the solution conformational entropy of O-linked disaccharides. J Am Chem Soc 2003;125:4146–8 [see Erratum in J. Am. Chem. Soc. 2004:126:4740]. [14] Boone MA, Striegel AM. Influence of anomeric configuration, degree of polymerization, hydrogen bonding, and linearity versus cyclicity on the solution conformational entropy of oligosaccharides. Macromolecules 2006;39:4128–31. [15] Boone MA, Nymeyer H, Striegel AM. Determining the solution conformational entropy of O-linked oligosaccharides as quasi-physiological conditions: size-exclusion chromatography and molecular dynamics. Carbohydr Res 2008;343:132–8. [16] Buley TD, Striegel AM. Relation between the Δ2 effect and the solution conformational entropy of aldohexoses and select methyl glycosides. Carbo Polym 2010;79:241–9. [17] Striegel AM, Boone MA. Influence of glycosidic linkage on solution conformational entropy of oligosaccharides: Malto- vs. isomalto- and cello- vs. laminarioligosaccharides. Biopolymers 2011;95:228–33. [18] Morris MJ, Striegel AM. Does the “C3 effect” offset the Δ2 effect, as regards the solution flexibility of aldoses? Biopolymers 2014;101:703–11. [19] Morris MJ, Striegel AM. Determining the solution conformational entropy of oligosaccharides by SEC with on-line viscometry detection. Carbohydr Polym 2014;106:230–7. [20] Striegel AM, Trainoff SP. Entropic-based separation of diastereomers: size-exclusion chromatography with online viscometry and refractometry detection for analysis of blends of mannose and galactose methyl-α-pyranosides at “ideal” size-exclusion conditions. Chromatographia 2021;84:37–45. [21] Richard D.J., Striegel A.M.. The obstruction factor in size-exclusion chromatography. 1. The intraparticle obstruction factor. J Chromatogr A 2010;1241:7131–7137. [22] Little JN, Waters JL, Bombaugh KJ, Pauplis WJ. Fast gel-permeation chromatography. I. A study of operational parameters. J Polym Sci A-2 1969;7:1775–83. [23] Yau WW, Malone CP, Fleming SW. The equilibrium distribution coefficient in gel permeation chromatography. J Polym Sci B: Polym Lett 1968;6:803–7. [24] Striegel AM. Thermodynamic equilibrium of the solute distribution in size-exclusion chromatography. J Chromatogr A 2004;1033:241–5. [25] Striegel AM. Hydrodynamic chromatography: packed columns, multiple detectors, and microcapillaries. Anal Bioanal Chem 2012;402:77–81. [26] Striegel AM, Brewer AK. Hydrodynamic chromatography. Ann Rev Anal Chem 2012;5:15–24.

References

[27] Brewer AK. Hydrodynamic chromatography: the underutilized size-based separation technique. Chromatographia 2021;84:807–11. [28] Uliyanchenko E, van der Wal S, Schoenmakers PJ. Deformation and degradation of polymers in ultra-high-pressure liquid chromatography. J Chromatogr A 2011;1218:6930–42. [29] Striegel AM. Longitudinal diffusion in size-exclusion chromatography: a stop-flow sizeexclusion chromatography study. J Chromatogr A 2001;932:21–31. [30] Striegel AM. Observations regarding on-column, flow-induced degradation during SEC analysis. J Liq Chromatogr Rel Techn 2008;31:3105–14. [31] Striegel AM, Isenberg SL, C^ote GL. An SEC/MALS study of alternan degradation during size-exclusion chromatographic analysis. Anal Bioanal Chem 2009;394:1887–93. [32] Isenberg SL, Brewer AK, C^ote GL, Striegel AM. Hydrodynamic versus size exclusion chromatography characterization of alternan and comparison to off-line MALS. Biomacromolecules 2010;11:2505–11. [33] Brewer AK, Striegel AM. Characterizing string-of-pearls colloidal silica by multidetector hydrodynamic chromatography and comparison to multidetector size-exclusion chromatography, off-line multiangle static light scattering, and transmission electron microscopy. Anal Chem 2011;83:3068–75. [34] Striegel AM. Do column frits contribute to the on-column, flow-induced degradation of macromolecules? J Chromatogr A 2014;1359:147–55. [35] Giddings JC. Dynamics of chromatography. New York: Marcel Dekker; 1965. [36] Giddings JC. Unified separation science. New York: Wiley; 1991. [37] Snyder LR, Kirkland JJ, Dolan JW. Introduction to modern liquid chromatography. 3rd ed. New York: Wiley; 2010. [38] Karger BI, Snyder LR, Horvath C. An introduction to separation science. New York: Wiley; 1973. [39] Yau WW, Kirkland JJ, Bly DD, Stoklosa HJ. Effect of column performance on the accuracy of molecular weights obtained from size exclusion chromatography (gel permeation chromatography). J Chromatogr 1976;125:219–30. [40] Grubisic Z, Rempp P, Benoit H. A universal calibration for gel permeation chromatography. J Polym Sci Polym Lett 1967;5:753–9. [41] Haidar Ahmad IA, Striegel AM. Determining the absolute, chemical-heterogeneitycorrected molar mass averages, distribution, and solution conformation of random copolymers. Anal Bioanal Chem 2010;396:1589–98.  [42] Benoit H, Grubisic Z, Rempp P, Decker D, Zilliox J- G. Etude par chromatographie en phase liquid de polystyre`nes lineaires et ramifies de structures connues. J Chim Phys 1966;63:1507–14. [43] Benoit HC. Reflections on “A universal calibration method for gel permeation chromatography”. Z. Grubisic, P. Rempp, and H. Benoit, J Polym Sci 5, 753 (1967), J Polym Sci B 1996;34:1703–4. [44] Haney MA. The differential viscometer. I. a new approach to the measurement of specific viscosities of polymer solutions. J App Polym Sci 1985;30:3023–36. [45] Haney MA. The differential viscometer. II. On-line viscosity detector for size-exclusion chromatography. J App Polym Sci 1985;30:3037–49. [46] Striegel AM. Long-chain branching macromolecules: SEC analysis. In: Cazes J, editor. Encyclopedia of chromatography. 3rd ed. New York: Taylor & Francis; 2010. p. 1417. [47] Dubin PL, Principi JM. Failure of universal calibration for size-exclusion chromatography of rodlike macromolecules vs random coils and globular proteins. Macromolecules 1989;22:1891–6.

535

536

CHAPTER 19 Size-exclusion chromatography

[48] Pannell J. Gel permeation chromatography: the behavior of polystyrenes with long-chain branching. Polymer 1972;13:277–82. [49] Temyanko E, Russo PS, Ricks HH. Study of rodlike homopolypeptides by gel permeation chromatography with light scattering detection: validity of universal calibration and stiffness assessment. Macromolecules 2001;34:582–6. [50] Striegel AM, Plattner RD, Willett JJ. Dilute solution behavior of dendrimers and polysaccharides: SEC, ESI-MS, and computer modeling. Anal Chem 1999;71:978–86. [51] Wyatt PJ. Light scattering and the absolute characterization of macromolecules. Anal Chim Acta 1993;272:1–40. [52] Ouano AC, Kaye W, Gel-permeation chromatography, X. Molecular weight detection by low-angle laser light scattering. J Polym Sci Polym Chem Ed 1974;12:1151–62. [53] (a) Wyatt PJ, Jackson C, Wyatt GK. Absolute GPC determinations of molecular weights and sizes from light scattering. Am Lab 1988;20:86–9. (b) Wyatt PJ, Hicks DL, Jackson C, Wyatt GK. Absolute GPC determinations of molecular weights and sizes. Am Lab 1988;20:108–11. [54] Zimm BH, Stockmayer WH. The dimensions of chain molecules containing branches and rings. J Chem Phys 1949;17:1301–14. [55] Striegel AM. Editor. Multiple detection in size-exclusion chromatography. ACS Symp Ser.; 893Washington. DC: American Chemical Society; 2005. [56] Striegel AM. Multiple detection in size-exclusion chromatography of macromolecules. Anal Chem 2005;77:104A–13A. [57] Striegel AM. Viscometric detection in size-exclusion chromatography: principles and select applications. Chromatographia 2016;79:945–60. [58] Haidar Ahmad IA, Striegel AM. Influence of second virial coefficient and persistence length on dilute solution polymer conformation. Anal Bioanal Chem 2011;399:1515–21. [59] Haidar Ahmad IA, Striegel DA, Striegel AM. How does sequence length heterogeneity affect the dilute solution conformation of copolymers? Polymer 2011;52:1268–77. [60] Ludlow M, Louden D, Handley A, Taylor S, Wright B, Wilson ID. Size-exclusion chromatography with on-line ultraviolet, proton nuclear magnetic resonance and mass spectrometric detection and on-line collection for off-line Fourier transform infrared spectroscopy. J Chromatogr A 1999;857:89–96. [61] Rowland SM, Striegel AM. Characterization of copolymers and blends by quintupledetector size-exclusion chromatography. Anal Chem 2012;84:4812–20. [62] Striegel AM. Separation science of macromolecules. Anal Bioanal Chem 2011;399:1399–400. [63] Striegel AM. Determining the vinyl alcohol distribution in poly(vinyl butyral) using normal-phase gradient polymer elution chromatography. J Chromatogr A 2002;971:151–8. [64] Striegel AM. Determining and correcting “moment bias” in gradient polymer elution chromatography. J Chromatogr A 2003;996:45–51. [65] Philipsen HJA. Determination of chemical composition distribution in synthetic polymers. J Chromatogr A 2004;1037:329–50. [66] Hutchings LR, Pagliarulo A. Temperature gradient interaction chromatography: a perspective. Chromatographia 2021;84:813–8. [67] Staal WJ. Gradient polymer elution chromatography (GPEC): under-estimated powerful tool for polymer characterization: a perspective. Chromatographia 2022. [68] Pasch H. Thermal field-flow fractionation as a powerful tool for the fractionation of complex synthetic polymers: a perspective. Chromatographia 2021;84:525–30.

References

[69] Malik AM. Liquid chromatography at critical conditions in polymer analysis: a perspective. Chromatographia 2021;84. [70] Striegel AM. Multi-detector hydrodynamic chromatography of colloids: following in Hamish Small’s footsteps. Heliyon 2021;7, e06691. [71] Podzimek S, Johann C. Asymmetric flow field-flow fractionation: current status, possibilities, analytical limitations and future trends. Chromatographia 2021;84:531–4. [72] Viktor Z, Pasch H. Two-dimensional fractionation of complex polymers by comprehensive online-coupled thermal field-flow fractionation and size exclusion chromatography. Anal Chim Acta 2020;1107:225–32. [73] Berek D. Two-dimensional liquid chromatography of synthetic polymers. Anal Bioanal Chem 2010;396:421–41. [74] Baumgaertel A, Altuntas¸ E, Schubert US. Recent developments in the detailed characterization of polymers by multidimensional chromatography. J Chromatogr A 2012;1240:1–20. [75] Pasch H. Hyphenated separation techniques for complex polymers. Polym Chem 2013;4:2628–50. [76] Berek D. Size exclusion chromatography - a blessing and a curse of science and technology of synthetic polymers. J Sep Sci 2010;33:315–35. [77] Striegel AM. Size-exclusion chromatography: a 21st century perspective. Chromatographia 2022. [78] Dong S, Striegel AM. Monte Carlo simulation of the sequence length and junction point distributions in random copolymers obeying Bernoullian statistics. Int J Polym Anal Charac 2012;17:247–56.

537

CHAPTER

Affinity chromatography

20

David S. Hage, Jeanethe A. Anguizola, Rong Li, Ryan Matsuda, Efthimia Papastavros, Erika Pfaunmiller, Matthew Sobansky, and Xiwei Zheng University of Nebraska, Lincoln, NE, United States

20.1 Introduction Affinity chromatography is a type of liquid chromatography in which a biologically related agent is used in a column as a stationary phase to purify or analyze the components of a sample or complex mixture [1–5]. The ability of this method to selectively bind and purify its target compounds is based on the specific and reversible interactions that are present in many biological systems, such as the binding of a hormone to a receptor or an antibody to its antigen [1–4]. To develop a method based on affinity chromatography, one of a pair of interacting species is first immobilized to a solid support, such as agarose beads or silica particles. The immobilized agent, which is called the affinity ligand, acts as the stationary phase for the affinity column [1,3]. The target compound for the affinity ligand is then injected onto the affinity column or passed through this column in the presence of an application buffer, which has a pH and composition that allows the desired target to bind to the immobilized ligand (see Figure 20.1) [6]. After nonretained sample components have been washed from the column, the retained target can be released in the presence of an elution buffer [2,3,6]. The elution buffer may have a different pH or composition than the application buffer or it may contain a competing agent that causes the retained target to be released from the affinity ligand [6]. If the retained compound has only weak or moderate binding to the immobilized ligand, it is also possible to use the same application buffer to both apply and elute this target under isocratic conditions; this approach is known as weak affinity chromatography (WAC) [6–12]. As the target elutes, it may be collected for further use or analyzed by an on-line or off-line detector. The column can then be regenerated by reapplying the application buffer before the next sample is injected.

Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00034-5 Copyright # 2023 Elsevier Inc. All rights reserved.

539

540

CHAPTER 20 Affinity chromatography

FIGURE 20.1 Typical scheme for the application of a sample to an affinity column, elution of the retained targets, and regeneration of the affinity column [1].

20.2 Basic components of affinity chromatography The success of any affinity separation depends largely on the selection of the affinity ligand, or binding agent, that is immobilized in the column. A summary of common binding agents that are used in affinity chromatography is given in Table 20.1. Many of the ligands used in affinity chromatography are obtained from a biological source; examples of these ligands are antibodies, serum proteins, and lectins. Other binding agents and ligands that are useful in affinity chromatography are boronic acid, metal chelates, and triazine dyes, which are synthetic agents or inorganic molecules [1–5]. Affinity ligands can be divided into two main categories: high-specificity ligands and general ligands [3]. High-specificity ligands are binding agents that retain only one or a few closely related targets. These ligands are used when the goal is to isolate or separate a specific solute from a sample or mixture. Examples of high-specificity ligands include antibodies for the capture of antigens, substrates or inhibitors for binding to given enzymes, and single-strand nucleic acids for the retention of

20.2 Basic components of affinity chromatography

Table 20.1 Common ligands used in affinity chromatography. Type of ligand

Retained targets

Biological Ligands Antibodies Inhibitors and substrates Cofactors and coenzymes Lectins Nucleic acids Protein A/protein G

Antigens (drugs, hormones, peptides, proteins, viruses, and cell components) Enzymes Sugars, glycoproteins, and glycolipids Complementary nucleic acids and DNA/RNA-binding proteins Antibodies

Non-biological ligands Boronates Triazine dyes Metal-ion chelates

Sugars, glycoproteins, and diol-containing compounds Nucleotide-binding proteins and enzymes Metal-ion-binding amino acids, peptides, and proteins

This table is based on information provided in refs. [1–5].

complementary sequences of DNA or RNA [1]. General ligands are binding agents that retain a class of related molecules or structurally similar targets. Examples of general ligands are lectins and boronates for the binding of carbohydrate-containing agents, some types of dyes for the retention of enzymes and proteins, and protein A or protein G for the binding of immunoglobulins [3]. When designing a system for use with affinity chromatography, an important aspect to consider is the selection of the support material that is used for ligand attachment. This support material should have low non-specific binding for sample components but be easy to modify for chemical activation and ligand attachment. In addition, the support should be able to withstand the pressures and flow rates that will be used in the final separation method [1]. Agarose is often used as a support in traditional affinity columns [1,4]. This material consists of polymeric chains comprised of D-galactose and 3,6-anhydro-L-galactose [13]. Another common type of polysaccharide support is cellulose. Work has also been conducted in the use of affinity ligands with supports for high-performance liquid chromatography (HPLC), resulting in a method known as high-performance affinity chromatography (HPAC) or high-performance liquid affinity chromatography (HPLAC) [1,4,7]. Supports used in this latter method include silica particles, modified polystyrene supports, silica monoliths, and organic-based monoliths [4,7,13–16]. Affinity ligands can also be used with monolith supports, giving a technique known as affinity monolith chromatography (AMC) [17–19]. Several techniques are available for attaching an affinity ligand to a chromatographic support. These techniques include both covalent and non-specific immobilization methods [20]. Non-specific immobilization techniques involve the physical adsorption of an affinity ligand to a support [21,22]. Biospecific adsorption is a form

541

542

CHAPTER 20 Affinity chromatography

of non-covalent immobilization that uses the binding between two affinity ligands, one of which has been previously bound to the support and the second of which is adsorbed to the first ligand and used to bind the final desired target. An example of this approach is the use of covalently immobilized avidin or streptavidin for the adsorption of biotinylated proteins [20,23]. Another example of biospecific adsorption is the use of immobilized protein A or G to adsorb antibodies for the creation of immunoaffinity supports [20,24,25]. Covalent immobilization involves the chemical attachment of a ligand to a chromatographic support. In this method, the support must first be activated for ligand attachment. Several functional groups can be used for covalent immobilization. These methods may involve amine groups, carboxyl groups, sulfhydryl groups, hydroxyl groups, or aldehyde groups [20]. Other possible routes for the immobilization of an affinity ligand include the use of coulombic interactions or coordination complexes and the use of entrapment or encapsulation [20,26,27]. The application buffer used with an affinity column should have an appropriate pH and ionic strength to promote binding between the immobilized ligand and target [6]. The elution buffer is a mobile phase that disrupts the binding of the target with the ligand. This elution may be accomplished by changing the pH, ionic strength, or amount of organic solvent in the mobile phase. This elution format, referred to as non-specific elution, is often used in analytical methods for the quick removal of a target from an affinity column. Alternatively, a competing agent may be placed in the mobile phase to displace the target from the column by binding with either the target or the ligand, thus preventing their further interaction. This method is called biospecific elution. Although biospecific elution is often slower than nonspecific elution, it is useful when gentle removal and purification of an active target are desired [6].

20.3 Bioaffinity chromatography Bioaffinity chromatography is a type of affinity chromatography that employs a biological agent as the stationary phase [28]. This technique was first used in 1910 by Starkenstein to purify alpha-amylase by using a column that contained insoluble starch as both the support and affinity ligand [29,30]. Bioaffinity chromatography is now commonly used as a purification technique for numerous compounds [28,31–34]. Because many biological agents are found in nature, bioaffinity chromatography is one of the most diverse forms of affinity chromatography. Several examples of biological agents that can be used as ligands in bioaffinity applications are listed in Table 20.1. Many of these agents are commercially available in an immobilized form for use in affinity columns [3,28,31]. Enzyme purification was the first application of bioaffinity chromatography [30] and remains an important use of this technique. In this type of separation, ligands such as enzyme inhibitors, coenzymes, or cofactors are used to purify and separate enzymes [28,33]. For instance, in 1968 and the first report of “modern” affinity

20.3 Bioaffinity chromatography

chromatography, Cuatrecasas, Wilchek, and Anfinsen employed specific enzyme inhibitors to selectively isolate the corresponding enzymes that could bind these inhibitors [1,5]. This general approach has also been used with a support containing flavin mononucleotides for the purification of flavin adenine dinucleotide synthetase [33]. Other examples have included the use of mono-, di-, and triphosphate nucleotides for the purification of kinases and the use of nicotinamide adenine dinucleotide for the isolation of dehydrogenases [33,34]. Lectins are another group of ligands that are frequently utilized in bioaffinity chromatography. Lectins are non-immune system proteins that can bind to specific carbohydrate groups [28]. The most common lectins used in bioaffinity chromatography are concanavalin A (Con A), wheat-germ agglutinin (WGA), and jacalin [4,28,32]. Con A binds to targets that contain α-D-mannose or α-D-glucose residues [4,32]. WGA specifically binds to D-N-acetylglucosamine residues, and jacalin binds to galactose or mannose residues [32,35]. The most popular application for lectins has been in glycomics, where these ligands have been used to isolate and separate polysaccharides, glycoproteins, glycopeptides, and glycolipids [4,31,32]. For instance, supports containing multiple lectins (e.g., Con A, WGA, and jacalin) have been used for the separation of glycoproteins in plasma samples and to help map glycosylation patterns for disease detection [36,37]. Immunoglobulin-binding proteins are another class of ligands that are employed in bioaffinity chromatography. Proteins A and G are two examples of such binding agents. Protein A is produced by Staphylococcus aureus, and protein G is produced by group G streptococci [28,38–40]. Both proteins bind to the Fc region of immunoglobulins [4,28], which makes these ligands useful for antibody purification and the biospecific adsorption of antibodies to affinity supports [4,28]. Protein A and protein G have some differences in the species and classes of antibodies to which they will bind [28,38–40]. For example, protein A can bind to human antibodies that belong to the classes of immunoglobulin G (IgG), IgM, and IgA; however, protein G can bind to only IgG-class antibodies from humans [28,38–40]. Both these proteins bind strongly to immunoglobulins at a neutral pH but dissociate from their targets at a mildly acidic pH [28]. Hybrid forms of proteins A and G, such as protein A/G, are also available, which can increase the types of immunoglobulins to which such a ligand is able to bind [28,41]. Nucleic acids and polynucleotides are additional binding agents that can be used in bioaffinity chromatography. This combination produces a method referred to as DNA affinity chromatography [42–44]. Methods in DNA affinity chromatography can be classified into two categories: non-specific techniques and sequencespecific techniques. Ligands used in non-specific DNA chromatography are generally composed of fragmented nuclear DNA (e.g., salmon sperm DNA or calf thymus DNA) [42]. In this technique, DNA-binding proteins can be bound to the immobilized ligand for their separation from other sample components [42]. Sequence-specific DNA chromatography is used for the purification and isolation of targets that bind to specific sequences of the polynucleotide that is being used as the ligand [42–44].

543

544

CHAPTER 20 Affinity chromatography

20.4 Immunoaffinity chromatography Immunoaffinity chromatography (IAC) is a special type of bioaffinity chromatography that uses antibodies or antibody-related molecules as the stationary phase. IAC columns typically have strong and selective binding between antibodies and their target antigens. This feature, as well as the ability to produce antibodies against a wide range of targets, has made IAC popular for use in both the purification and analysis of chemicals in a variety of complex samples [45–48]. The most common technique used in IAC is the on–off elution mode, as illustrated earlier in Figure 20.1 [46,47]. In this format, a sample is injected into the IAC column under conditions in which the target has a strong affinity to antibodies immobilized within the column. Only the target and closely related compounds are ideally retained by the IAC column under these conditions, while other sample components are not retained. After the non-retained sample components have been washed away, an elution buffer is applied to the column to dissociate the bound target from the column. In this format, it is often easy to obtain baseline resolution between the non-retained and retained peaks by changing the time at which the elution buffer is applied. Proteins, glycoproteins, carbohydrates, lipids, bacteria, viral particles, drugs, and environmental agents have all been isolated by IAC through the use of this format [45,47–53]. The main advantage of this approach is its speed and simplicity, especially when it is carried out as part of an HPLC system. The use of antibodies or related binding agents with HPLC supports is referred to as high-performance immunoaffinity chromatography (HPIAC) [54,55]. The use of HPIAC or IAC to selectively remove and enrich a target from a sample is often referred to as immunoextraction. The use of these methods to remove an interfering substance from a sample (e.g., a major protein during the analysis of a trace protein) is sometimes called immunodepletion [46–48,53]. Chromatographic (or flow-injection) immunoassays represent another important mode of IAC. This technique uses immobilized antibodies or antigens in a column to perform various competitive or non-competitive immunoassays. Detection in these methods involves the use of a labeled antibody or labeled analyte analog (often referred to as simply the “label”) to indirectly measure the amount of a target in a sample, as demonstrated in Figure 20.2. Detection in such an assay may be based on measurements of such things as the label’s fluorescence, chemiluminescence, electrochemical activity, radioactivity, or thermal properties. The label may consist of an enzyme, fluorescent tag, chemiluminescent agent, liposome, or radioisotope, among other possibilities [54–56]. Competitive binding immunoassays involve competition between a target in a sample and a labeled analyte analog for a limited amount of antibodies that are capable of binding to both the target and labeled analog [56]. After they have been allowed to compete for binding sites on the antibodies, the bound and free portions of the target and labeled analog are separated. The amount of labeled analog in the bound or free fraction is then measured. The absence of any target results in the maximum measured amount of labeled analog in the bound fraction. An increase in the amount of target results in a decrease in the measured level of the bound labeled

20.4 Immunoaffinity chromatography

FIGURE 20.2 Chromatographic immunoassay based on a sandwich, or two-site immunometric, format.

analog, which provides a signal that is indirectly related to the amount of target in the original sample. Several formats are available for a competitive binding immunoassay in IAC or HPIAC. The simultaneous injection format involves the injection of the target and a labeled analog at the same time onto an immobilized antibody column [55–58]. In the sequential injection format, the sample is injected first, followed later by the injection of the labeled analog [55–59]. A displacement immunoassay involves the adsorption of a large amount of the labeled analog to an immobilized antibody column, followed by displacement of some of this labeled analog upon sample injection [55]. A reverse displacement immunoassay that instead uses an immobilized target analog and labeled antibodies or antibody fragments in IAC has also been described [60]. Simultaneous injection immunoassays have been used in the analysis of human serum albumin (HSA), IgG, theophylline, caffeine, and atrazine. Sequential injection immunoassays have also been used to measure HSA, IgG, and atrazine. Displacement immunoassays have been employed for the analysis of cocaine, benzoylecgonine, di- and trinitrotoluene, and cortisol [54–56]. Non-competitive immunoassays (or immunometric assays) can also be performed using IAC or HPIAC. There are two main types of non-competitive immunoassays: the one-site immunometric assay and the sandwich immunoassay, or the two-site immunometric assay [54–56]. A one-site immunometric assay is carried out by incubating a sample with a known excess of labeled antibodies that are specific to the target. This mixture is then applied to a column containing an immobilized analog of the target. The column is used to capture the non-bound portion of the labeled antibodies while allowing antibodies that are bound to the target to pass through non-

545

546

CHAPTER 20 Affinity chromatography

retained. Measurement of the target is performed indirectly by monitoring either the change in the response for the non-retained labeled antibodies or the amount of retained antibodies that are captured by the column and released during the elution step. This type of analysis has been used in the measurement of digoxin, thyroxine, 17β-estradiol, and HSA [55,61]. Sandwich immunoassays use two antibodies that can simultaneously bind to the same target. One of these antibodies is immobilized to the support in an IAC column and is used to extract the target from a sample. The second antibody is labeled for detection and is either mixed with the sample prior to application or applied to the column after the target is bound to the IAC column. The binding of the target by the two antibodies creates a “sandwich” complex in the column, as illustrated in Figure 20.2. After this complex has been created, an elution buffer is passed through the column to dissociate the target and labeled antibodies. Measurement of the labeled antibodies that elute with the target from the column should give a response that is directly proportional to the amount of target that was in the sample. Sandwich immunoassays based on IAC have been reported for the determination of HSA, IgG, parathyroid hormone, and human chorionic gonadotropin [54–56]. Another use of IAC columns is to monitor a target that is eluting from other types of columns. This approach is referred to as post-column immunodetection [46,55]. This technique typically involves the use of a post-column reactor and an immobilized antibody or antigen column that is attached to the outlet of an analytical HPLC column. As the target elutes from the analytical column, it is mixed with an excess of labeled antibodies that can bind to the target. The remaining, non-bound labeled antibodies are removed from this mixture by using a column that contains an immobilized analog of the target. The non-retained labeled antibodies that are already bound to the target pass through the analog column and give a signal that is proportional to the target’s concentration in the sample. This type of detection has been used for various applications, including the measurement of growth hormone-releasing factor, digoxin, digoxigenin, and human granulocyte colony-stimulating factor [47].

20.5 Dye-ligand and biomimetic affinity chromatography Synthetic dyes and chlorotriazine-linked biomimetic ligands are other ligands that can be used in affinity chromatography. These ligands have been used in numerous applications to purify enzymes and proteins. Dyes and biomimetic ligands are easy to immobilize and are inexpensive and stable, and provide stationary phases with high binding capacities. These features make such ligands of interest in large-scale or high-throughput separation techniques for the development of protein-based drugs or for the examination of protein libraries [62–66]. The method of dye-ligand affinity chromatography often uses triazine dyes to purify albumin and other blood proteins, as well as enzymes and pharmaceutical proteins [62,64,66,67]. Figure 20.3 shows some typical dye-ligands that are used in dye-ligand affinity chromatography,

20.5 Dye-ligand and biomimetic affinity chromatography

FIGURE 20.3 Structures of Cibacron Blue 3GA and Procion Blue, two common binding agents used in dyeligand affinity chromatography [64].

including Cibacron Blue 3GA and Procion Blue. These dyes have two main units that are joined through an amino bridge. The first unit is used to bind the target and often contains an anthraquinone, azo, or phthalocyanine group. The second unit usually contains a triazine ring and forms a scaffold for the binding domain and groups that can be used to attach the ligand to a support [62,65,68]. These dye ligands often have negatively charged sulfonic groups, which gives them some cation-exchange properties [64,69]. Retained proteins and targets may be dissociated from these columns by using non-specific elution; however, biospecific elution through the addition of a competing agent to the mobile phase is usually preferred [64,70]. To increase the speed and efficiency of purification based on a dye-ligand, a method known as polymer-shielded dye-affinity chromatography is sometimes employed [71,72]. In this technique, the stationary phase is treated with a watersoluble polymer to prevent non-specific interactions between proteins and the dye, as has been used in enzyme purification [59]. Other applications of dye-ligands have included their use in removing toxic or undesired components in biological fluids, such as prion proteins, human immunodeficiency virus-1, and hepatitis B particles [73–75]. There is also growing interest in the use of computational and combinatorial chemistry, as used in a method known as biomimetic affinity chromatography, to develop improved dye-based ligands and related binding agents for use in the purification of pharmaceutical proteins [62].

547

548

CHAPTER 20 Affinity chromatography

20.6 Immobilized metal-ion affinity chromatography Immobilized metal-ion affinity chromatography (IMAC) is also known as metal chelate affinity chromatography (MCAC). This method was first proposed by Porath et al. in 1975 [76] and is based on the specific interactions that can take place between immobilized metal ions and residues such as the amino acids histidine, tryptophan, or cysteine in proteins or peptides [76]. IMAC has become an important tool for the detection and purification of metalloproteins, histidine-tagged proteins, and phosphorylated proteins. Areas in which this method is now used include proteomics [77–79], work with recombinant proteins [80–82], and disease diagnosis [83,84]. The stationary phase in IMAC consists of an immobilized chelating agent that is complexed with a metal ion. The chelating agent is attached covalently to a support and used to entrap metal ions through coordinate bonds. Agarose was the first support used in IMAC, but supports based on other materials, such as silica, are also now employed in this method [85,86]. Other reports have described the development of supports for IMAC based on cryogels, silica monoliths, and polymethacrylate monoliths [87–89]. Examples of chelating ligands that can be used in IMAC are iminodiacetic acid (IDA), nitrilotriacetic acid (NTA), carboxymethylated-aspartic acid (CM-Asp), 8-hydroxyquinoline, o-phosphoserine, and N,N,N0 -tris(carboxymethyl)ethylenediamine [85,86]. IDA is the most common chelating agent employed in IMAC [85], followed by NTA [90] and CM-Asp [91] (see examples in Figure 20.4). According to the principles of Lewis acids and bases, the metal ions that are used with these

FIGURE 20.4 IMAC supports based on iminodiacetic acid (IDA) or nitrilotriacetic acid (NTA) as chelating agents for metal ions [85,91,92].

20.7 Analytical affinity chromatography

chelating groups can be divided into three categories. Hard metal ions prefer to bind to targets that contain phosphorus, aliphatic nitrogen, and oxygen. Soft metal ions prefer targets that contain sulfur. “Borderline” metal ions tend to bind targets that contain aromatic nitrogens, oxygens, and sulfur groups [85,92]. Columns containing IDA or NTA that is complexed with Ni(II), Cu(II), Zn(II), Fe(III), or Ga(III) are typically used for the isolation and purification of histidine-tagged proteins and phosphorylated proteins [78,83,84,93,94]. Columns and supports that contain Zr(IV) or Ti(IV) are useful for phosphopeptide isolation and phosphoproteome analysis [88,95].

20.7 Analytical affinity chromatography The terms analytical affinity chromatography and quantitative affinity chromatography refer to the use of an affinity column to provide information on the thermodynamics, kinetics, or mechanism of a biological interaction. One approach for this type of study is zonal elution. In this technique, a small plug of a target is injected into an affinity column that contains an immobilized ligand that binds to this target. The retention time and elution profile of the target are then used to obtain data on the interaction between the analyte and the ligand [96]. This technique was first employed by Andrews, Kitchen, and Winzor in 1973 and Dunn and Chaikin in 1974 for studying the interactions of lactate synthetase and staphylococcal nuclease [97,98]. Since then, zonal elution has been utilized to examine enzyme-inhibitor binding, protein–protein interactions, and drug-protein binding [96,99–103]. Applications of zonal elution include its use in comparing the relative affinities of a ligand for injected solutes, determining the strength of these interactions, studying the effects of changing reaction conditions (e.g., pH, solvent, and temperature) on solute-ligand binding, and determining the number and location of binding sites for a solute on a ligand [96]. Displacement and competitive binding studies are the most frequent uses of zonal elution in analytical affinity chromatography. An example of such a study is shown in Figure 20.5A [104]. In this type of application, a fixed and known concentration of a competing agent is placed into the mobile phase. The retention of an injected target is then measured which, if the target and competing agent have common binding sites, will be determined by the concentration of the competing agent, the ability of the target and competing agent to bind the ligand, and the relative amount of active ligand that is present in the column [103]. It is possible with this type of experiment to measure the site-specific equilibrium constants for the ligand with the target and determine the type of competition that is occurring between these two solutes for the ligand. Frontal analysis, or frontal affinity chromatography (FAC), is another method that can be employed in analytical affinity chromatography. In this technique, a solution containing a known concentration of a target is continuously applied to an affinity column. As the immobilized ligand becomes saturated with the applied target, a

549

550

CHAPTER 20 Affinity chromatography

FIGURE 20.5 Typical chromatograms obtained by analytical affinity chromatography on an HSA column during (A) zonal elution-competitive binding experiments, in which R-warfarin was the injected analyte and glimepiride was a mobile phase additive, and (B) frontal analysis experiments, in which glimepiride was applied as the target. The concentrations shown represent the amount of glimepiride that was applied to the column in the mobile phase. Adapted with permission from ref. Matsuda et al. [104].

breakthrough curve is obtained. This type of curve is shown in Figure 20.5B. If this experiment is done under conditions that allow a local equilibrium to be established, the moles of the target required to reach the mean position of the breakthrough curve can be related to the target’s concentration, the number of binding sites in the column, and the equilibrium constants for the target-ligand interaction [96]. This method was first used for affinity chromatography in 1975 by Kasai and Ishii [105]. A major advantage of frontal analysis over zonal elution is the former method can simultaneously provide information on both the equilibrium constants for a

20.8 Miscellaneous methods and newer developments

solute-ligand interaction and the binding capacity of the column. However, frontal analysis also usually needs a larger amount of target than zonal elution for a binding study [102]. Several other methods have been developed or modified for use in affinity chromatography to obtain information on solute-ligand interactions. Approaches that have been developed for kinetic studies include band-broadening measurements, the split-peak method, peak-fitting methods, peak decay analysis, and free fraction analysis [100–102,106–124]. These methods have been used to study the rates of drug interactions with serum proteins [121–124], the binding of lectins with sugars [100], the interaction rates between antibodies and antigens or antibody-binding proteins [100,106], and the interactions of various solutes with immobilized receptors [113,114]. Affinity chromatography has further been used for kinetic and thermodynamic studies in the area of chiral separations based on enzymes or serum proteins as stationary phases [115–117]. For example, band-broadening measurements have been used to study the kinetics of protein-based chiral stationary phases [107,117], and free fraction measurements have been employed to analyze the binding of chiral compounds with proteins in solution [110].

20.8 Miscellaneous methods and newer developments In addition to the methods already described, many other techniques have used affinity chromatography for new and improved separation or analysis methods. One example is the combined use of affinity ligands and columns with mass spectrometry in either off-line or on-line methods for work with complex samples. For instance, immunoaffinity columns have been used for sample pretreatment with gas chromatography–mass spectrometry (GC–MS) and liquid chromatography-mass spectrometry (LC–MS). In the case of GC–MS, the analytes must be derivatized after extraction to improve their stability or volatility prior to their analysis by GC. Affinity chromatography can be coupled on-line with electrospray ionization (ESI) and examined by MS as long as the elution buffer contains only volatile salts, such as ammonium acetate or ammonium formate. For work with matrix-assisted laser desorption-ionization time-of-flight mass spectrometry (MALDI-TOF MS), off-line analysis can be conducted by using an affinity column to separate sample components, which are then collected as fractions and combined with a matrix for use in MALDI. On-line analysis procedures are also possible if an affinity ligand is immobilized onto a MALDI target and used to bind specific compounds from a sample. Fast atom-bombardment mass spectrometry (FAB-MS) can also be used with offline or on-line affinity extraction [125]. Several reports have combined mass spectrometry with affinity ligands. For instance, antibody-conjugated nanoparticles have been used for the affinity extraction of plasma antigens prior to analysis by MALDI-TOF MS [126]. Magnetic nanoparticles have been used as laser desorption-ionization components and as solidphase extraction probes in MALDI-TOF MS. A nanoprobe-based assay has been

551

552

CHAPTER 20 Affinity chromatography

employed for the robotic screening of small molecules in serum, based on the weak interactions between mannose and Con A [127]. These techniques tend to be relatively simple, cost effective, and rapid and are of interest for targeted proteome profiling [128]. In addition, surface-non-covalent-affinity mass spectrometry (SNA-MS) has been used to isolate, enrich, and sequence glycosaminoglycans that bind to specific proteins [129]. In recent years, affinity columns and ligands have been incorporated into miniaturized devices. Examples include devices such as nucleic acid microarrays, protein arrays, antibody arrays, and the use of affinity ligands in microfluidic devices or microcolumns for capillary electrophoresis or liquid chromatography [130–134]. Nucleic acid-based microarrays have been useful in genomic studies, and protein arrays (e.g., antibody arrays) have been employed for proteomics. These systems have been used with various detection methods, including absorbance, fluorescence, chemiluminescence, electrochemical detection, and mass spectrometry [130]. For instance, the separation, binding, and elution of immunoglobulins on beads coated with protein G have been studied using dual-phase detection [132]. Affinity microcolumns have been developed and employed in drug-protein binding studies for personalized medicine and chromatographic immunoassays [58,61,134,135]. Another area of growing interest has been the creation of synthetic ligands for affinity methods. As an example, molecularly imprinted polymers (MIPs) have been utilized as alternatives to antibodies for use in affinity separations, binding assays, and biosensors. Binding and recognition sites are formed during the preparation of these polymers by a process like the one illustrated in Figure 20.6. During this process, functional monomers form a complex with a template, which is usually the target analyte or a related analog. Cross-linking of the monomers creates a polymer about this template. The template is then extracted, leaving behind a binding site that recognizes and retains the target when it is applied to the support. Although MIPs have been used in some applications for affinity chromatography, most of their use has involved solid-phase extraction methods [136,137]. MIPs have recently been employed in microanalytical devices [131], such as for the solid-phase extraction of riboflavin from food [138] and the extraction of chlorotriazine herbicides plus their metabolites from environmental samples [139]. Another type of alternative affinity ligand is an aptamer. Aptamers are nucleic acid ligands that have been screened to selectively bind a specific target. Aptamers are generated by separating oligonucleotides that bind to the desired target from a large, random pool of single-stranded DNA or RNA. These oligonucleotides are selected and enriched by a process called SELEX (i.e., the systematic evolution of ligands by exponential enrichment) [62]. The SELEX process begins by immobilizing a protein or the desired target within a column or on a support. A random library of oligonucleotides is applied to the column or support, and the oligonucleotides that are bound by the target are later eluted and amplified. Additional rounds of selection and enrichment are carried out with fresh columns until an aptamer with the desired selectivity and binding strength is obtained. Aptamers have been used in

20.8 Miscellaneous methods and newer developments

FIGURE 20.6 General process for the preparation of a molecularly imprinted polymer [135].

affinity chromatography to purify a recombinant human L-selectin-immunoglobulin fusion protein from Chinese hamster ovary cells [140] and have been considered for use in pharmacology [141], chemical analysis [142,143], and cell biology [62]. In addition, aptamers have been used in microfluidic chips for affinity extraction, separation, and detection. Microfluidic chips have been further used to study aptamertarget interactions and for aptamer selection [144]. Aptamers have also been immobilized within monoliths [145,146] and used in biosensors with modified nanoparticles [147].

553

554

CHAPTER 20 Affinity chromatography

Acknowledgment This work was supported, in part, by the National Institutes of Health under grant R01 DK069629 and the National Science Foundation (NSF) under grant CHE 2108881.

References [1] Hage DS, Ruhn PF. An introduction to affinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 3–13. [2] Turkova J. Affinity chromatography. Amsterdam: Elsevier; 1978. [3] Walters RR. Affinity chromatography. Anal Chem 1985;57:1099A–114A. [4] Rodriguez EL, Poddar S, Iftekhar S, Suh K, Woolfork A, Ovbude S, et al. Affinity chromatography: a review of trends and developments over the past 50 years. J Chromatogr B 2020;1157, 122332. [5] Cuatrecasas P, Wilchek M, Anfinsen CB. Selective enzyme purification by affinity chromatography. Proc Natl Acad Sci U S A 1968;68:636–43. [6] Hage DS, Xuan H, Nelson MA. Application and elution in affinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 79–97. [7] Zhang C, Rodriguez E, Bi C, Zheng X, Suresh D, Suh K, et al. High performance affinity chromatography and related separation methods for the analysis of biological and pharmaceutical agents. Analyst 2018;143:374–91. [8] Ohlson S, Lundblad A, Zopf D. Novel approach to affinity chromatography using “weak” monoclonal antibodies. Anal Biochem 1988;169:204–8. [9] Ohlson S, Bergstroem M, Pahlsson P, Lundblad A. Use of monoclonal antibodies for weak affinity chromatography. J Chromatogr A 1997;758:199–208. [10] Strandh M, Andersson HS, Ohlson S. Weak affinity chromatography. Methods Mol Biol 2000;147:7–23. [11] Vosseller K, Trinidad JC, Chalkley RJ, Specht CG, Thalhammer A, Lynn AJ, et al. O-linked N-acetylglucosamine proteomics of postsynaptic density preparations using lectin weak affinity chromatography and mass spectrometry. Mol Cell Proteomics 2006;5:923–34. [12] Duong-Thi M-D, Meiby E, Bergstroem M, Fex T, Isaksson R, Olson S. Weak affinity chromatography as a new approach for fragment screening in drug discovery. Anal Biochem 2011;414:138–46. [13] Gustavsson P-E, Larsson P-O. Support materials for affinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 16–32. [14] Mallik R, Jiang T, Hage DS. High-performance affinity monolith chromatography: development and evaluation of human serum albumin columns. Anal Chem 2004;76:7013–22. [15] Mallik R, Hage DS. Development of an affinity silica monolith containing human serum albumin for chiral separations. J Pharm Biomed Anal 2008;46:820–30. [16] Noppe WP, Plieva FM, Vanhoorelbeke K, Deckmyn H, Tuncel M, Tuncel A, et al. Macroporous monolithic gels, cryogels, with immobilized phages from phage-display

References

[17]

[18] [19] [20]

[21] [22] [23]

[24]

[25] [26] [27]

[28]

[29] [30] [31] [32] [33]

[34] [35]

[36]

library as a new platform for fast development of affinity adsorbent capable of target capture from crude feeds. J Biotechnol 2007;131:293–9. Pfaunmiller EL, Paulemond ML, Dupper CM, Hage DS. Affinity monolith chromatography: a review of principles and recent analytical applications. Anal Bioanal Chem 2013;405:2133–45. Li Z, Rodriguez E, Azaria S, Pekarek A, Hage DS. Affinity monolith chromatography: a review of general principles and applications. Electrophoresis 2017;38:2837–50. Mallik R, Hage DS. Affinity monolith chromatography. J Sep Sci 2006;29:1686–704. Kim HS, Hage DS. Immobilization methods for affinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 35–78. Chibata I. Immobilized enzymes. New York: Wiley; 1978. Messing RA. Adsorption and inorganic bridge formations. Methods Enzymol 1976;44:148–69. Raducanu VS, Tehseen M, Shirbini A, Raducanu DV, Hamdan SM. Two chromatographic schemes for protein purification involving the biotin/avidin interaction under native conditions. J Chromatogr A 2020;1621, 461051. Gupalova TVI, Palagnuk VG, Totolian AA, Tennikova TB. Quantitative investigation of the affinity properties of different recombinant forms of protein G by means of highperformance monolithic chromatography. J Chromatogr A 2002;949:185–93. Jackson A, Karle EM, Hage DS. Preparation of high capacity supports containing protein G immobilized to porous silica. Anal Biochem 2010;406:235–7. Jackson A, Xuan H, Hage DS. Entrapment of proteins in glycogen-capped and hydrazide-activated supports. Anal Biochem 2010;404:106–8. Vargas-Badilla J, Poddar S, Azaria S, Zhang C, Hage DS. Optimization of protein entrapment in affinity microcolumns using hydrazide-activated silica and glycogen as a capping agent. J Chromatogr B 2019;1121:1–8. Hage DS, Bian M, Burks R, Karle E, Ohnmacht C, Wa C. Bioaffinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 101–26. Hage DS, Matsuda R. Affinity chromatography—a historical perspective. In: Reichelt S, editor. Affinity chromatography. Methods Mol Biol 2015;1286:1–19. Starkenstein E. Fermentation and the influence upon it of neutral salts. Biochem Z 1910;24:210–8. Hermanson GT, Mallia AK, Smith K. Immobilized affinity ligand techniques. New York: Academic Press; 1992. Alley WR, Mann BF, Novotny MV. High-sensitivity analytical approaches for the structural characterization of glycoproteins. Chem Rev 2013;113:2668–732. Friedberg F, Rhoads AR. Affinity chromatography of enzymes. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, EL: CRC Press; 2006. p. 313–46. Parikh I, Cuatrecasas P. Affinity chromatography. Chem Eng News 1985;63:17–32. Bourne Y, Astoul CH, Zamboni V, Peumans WJ, Menu-Bouaouiche I, Van Damme EJM, et al. Structural basis for the unusual carbohydrate-binding specificity of jacalin towards galactose and mannose. J Biochem 2002;364:173–80. Kullolli M, Hancock WS, Hincapie M. Automated platform for fractionation of human plasma glycoproteome in clinical proteomics. Anal Chem 2010;82:115–20.

555

556

CHAPTER 20 Affinity chromatography

[37] Kullolli M, Hancock WS, Hincapie M. Preparation of a high-performance multi-lectin affinity chromatography (HP-M-LAC) adsorbent for the analysis of human plasma glycoproteins. J Sep Sci 2008;31:2733–9. [38] Lindmark R, Biriell C, Sjoequist J. Quantitation of specific IgG antibodies in rabbits by a solid-phase radioimmunoassay with 125I-protein A from Staphylococcus aureus. Scand J Immunol 1981;14:409–20. [39] Ey PL, Prowse SJ, Jenkin CR. Isolation of pure IgG, IgG2a and IgG2b immunoglobulins from mouse serum using protein A-Sepharose. Immunochemistry 1978;15:429–36. [40] Bjorck L, Kronvall G. Purification and some properties of streptococcal protein G, a novel IgG-binding reagent. J Immunol 1984;133:969–74. [41] Akerstrom B, Bjorck L. A physiochemical study of protein G, a molecule with unique immunoglobulin G-binding properties. J Biol Chem 1986;261:10240–7. [42] Moxley RA, Oak S, Gadgil H, Jarrett HW. DNA affinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 173–214. [43] Potuzak H, Dean PDG. Affinity chromatography on columns containing nucleic acids. FEBS Lett 1978;88:161–6. [44] Alberts B, Johnson B, Lewis J, Raff M, Roberts K, Walter P. Molecular biology of the cell. 4th ed. New York: Garland Science; 2002. [45] Carlton GJ. Immunosorbent separations. Methods Enzymol 1985;104:381–7. [46] Hage DS. Survey of recent advances in analytical applications of immunoaftinity chromatography. J Chromatogr B 1998;715:3–28. [47] Hage DS, Phillips TM. Immunoaffinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 127–72. [48] Jackson AJ, Sobansky MR, Hage DS. Principles and applications of immunoaffinity chromatography. In: Meulenberg EP, editor. Antibodies. Netherlands: Bentham; 2012 [Chapter 10]. [49] Ehle H, Horn A. Immunoaffinity chromatography of enzymes. Bioseparation 1990;1:97–110. [50] Weller MG. Immunochromatographic techniques—a critical review. Fresenius J Anal Chem 2000;366:635–45. [51] Nakajima M, Yamaguchi I. Purification of plant hormones by immunoaffinity chromatography. Kagaku to Seibutsu 1991;29:270–5. [52] Gallant SR. Immunoaffinity chromatography of proteins. Methods Mol Biol 2004;251:103–9. [53] Zolotarjova N, Boyes B, Martosella J. Immunoaffinity depletion of high-abundant proteins for proteomic sample preparation. In: Smejkal GB, Lazarev A, editors. Separation methods in proteomics. Boca Raton, FL: CRC Press; 2006. p. 63–79. [54] Moser AC, Hage DS. Chromatographic immunoassays. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 789–836. [55] Moser AC, Hage DS. Immunoaffinity chromatography: an introduction to applications and recent developments. Bioanalysis 2010;2:769–90. [56] Matsuda R, Rodriguez E, Suresh D, Hage DS. Chromatographic immunoassays: strategies and recent developments in the analysis of drugs and biological agents. Bioanalysis 2015;7:2947–66. [57] Hage DS, Thomas DH, Chowdhuri AR, Clarke W. Development of a theoretical model for chromatographic-based competitive binding immunoassays with simultaneous injection of sample and label. Anal Chem 1999;71:2965–75.

References

[58] Pfaunmiller EL, Anguizola JA, Milanuk ML, Carter N, Hage DS. Use of protein G microcolumns in chromatographic immunoassays: a comparison of competitive binding formats. J Chromatogr B 2016;1021:91–100. [59] Hage DS, Thomas DH, Beck MS. Theory of a sequential addition competitive binding immunoassay based on high-performance immunoaffinity chromatography. Anal Chem 1993;65:1622–30. [60] Schiel JE, Tong Z, Sakulthaew C, Hage DS. Development of a flow-based ultrafast immunoextraction and reverse displacement immunoassay: analysis of free drug fractions. Anal Chem 2011;83:9384–90. [61] Pfaunmiller EL, Anguizola TA, Milanuk ML, Papastavros E, Carter N, Matsuda R, et al. Development of microcolumn-based one-site immunometric assays for protein biomarkers. J Chromatogr A 2014;1366:92–100. [62] Labrou NE, Mazitsos K, Clonis YD. Dye-ligand and biomimetic affinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 231–55. [63] Lowe CR. Combinatorial approaches to affinity chromatography. Curr Opin Chem Biol 2001;5:248–56. [64] Labrou NE, Clonis YD. Immobilised synthetic dyes in affinity chromatography. In: Vijayalakshmi MA, editor. Theory and practice of biochromatography. London: Taylor & Francis; 2002. p. 335–51. [65] Labrou NE, Clonis YD. The affinity technology in downstream processing. J Biotechnol 1994;36:95–119. [66] Clonis YD, Labrou NE, Kotsira V, Mazitsos K, Melissis S, Gogolas G. Biomimetic dyes as affinity chromatography tools in enzyme purification. J Chromatogr A 2000;891:33–44. [67] Janson GC, editor. Protein purification: principles, high resolution methods, and applications. 3rd ed. Hoboken, NJ: Wiley; 2011. [68] Denizli A, Piskin E. Dye-ligand affinity systems. J Biochem Biophys Methods 2001;49:391–416. [69] Labrou NE. Dye-ligand affinity chromatography for protein separation and purification. In: Walker JM, editor. Methods in molecular biology, affinity chromatography: methods and protocols. Totowa, NJ: Humana Press; 2000. p. 129–39. [70] Firer MA. Efficient elution of functional proteins in affinity chromatography. J Biochem Biophys Methods 2001;49:433–42. [71] Mattiasson B, Galaev IY, Garg N. Polymer-shielded dye affinity chromatography. J Mol Recognit 1996;9:509–14. [72] Garg N, Galaev IY, Mattiasson B. Polymer-shielded dye affinity chromatography of lactate dehydrogenase from porcine muscle in an expanded bed system. Bioseparation 1996;6:193–9. [73] Caspi S, Halimi M, Yanai A, Sasson SB, Taraboulos A, Gabizon R. The anti-prion activity of Congo Red. Putative mechanism. J Biol Chem 1998;73:3484–9. [74] Hattori T, Zhang X, Weiss C, Xu Y, Kubo T, Sato Y, et al. Triazine dyes inhibit HIV-1 entry by binding to envelope glycoproteins. Microbiol Immunol 1997;41:717–24. [75] Brown RA, Combridge BS. Binding of hepatitis virus particles to immobilized Procion Blue-HB and Cibacron Blue 3GA. J Virol Methods 1986;14:267–74. [76] Porath J, Carlsson J, Olsson I, Belfrage G. Metal chelate affinity chromatography: a new approach to protein fraction. Nature 1975;258:598–9. [77] McNulty DE, Annan RS. Hydrophilic interaction chromatography for fractionation and enrichment of the phosphoproteome. Methods Mol Biol 2009;527:93–105.

557

558

CHAPTER 20 Affinity chromatography

[78] Tsai C-F, Wang Y-T, Chen Y-R, Lai C-Y, Lin P-Y, Pan K-T, et al. Immobilized metal affinity chromatography revisited: pH/acid control toward high selectivity in phosphoproteomics. J Proteome Res 2008;7:4058–69. [79] Wilson-Grady JT, Villen J, Gygi SP. Phosphoproteome analysis of fission yeast. J Proteome Res 2008;7:1088–97. [80] Dong X-Y, Chen L-J, Sun Y. Refolding and purification of histidine-tagged protein by artificial chaperone-assisted metal affinity chromatography. J Chromatogr A 2009;1216:5207–13. [81] Wang C-Z, Wang L-L, Geng X-D. Optimization of refolding with simultaneous purification of recombinant human granulocyte colony-stimulating factor from Escherichia coil by immobilized metal ion affinity chromatography. Biochem Eng J 2009;43:197–202. [82] Yin SM, Zheng Y, Tien P. On-column purification and refolding of recombinant bovine prion protein: using its octarepeat sequences as a natural affinity tag. Protein Expr Purif 2003;32:104–9. [83] Callesen AK, Vach W, Jorgensen PE, Cold S, Tan QH, Rene DPC, et al. Combined experimental and statistical strategy for mass spectrometry based serum protein profiling for diagnosis of breast cancer: a case-control study. J Proteome Res 2008;7: 1419–26. [84] Lee YM, Venkataraman K, Hwang S, Han DK, Hla T. A novel method to quantify sphingosine 1-phosphate by immobilized metal affinity chromatography (IMAC). Prostaglandins Other Lipid Mediat 2007;84:154–62. [85] Gutierrez R, Martin del Valle EM, Galan MA. Immobilized metal-ion affinity chromatography: status and trends. Sep Purif Rev 2007;36:71–111. [86] Prasanna RR, Vijayalakshmi MA. Immobilized metal-ion affinity systems for recovery and structure—function studies of proteins at molecular, supramolecular, and cellular levels. Pure Appl Chem 2010;82:39–55. [87] Efremenko E, Votchitseva Y, Plieva F, Galaev I, Mattiasson B. Purification of His6— organophosphate hydrolase using monolithic supermacroporous polyacrylamide cryogels developed for immobilized metal affinity chromatography. Appl Microbiol Biotechnol 2006;70:558–63. [88] Hou C-Y, Ma J-F, Tao D-Y, Shan Y-C, Liang Z, Zhang L-H, et al. Organic—inorganic hybrid silica monolith based immobilized titanium ion affinity chromatography column for analysis of mitochondrial phosphoproteome. J Proteome Res 2010;9:4093–101. [89] Peterka M, Jarc M, Banjac M, Frankovic V, Bencina K, Merhar M, et al. Characterisation of metal-chelate methacrylate monoliths. J Chromatogr A 2006;1109:80–5. [90] Hochuli E, Bannwarth W, Dobeli H, Gentz R, Stuber D. Genetic approach to faciltate purification of recombinant proteins with a novel metal chelate adsorbent. Nat Biotechnol 1988;6:1321–5. [91] Arnold FH. Metal-affinity separations—a new dimension in protein processing. Biotechnology 1991;9:151–6. [92] Ueda EKM, Gout PW, Morganti L. Current and prospective applications of metal ionprotein binding. J Chromatogr A 2003;988:1–23. [93] Vorackova I, Suchanova S, Ulbrich P, Diehl WF, Ruml T. Purification of proteins containing zinc finger domains using immobilized metal ion affinity chromatography. Protein Expr Purif 2011;79:88–95. [94] Ye J-Y, Zhang X-M, Young C, Zhao X-L, Hao Q, Cheng L, et al. Optimized IMAC protocol for phosphopeptide recovery from complex biological samples. J Proteome Res 2010;9:3561–73.

References

[95] Zou H-F, Ye M-L, Jiang X-N, Han G-G, Wang F-J. Development of technology and methods for large-scale phosphoproteome analysis. In: Pacifichem 2010; Honolulu, HI: International chemical congress of pacific basin societies; 2010. December 15–20; 2010. ANYL-482. [96] Hage DS, Chen J. Quantitative affinity chromatography: practical aspects. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 596–628. [97] Andrews P, Kitchen BJ, Winzor D. Use of affinity chromatography for the quantitative study of acceptor-ligand interactions: the lactose synthetase system. Biochem J 1973;135:897–900. [98] Dunn BM, Chaiken IM. Quantitative affinity chromatography. Determination of binding constants by elution with competitive inhibitors. Proc Natl Acad Sci U S A 1974;71:2382–5. [99] Winsor DJ. Quantitative affinity chromatography: Recent theoretical developments. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 630–62. [100] Chaiken IM, editor. Analytical affinity chromatography. Boca Raton, FL: CRC Press; 1987. [101] Bertucci C, Pistolozzi M, Felix G, Danielson UH. HSA binding of HIV protease inhibitors: a high-performance affinity chromatography study. J Sep Sci 2009;32:1625–31. [102] Schiel JE, Joseph KS, Hage DS. Biointeraction affinity chromatography. Adv Chromatogr 2009;48:145–93. [103] Hage DS, Jackson A, Sobansky MR, Schiel JE, Yoo MJ, Joseph KS. Characterization of drug-protein interactions in blood using high-performance affinity chromatography. J Sep Sci 2009;32:835–53. [104] Matsuda R, Li Z, Zheng X, Hage DS. Analysis of multi-site drug-protein interactions by high-performance affinity chromatography: binding by glimepiride to normal or glycated human serum albumin. J Chromatogr A 2015;1408:133–44. [105] Kasai K, Ishii S. Quantitative analysis of affinity chromatography of trypsin. A new technique for investigation of protein-ligand interaction. J Biochem 1975;77:261–4. [106] Hage DS, Walters RR, Hethcote HW. Split-peak affinity chromatographic studies of the immobiilization-dependent adsorption kinetics of protein a. Anal Chem 1986;58:274–9. [107] Yang J, Hage DS. Effect of mobile phase composition on the binding kinetics of chiral solutes on a protein-based HPLC column: interactions of D- and L-tryptophan with immobilized human serum albumin. J Chromatogr A 1997;766:15–25. [108] Denizot FC, Delaage MS. Statistical theory of chromatography: new outlooks for affinity chromatography. Proc Nati Acad Sci USA 1975;72:4840–3. [109] Talbert AM, Tranter GE, Holmes E, Francis PI. Determination of drug-plasma protein binding kinetics and equilibria by chromatographic profiling: exemplification of the method using L-tryptophan and albumin. Anal Chem 2002;74:446–52. [110] Clarke W, Schiel JE, Moser A, Hage DS. Analysis of free hormone fractions by an ultrafast immunoextraction/displacement immunoassay: studies using free thyroxine as a model system. Anal Chem 2005;77:1856–66. [111] Ohnmacht CM, Schiel JE, Hage DS. Analysis of free drug fractions using near infrared fluorescent labels and an ultrafast immunoextraction/displacement assay. Anal Chem 2006;78:7547–56. [112] Mallik R, Yoo MJ, Brisoe CJ, Hage DS. Analysis of drug-protein binding by ultrafast affinity chromatography using immobilized human serum albumin. J Chromatogr A 2010;1217:2796–803.

559

560

CHAPTER 20 Affinity chromatography

[113] Moaddel R, Wainer IW. Conformational mobility of immobilized proteins. J Pharm Biomed Anal 2007;43:399–406. [114] Jozwiak K, Haginaka J, Moaddel R, Wainer IW. Displacement and nonlinear chromatographic techniques in the investigation of noncompetitive inhibitors with an immobilized alpha3beta4 nicotinic acetylcholine receptor liquid chromatographic stationary phase. Anal Chem 2002;74:4618–24. [115] Patel S, Wainer IW, Lough WJ. Affinity-based chiral stationary phases. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 571–92. [116] Patel S, Wainer IW, Lough WJ. Chromatographic studies of molecular recognition and solute binding to enzymes and plasma proteins. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 663–83. [117] Loun B, Hage DS. Chiral separation mechanisms in protein-based HPLC columns, II. Kinetic studies of R- and S-warfarin binding to immobilized human serum albumin. Anal Chem 1996;68:1218–25. [118] Iftekhar S, Ovbude ST, Hage DS. Kinetic analysis by affinity chromatography. Front Chem 2019;7:673. [119] Schiel JE, Ohnmacht CM, Hage DS. Measurement of drug-protein dissociation rates by high-performance affinity chromatography and peak profiling. Anal Chem 2009;81:4320–33. [120] Zheng X, Bi C, Li Z, Podariu M, Hage DS. Analytical methods for kinetic studies of biological interactions: a review. J Pharm Biomed Anal 2015;113:163–80. [121] Tong Z, Schiel JE, Papastavros E, Ohnmacht CM, Smith QR, Hage DS. Kinetic studies of drug-protein interactions by using peak profiling and high-performance affinity chromatography: examination of multi-site interactions of drugs with human serum albumin columns. J Chromatogr A 2011;1218:2065–71. [122] Moore RM, Walters RR. Peak-decay method for the measurement of dissociation rate constants by high-performance affinity chromatography. J Chromatogr A 1987;384: 91–103. [123] Chen J, Schiel JE, Hage DS. Noncompetitive peak decay analysis of drug-protein dissociation by high-performance affinity chromatography. J Sep Sci 2009;32:1632–41. [124] Yoo MJ, Hage DS. Use of peak decay analysis and affinity microcolumns containing silica monoliths for rapid determination of drug-protein dissociation rates. J Chromatogr A 2011;1218:2072–8. [125] Briscoe CJ, Clarke W, Hage DS. Affinity mass spectrometry. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 737–61. [126] Lin P-C, Chou P-H, Chen S-H, Liao H-K, Wang K-Y, Chen Y-J, et al. Ethylene glycolprotected magnetic nanoparticles for a multiplexed immunoassay in human plasma. Small 2006;2:485–9. [127] Lin P-C, Tseng M-C, Su A-K, Chen Y-T, Lin C-C. Functionalized magnetic nanoparticles for small-molecule isolation, identification, and quantification. Anal Chem 2007;79:3401–8. [128] Wei C, Yamato M, Wei W, Zhao X, Tsumoto K, Yoshimura T, et al. Genetic nanomedicine and tissue engineering. Med Clin N Am 2007;91:889–98. [129] Prabhakar V, Capila I, Sasiseldiaran R. Glycosaminoglycan characterization methodologies: probing biomolecular interactions. Methods Mol Biol 2009;534:331–40.

References

[130] Phillips TM. Microanalytical methods based on affinity chromatography. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 763–87. [131] Miro M, Oliveira HM, Sgundo MA. Analytical potential of microfluidic lab-on-a-valve as front end to column-separation systems. Trends Anal Chem 2011;30:153–64. [132] Gutzman Y, Carroll AD, Ruzicka J. Bead injection for biomolecular assays: affinity chromatography enhanced by bead injection spectroscopy. Analyst 2006;131:809–15. [133] Decuir M, Laehdesmaeki I, Carroll AD, Ruzicka J. Automated capture and on-column detection of biotinylated DNA on a disposable solid support. Analyst 2007;132:818–22. [134] Zheng X, Li Z, Beeram S, Matsuda R, Pfaunmiller EL, Podariu M, et al. Analysis of biomolecular interactions using affinity microcolumns: a review. J Chromatogr B 2014;968:49–63. [135] Hage DS. Analysis of biological interactions by affinity chromatography: clinical and pharmaceutical applications. Clin Chem 2017;63:1083–93. [136] Haupt K. Molecularly imprinted polymers: Artificial receptors for affinity separations. In: Hage DS, editor. Handbook of affinity chromatography. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 837–56. [137] Haginaka J. Monodispersed, molecularly imprinted polymers as affinity-based chromatography media. J Chromatogr B 2008;866:3–13. [138] Oliveira HM, Segundo MA, Lima JLFC, Miro M, Cerda V. Exploiting automatic online renewable molecularly imprinted solid-phase extraction in lab-on-valve format as front end to liquid chromatography: application to the determination of riboflavin in foodstuffs. Anal Bioanal Chem 2010;3:77–86. [139] Boonjob W, Yu Y, Miro M, Segundo MA, Wang J, Cerda V. Online hyphenation of multimodal microsolid phase extraction involving renewable molecularly imprinted and reversed-phase sorbents to liquid chromatography for automatic multiresidue assays. Anal Chem 2010;82:3052–60. [140] Romig TS, Bell C, Drolet DW. Aptamer affinity chromatography: combinatorial chemistry applied to protein purification. J Chromatogr B 1999;731:275–84. [141] Bock LC, Griffin LC, Latham JA, Vermaas EH, Toole JJ. Selection of single-stranded DNA molecules that bind and inhibit human thrombin. Nature 1992;355:564–6. [142] German I, Buchanan DD, Kennedy RT. Aptamers as ligands in affinity probe capillary electrophoresis. Anal Chem 1998;70:4540–5. [143] Potyrailo RA, Conrad RC, Ellington AD, Hieftje GM. Adapting selected nucleic acid ligands (aptamers) to biosensors. Anal Chem 1998;70:3419–25. [144] Xu Y, Yang X, Wang E. Review: aptamers in microfluidic chips. Anal Chim Acta 2010;683:12–20. [145] Zhao Q, Li X-F, Shao Y, Le XC. Aptamer-based affinity chromatographic assays for thrombin. Anal Chem 2008;80:7586–93. [146] Zhao Q, Li X-F, Le XC. Aptamer-modified monolithic capillary chromatography for protein separation and detection. Anal Chem 2008;80:3915–20. [147] Mairal T, Oezalp VC, Sanchez PL, Mir M, Katakis I, O’Sullivan CK. Aptamers: molecular tools for analytical applications. Anal Bioanal Chem 2008;390:989–1007.

561

CHAPTER

Multidimensional liquid chromatography

21

Francesco Cacciolaa, Katia Arenab, Paola Dugob,c, and Luigi Mondellob,c,d a

Department of Biomedical, Dental, Morphological and Functional Imaging Sciences, University of Messina, Messina, Italy, bDepartment of Chemical, Biological, Pharmaceutical and Environmental Sciences, University of Messina, Messina, Italy, cChromaleont s.r.l., c/o Department of Chemical, Biological, Pharmaceutical and Environmental Sciences, University of Messina, Messina, Italy, d Department of Sciences and Technologies for Human and Environment, University Campus Bio-Medico of Rome, Rome, Italy

21.1 Introduction One-dimensional liquid chromatography (1D-LC) is widely applied to the analysis of real-world samples in several fields. However, for the analysis of many real-world samples, for example, biological, food, and environmental samples, “conventional” 1D-LC methods do not provide rewarding results for enabling compound identification and quantification. Multidimensional liquid chromatography (MD-LC) has emerged as an interesting alternative for the analysis of complex samples. Peak capacity (nc) enhancement achievable by MD-LC is by far higher than by any improved 1D-LC method. MD-LC implies the combination of two or more independent or nearly independent separation steps. Although there is no inherent restriction to the number of independent separation methods used in MD-LC separations, practical constraints have limited the vast majority of MD-LC separations reported to date to two dimensions. From an historical point of view, introduced over 40 years ago, a switching valve to couple LC columns was experienced to selectively transfer the target components from the first (1D) to the second dimension (2D) [1]. The most notable development in this area has been the “continuous” approach, which led to comprehensive twodimensional liquid chromatography (LC  LC) systems (introduced for the first time in 1978 by Erni and Frei) [2], where the whole sample (a Senna extract) was analyzed on-line by size exclusion chromatography  reversed-phase liquid chromatography (SEC  RP-LC) system. Although the common coupled-column LC system can be successfully used to isolate and resolve selected regions of the sample, the LC  LC mode is capable of giving an overall overview of the sample. Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00014-X Copyright # 2023 Elsevier Inc. All rights reserved.

563

564

CHAPTER 21 Multidimensional liquid chromatography

The most general set-up of an MD-LC system consists of two pumps, two columns, injector, an interface (a modulator or sampling device), and a detector. The potential of MD-LC techniques can be further increased by means of on-line coupling of different detection systems, for example, photo diode array (PDA) and MS, thus greatly enhancing the identification power. MD-LC techniques can be performed in either off-line or on-line modes. The combination of some LC modes may present a series of well-known issues, for example, mobile phase immiscibility and/or incompatibility of the mobile phases used in the 1D and 2D as experienced for normal-phase  reversed-phase (NP  RP) or hydrophilic interaction liquid chromatography  reversed-phase (HILIC  RP). From a practical point of view, MD-LC techniques can be divided into four groups of techniques: off-line heart-cutting (LC//LC), off-line comprehensive LC (off-line LC  LC), on-line heart-cutting (LC–LC), and on-line comprehensive LC (on-line LC  LC). In LC//LC and off-line LC  LC, one or more fractions eluting from the 1D are collected manually, evaporated, and injected into the 2D. The results of these analyses are a series of 2D separations equal to the number of fractions collected from the 1D. Main advantages are the ease of being carried out without the need of special interfaces or switching valves and the reduction of solvent incompatibility between the two dimensions allowing the coupling of a great variety of different LC modes. The off-line approach is labor-intensive and time-consuming, and could be affected by sample contamination or artifact formation during long treatment. This technique is mainly used when only parts of the 1D require a separation in the 2D. The previous knowledge of the retention of specific sample components, before the fractionation can take place, is normally required, and it is commonly used for the separation of no. components in a sample. In LC–LC and on-line LC  LC, the interfaces used should be capable to automatically transfer the 1D fractions into the 2D. The on-line approach is characterized by the following advantages if compared to the off-line mode: it is quite easy to automate thus offering improved sample throughput and minimizing sample loss or contamination (due to shorter sample treatment) and overall analysis time. Furthermore, it is more reproducible, and there is a great potential for the identification of “unknowns” (due to the formation of group-type patterns on the two-dimensional plane). However, this technique presents a number of challenges, for example, the need of automated or semiautomated systems, need of specific interfaces, and immiscibility/incompatibility of different mobile phases. LC  LC is a powerful technique, frequently used also for sample clean-up, because it can be used for efficient matrix elimination and hence is characterized by less time-consuming sample treatment steps than in off-line approaches. Further, transfer systems and chromatographic separation modes could be conveniently changed during the optimization process. The goals in applying LC  LC can be to increase resolution, selectivity, and sensitivity, to enrich trace amounts of the sample, to protect sensitive detectors, or to speed up column equilibration.

21.2 Fundamentals

The knowledge of the sample composition ahead is always beneficial when using such technique, and then careful method optimization and several related practical aspects should be considered [3–5]. Due to the tremendously higher separation power of MD-LC than “conventional” 1D-LC, the technique has been employed in several fields, for example, polymers, pharmaceuticals and biological mixtures, natural products, and environmental and petrochemical samples, and has been the subject of various reviews [3–34].

21.2 Fundamentals MD-LC methods improve the likelihood for separation of a higher number of sample components; the nc of an LC  LC separation is multiplicative of the product of the peak capacity values of the two single dimensions (n2D ¼ nD1  nD2) [35–37]. The nc values of the two dimensions are shown as the number of adjacent Gaussian profiles that can be accommodated into the space along with the respective separation coordinates. The resolution is represented by rectangular boxes, which divide the separation plane. The n2D is therefore approximately equal to the number of such boxes [37]. In the last two decades, many works have dealt with all key factors affecting resolution and performance in MD-LC, especially in LC  LC methods. An important role in the dependence of the two dimensions on each other is related to the analysis speed in the 2D affecting the 1D sampling rate on the peak capacity of the overall LC  LC separations. In 1998, Murphy et al. investigated the effect of the sampling rate on the effective 1D peak width and modeled a Gaussian peak as a histogram profile of the average concentration within every sampling period [38]. The outcome of the research was that approximately three to four modulations per each 1D peak (8σ) are needed in order to avoid serious loss of performance of the LC  LC system as a result of “undersampling” of 1D peaks. Some years later, Seeley et al. published a similar work on this topic investigating the sampling frequency for modulators with various duty cycles (2002) introducing the (average) peak broadening factor, σ*, related to selected parameters such as sampling period, τΖ, and sampling phase, Φ [39]. τΖ is determined by the modulation period or sampling time ts and 1σ of the 1D peak, while Φ represents the way in which the primary peak is cut into aliquots. The sampling phase Φ is the time difference between the center of the sampling cycle nearest to the peak maximum, and the peak maximum itself, divided by the sampling period. Peak broadening was less significant when the peak maximum was centered in one of the sampling periods and for low duty cycles. For systems with a duty cycle of 1, as is the case for most LC  LC separations, the peak broadening was practically independent of Φ. Basically, Seeley drew the same conclusions as Murphy, extending his investigations to duty cycles less than 1. Afterward, Horie et al. (2007) demonstrated that modulation periods equal to 2.2–4 are sufficient to minimize “peak broadening”. In addition, they introduced a new parameter, eM (Nobuo factor), which is defined as

565

566

CHAPTER 21 Multidimensional liquid chromatography

the ratio of the apparent peak capacity of the 1D derived from σ* to its actual peak capacity in the 1D [40]. In 2006, Schoenmakers and co-workers introduced a useful “protocol” for optimizing LC  LC separations based on proper sampling, taking into account suitable chromatographic parameters such as column dimensions, injection volumes, and flow rates [41]. Carr and co-workers put great efforts toward understanding LC  LC by extending their investigations to the overall chromatogram [42–47]. For the determination of the average 1D broadening factor, β, as a function of the sampling time (ts) and the standard deviation of the 1D peaks prior to 1σ, the 2D statistical overlap theory along with the number of observed peaks in simulated LC  LC separations was used. Moreover, to better evaluate possible peak clusters along the LC  LC analysis and, thus, to estimate LC  LC space coverage, an orthogonality degree (AO) was considered to offer the denominated “corrected” (also known as effective) peak capacity. Among the different approaches reported so far for the evaluation of the orthogonality degree of a two-dimensional set-up, a valuable method was proposed by Camenzuli and Schoenmakers [48] in 2014. Such a procedure takes into account the spread of each peak along the four imaginary lines that cross the LC  LC, forming an asterisk, Z1, Z2 (vertical and horizontal lines) and Z, Z+ (diagonal lines of the asterisk). Z parameters describe the use of the separation space with respect to the corresponding Z line, allowing to semiquantitatively diagnose areas of the separation space, where sample components are clustered, reducing in practice orthogonality. For the determination of each Z parameter, the SZx value is calculated as the measure of spreading around the Zx line using the retention times of all the separated peaks in each LC  LC analysis. In terms of resolution, an established metric for LC  LC separations was proposed by Peters et al. [49] to calculate a representative resolution value and the separation quality of an LC  LC set-up. Such a measure is based on the valley-to-peak ratio between two neighbor peaks. To establish the valley-to-peak ratio between two peaks (peak 1 and peak 2), three maximum intensities are considered: the maximum of the peak 1 (max1), the saddle point between both peaks (S) and the maximum of peak 2 (max2), as well as the distances between max1 and S, d1,S, and the distance between S and max2, dS,2.

21.3 Instrumental set-up and data analysis In the most common set-up for MD-LC systems, small volume fractions of the effluent from the 1D are transferred, via an interface into the 2D. An alternative to sampling loops is the use of either packed loops (trapping columns) [50] or two parallel analytical 2D columns [51]. Significant implementations have been carried out in the last 8 years to improve the performance of the MD-LC or LC  LC methodology. First, implementations have been recently carried out in Schoenmakers and Stoll’s research groups. They first proposed a so-called “actively modulated LC  LC” (LC/ a  m/LC) in order to overcome the issues related to the employment of different 1D and 2D column diameters. In such a set-up, the fractions collected from the 1D are

21.3 Instrumental set-up and data analysis

subjected to a dilution flow prior to be injected into the 2D with the befit to lower the 1 D mobile phase strength. Such a system includes “trap” columns, instead of sampling loops, with linear velocities in the 1D closer to the optimum without the need of flow splitting [52–54]. Similar set-ups have also been investigated by other research groups [55,56], and a typical scheme is reported in Figure 21.1. [57]. A novel modulation technique with trap loops named pulsed elution (PE) has been developed lately to improve the LC  LC separations in RP  RP [58] and RP  HILIC [59] systems. The PE gradient implies a set of pulses in which the solvent strength increases gradually. The use of a PE gradient can be advantageously used in the 1D separation with the advantage of ensuring a proper fractionation of the 1D effluent. The number of fractions and the stop-flow period can be regulated independently allowing for greater flexibility in the modulation time without loss of 1D separation efficiency. Figure 21.2. illustrates the comparison of the separation of saponins in a Panax ginseng root extract with a linear program gradient and PE gradient [59]. The proposed RP-LC  HILIC system showed excellent orthogonality, and 20% more peaks could be obtained with the PE method than with the traditional gradient program.

FIGURE 21.1 Scheme of the time-decoupled online comprehensive 2D-UHPLC-ion mobility MS method. Reproduced from Zhang et al. [57] with permission from Elsevier.

567

568

CHAPTER 21 Multidimensional liquid chromatography

FIGURE 21.2 Comparison of the separation of saponins in Panax ginseng root extract with a linear program gradient and PE gradient. Reproduced from Chen et al. [59] with permission from Elsevier.

A novel implementation of existing valve technology was exploited by Stoll and co-workers who developed a “selective” LC  LC system (sLC  LC) [60–62]. In the sLC  LC approach, fractions of the 1D effluent, prior to be injected into the 2D, are temporarily stored until the 1D separation is accomplished. This approach allows solving the issue related to the 1D peak widths and 2D analysis, thus allowing highly efficient 1D separations. The added time dimension of the sLC  LC datasets comes by sampling 1D peaks multiple times across their widths and enables the use of sophisticated chemometric algorithms to mathematically resolve chromatographically unresolved peaks. Moreover, the transfer and subsequent 2D separations of multiple fractions of a particular 1D peak produces a two-dimensional chromatogram, which reveals the coordinates of the peak in both dimensions. The sLC  LC approach is very repeatable, even at very short 1D sampling times, with values as low as 1 s. A valve-based modulation approach was also experienced by Schmitz and co-workers for post-1D dilution using a double-loop modulator [63]. The 1D effluent with a high content of organic solvent is diluted by a make-up flow with water and then accurately split. The 1D effluent dilution allows focusing the analytes on the top of the 2D column, avoiding the negative effects in the subsequent 2 D separation. In order to improve the transfer of the total 1D effluent, a fixed solvent modulator (FSM) was proposed by Petersson et al. [64] in 2016. This type of modulator does not require a third pump to generate the dilution effect since the 2D flow is split into two parts: one part flows through the sample loop and the other part bypasses the modulator, and both flows are combined behind the modulator by a T-type connection; the mixed fraction is afterward injected into the 2D column. The main advantage is that the transfer fraction from the sample loop is on-line diluted by the 2D weak

21.3 Instrumental set-up and data analysis

mobile phase coming directly from the bypass way without causing sample loss. The principal disadvantage is due to the time difference between the mobile phases flowing through these two paths, thus altering the ratio of the different mobile phases, leading to baseline shifts during the run hampering the detection of less abundant signals. Such an issue was solved by the introduction of the so-called active solvent modulation (ASM), which was designed to control the on and off of the bypass at each modulation period [65]. Under these conditions, the split of the 2D mobile phase only occurs at the stage of the transfer of fractions, partially avoiding the negative effects on the 2D separation. With regard to mobile phase incompatibility of NP-LC  RP-LC separations a vacuum evaporation interface was employed by Guan and co-workers [66]. Such an approach allowed condensing the 1D eluents and the 2D solvent redissolved the residents at the inside wall of a loop for further separation in the 2D. The main pitfall was the potential sample loss risk for volatile components due to evaporation in the interface. Later, a newly developed vacuum evaporation-assisted adsorption (VEAA) interface, allowing fast removal of NP-LC solvent in the vacuum condition and successfully solving the solvent incompatibility problem between NP-LC and RP-LC, was constructed for preparative purposes [67]. In such field, two remarkable improvements have been introduced by Verstraeten et al. [68] and Fornells et al. [69]. In the first case, based on the thermal desorption concept used in comprehensive two-dimensional gas chromatography (GC  GC) systems, preconcentration of neutral analytes eluting from the 1D was performed in a capillary “trap” column packed with highly retentive porous graphitic carbon particles, placed in an aluminum low–thermal mass LC heating sleeve. Remobilization of the trapped analytes was achieved by rapidly heating the trap column, by applying temperature ramps up to +1200 °C/min. The whole LC  LC system was validated by the analysis of a red wine sample, allowing to obtain 2D peaks about 30–55% narrower than the ones found with the “conventional” valve modulator. In the second case, in 2018, a membrane evaporative interface was developed (Figure 21.3.) [69]. The main pitfall is the occurrence of uneven evaporation rates, which is deleterious for regular LC  LC sampling. Such an issue could be solved by proper optimization of both flow rate and heating conditions. More recently, aiming to achieve high peak capacities in relatively short analysis times a TWo-dimensional Insertable Separation Tool (TWIST) concept has been introduced [70]. Such a platform, named spatial LC  LC (xLC  xLC), was realized by using 3D printing technology [71]. It is based on adequate flow control and confinement of the analytes to the desired regions, namely, confinement in the 1D direction and subsequently homogeneous flow in the 2D (Figure 21.4.). An interesting tutorial on spatial comprehensive three-dimensional chromatography components, separated within a three-dimensional separation space, was recently reported by Eeltink and co-workers [72]. Regarding detection, all conventional LC detectors, for example, PDA, evaporative light scattering (ELS), and/or MS can be conveniently hyphenated to an MD-LC system. Usually, a single detector is placed after the 2D column; in some

569

570

CHAPTER 21 Multidimensional liquid chromatography

Metallic case

Liquid channel

Hydrophobic membrane

Gas contact channel

Gas distribution channels

To inlet valve

To vacuum pump

FIGURE 21.3 Exploded view of the membrane evaporation device. Reproduced from Fornells et al. [69] with permission from Elsevier.

applications, an additional detector can be employed after the 1D column in order to monitor the separation prior to the 2D separation. 2D analysis is run at a high speed; thus, a very fast acquisition rate is needed for adequate sampling, especially if quantification is required. Either MD-LC or LC  LC data can be visualized in contour plots, where x- and y-axes represent the 1D and the 2D retention times, respectively, whereas the color of the spots is related to intensity. Quantification can be carried out either by summing the single 2D slices of a 1D chromatographic peak or by applying more advanced algorithms.

References

FIGURE 21.4 TWIST for flow confinement in spatial separations: the proposed device in assembly form, consisting of an internal (gray) and an external (blue) part. Insert: sketch of the internal part. Reproduced from Adamopoulou et al. [70] under the terms of the Creative Commons CC-BY license.

21.4 Conclusions and future perspectives Multidimensional liquid chromatography has been extensively exploited in the last three decades with remarkable updates in the last few years with reference to instrumental set-up and innovative modulation systems. Such improvements aimed to solve somehow the decreased detection sensitivity and compatibility issues associated with both conventional multidimensional and comprehensive two-dimensional applications. Also, the implementation of commercially available instruments as well as the design of more robust and flexible software for both qualitative and quantitative purposes is expected to grow in the near future, evolving toward a more “routine” use of both techniques.

References [1] Huber JFK, van der Linden R, Ecker E, Oreans M. Column switching in high-pressure liquid chromatography. J Chromatogr 1973;83:267–77. [2] Erni F, Frei RW. Two-dimensional column liquid chromatographic technique for resolution of complex mixtures. J Chromatogr 1978;149:561–9. [3] Stoll DR, Li X, Wang X, Carr PW, Porter SEG, Rutan SC. Fast, comprehensive twodimensional liquid chromatography. J Chromatogr A 2007;168:3–43. [4] Dugo P, Cacciola F, Kumm T, Dugo G, Mondello L. Comprehensive multi-dimensional liquid chromatography: theory and applications. J Chromatogr A 2008;1184:353–68. [5] Dugo P, Kumm T, Cacciola F, Dugo G, Mondello L. Multidimensional liquid chromatographic separations applied to the analysis of food samples. J Liq Chromatogr Relat Technol 2008;31:1758–807.

571

572

CHAPTER 21 Multidimensional liquid chromatography

[6] Jandera P. Column selectivity for two-dimensional liquid chromatography. J Sep Sci 2006;29:1763–83. [7] Shalliker RA, Gray MJ. Concepts and practice of multidimensional high-performance liquid chromatography. Adv Chromatogr 2006;44:177–236. [8] Shellie RA, Haddad PR. Comprehensive two-dimensional liquid chromatography. Anal Bioanal Chem 2006;386:405–15. [9] Guiochon G, Marchetti N, Mriziq K, Shalliker RA. Implementations of two- dimensional liquid chromatography. J Chromatogr A 2008;1189:109–68. [10] Pol J, Hyotylainen T. Comprehensive two-dimensional liquid chromatography coupled with mass spectrometry. Anal Bioanal Chem 2008;391:21–31. [11] Franc¸ois I, Sandra K, Sandra P. Comprehensive liquid chromatography: fundamental aspects and practical considerations – a review. Anal Chim Acta 2009;641:14–31. [12] Herrero M, Ibanez E, Cifuentes A, Bernal J. Multidimensional chromatography in food analysis. J Chromatogr A 2009;1216:7110–29. [13] Sandra K, Moshir M, D’hondt F, Tuytten R, Verleysen K, Kas K, Franc¸ois I, Sandra P. Highly efficient peptide separations in proteomics. Part 2: bi- and multidimensional liquid-based separation techniques. J Chromatogr B 2009;877:1019–39. [14] Berek D. Two-dimensional liquid chromatography of synthetic polymers. Anal Bioanal Chem 2010;396:421–41. [15] Pierce KM, Kehimkar B, Marney LC, Hoggard JC, Synovec RE. Review of chemometric analysis techniques for comprehensive two dimensional separations data. J Chromatogr A 2012;1255:3–11. [16] Pierce KM, Mohler RE. A review of chemometrics applied to comprehensive twodimensional separations from 2008-2010. Sep Purif Rev 2012;41:143–68. [17] Jandera P. Comprehensive two-dimensional liquid chromatography-practical impacts of theoretical considerations. A review. Cent Eur J Chem 2012;10:844–75. [18] Donato P, Cacciola F, Tranchida PQ, Dugo P, Mondello L. Mass spectrometry detection in comprehensive liquid chromatography: basic concepts, instrumental aspects, applications and trends. Mass Spectrom Rev 2012;1:523–59. [19] Cacciola F, Donato P, Beccaria M, Dugo P, Mondello L. Advances in LC-MS for food analysis. LC GC Eur 2012;25:15–24. [20] Tranchida PQ, Donato P, Cacciola F, Beccaria M, Dugo P, Mondello L. Potential of comprehensive chromatography in food analysis. TrAC Trends Anal Chem 2013;52:186–205. [21] Shi X, Wang S, Yang Q, Lu X, Xu G. Comprehensive two-dimensional chromatography for analyzing complex samples: recent new advances. Anal Meth 2014;6:7112–23. [22] Sarrut M, Cretier G, Heinisch S. Theoretical and practical interest in UHPLC technology for 2D-LC. TrAC-Trends Anal Chem 2014;63:104–12. [23] Cacciola F, Beccaria M, Donato P, Mondello L, Dugo P. Combining orthogonal comprehensive two-dimensional liquid chromatography methods in food analysis. LC GC Eur 2014;27:570–7. [24] Li D, Jakob C, Schmitz O. Practical considerations in comprehensive two-dimensional liquid chromatography systems (LCxLC) with reversed-phases in both dimensions. Anal Bioanal Chem 2015;407:153–67. [25] Cesla P, Krenkova J. Fraction transfer process in on-line comprehensive twodimensional liquid-phase separations. J Sep Sci 2017;40:109–23. [26] Cacciola F, Farnetti S, Dugo P, Marriott PJ, Mondello L. Comprehensive twodimensional liquid chromatography for polyphenol analysis in foodstuffs. J Sep Sci 2017;40:7–24.

References

[27] Cacciola F, Donato P, Sciarrone D, Dugo P, Mondello L. Comprehensive liquid chromatography and other liquid-based comprehensive techniques coupled to mass spectrometry in food analysis. Anal Chem 2017;89:414–29. [28] Cesla P, Krenkova´ J. Fraction transfer process in on-line comprehensive twodimensional liquid-phase separations. J Sep Sci 2017;40:109–23. [29] Stoll DR, Shoykhet K, Petersson P, Buckenmaier S. Active solvent modulation: a valvebased approach to improve separation compatibility in two-dimensional liquid chromatography. Anal Chem 2017;89:9260–7. [30] Pirok BWJ, Gargano AFG, Schoenmakers PJ. Recent developments in two-dimensional liquid chromatography: fundamental improvements for practical applications. J Sep Sci 2018;41:68–98. [31] Ji S, Wang S, Xu H, Su Z, Tang D, Qiao X, Ye M. The application of on-line twodimensional liquid chromatography (2DLC) in the chemical analysis of herbal medicines. J Pharm Biomed Anal 2018;160:301–13. [32] Pirok BWJ, Stoll DR, Schoenmakers PJ. Recent developments in two-dimensional liquid chromatography: fundamental improvements for practical applications. Anal Chem 2019;91:240–63. [33] Cacciola F, Rigano F, Dugo P, Mondello L. Comprehensive two-dimensional liquid chromatography as a powerful tool for the analysis of food and food products. TrAC-Trends Anal Chem 2020;127, 115894. [34] Cacciola F, Arena K, Mandolfino F, Donnarumma D, Dugo P, Mondello L. Reversed phase versus hydrophilic interaction liquid chromatography as first dimension of comprehensive two-dimensional liquid chromatography systems for the elucidation of the polyphenoliccontent of food and natural products. J Chromatogr A 2021;1645, 462129. [35] Karger BL, Snyder LR, Horvath C. An introduction to separation science. New York: Wiley & Sons; 1973. [36] Giddings JC. Two-dimensional separations: concept and promise. Anal Chem 1984;37:1258A–64A. [37] Giddings JC. Concepts and comparisons in multidimensional separation. J High Resolut Chromatogr Chromatogr Commun 1987;10:319–23. [38] Murphy RE, Schure MR, Foley JP. Effect of sampling rate on resolution in comprehensive two-dimensional liquid chromatography. Anal Chem 1998;70:1585–94. [39] Seeley JV. Theoretical study of incomplete sampling of the first dimension in comprehensive two-dimensional chromatography. J Chromatogr A 2002;962:21–7. [40] Horie K, Kimura H, Ikegami T, Iwatsuka A, Saad N, Fiehn O, Tanaka N. Calculating optimal modulation periods to maximize the peak capacity in two dimensional HPLC. Anal Chem 2007;79:3764–70. [41] Schoenmakers PJ, Vivo-Truyols G, Decrop WMC. A protocol for designing comprehensive two-dimensional liquid chromatography separation systems. J Chromatogr A 2006;1120:282–90. [42] Davis JM, Stoll DR, Carr PW. Effect of first-dimension undersampling on effective peak capacity in comprehensive two-dimensional separations. Anal Chem 2008;80:461–73. [43] Davis JM, Stoll DR, Carr PW. Dependence of effective peak capacity in comprehensive two-dimensional separations on the distribution of peak capacity between the two dimensions. Anal Chem 2008;80:8122–34. [44] Li X, Stoll DR, Carr PW. Equation for peak capacity estimation in two-dimensional liquid chromatography. Anal Chem 2009;81:845–50.

573

574

CHAPTER 21 Multidimensional liquid chromatography

[45] Potts LW, Stoll DW, Li X, Carr PW. The impact of sampling time on peak capacity and analysis speed in on-line comprehensive two-dimensional liquid chromatography. J Chromatogr A 2010;1217:5700–9. [46] Gu H, Huang Y, Carr PW. Peak capacity optimization in comprehensive two dimensional liquid chromatography: a practical approach. J Chromatogr A 2011;1218:64–73. [47] Schure MR, Davis JM. Orthogonal separations: comparison of orthogonality metrics by statistical analysis. J Chromatogr A 2015;1414:60–76. [48] Camenzuli M, Schoenmakers PJ. A new measure of orthogonality for multi-dimensional chromatography. Anal Chim Acta 2014;838:93–101. [49] Peters S, Vivo-Truyols G, Marriott PJ, Schoenmakers PJ. Development of a resolution metric for comprehensive two-dimensional chromatography. J Chromatogr A 2007;1146:232–41. [50] Cacciola F, Jandera P, Mondello L. Temperature effects on separation on zirconia columns: applications to one-and two-dimensional LC separations of phenolic antioxidants. J Sep Sci 2007;30:462–74. [51] Cacciola F, Jandera P, Hajdu` Z, Cesla P, Mondello L. Comprehensive two-dimensional liquid chromatography with parallel gradients for separation of phenolic and flavone antioxidants. J Chromatogr A 2007;1149:73–87. [52] Vonk RJ, Wouters S, Barcaru A, Vivo´-Truyols G, Eeltink S, de Koning LJ, Schoenmakers PJ. Post-polymerization photografting on methacrylate-based monoliths for separation of intact proteins and protein digests with comprehensive two-dimensional liquid chromatography hyphenated with high-resolution mass spectrometry. Anal Bioanal Chem 2015;407:3817–29. [53] Vonk RJ, Gargano AFG, Davydova E, Dekker HL, Eeltink S, de Koning LJ, Schoenmakers PJ. Comprehensive two-dimensional liquid chromatography with stationaryphase-assisted modulation coupled to high-resolution mass spectrometry applied to proteome analysis of Saccharomyces cerevisiae. Anal Chem 2015;87:5387–94. [54] Gargano AFG, Duffin M, Navarro P, Schoenmakers PJ. Reducing dilution and analysis time in online comprehensive two-dimensional liquid chromatography by active modulation. Anal Chem 2016;88:1785–93. [55] Sommella E, Ismail OH, Pagano F, Pepe G, Ostacolo C, Mazzoccanti G, Russo M, Novellino E, Gasparrini F, Campiglia P. Development of an improved online comprehensive hydrophilic interaction chromatographyreversed-phase ultra-high-pressure liquid chromatography platform for complex multiclass polyphenolic sample analysis. J Sep Sci 2017;40:2188–97. [56] Montero L, Ibanez E, Russo M, Rastrelli L, Cifuentes A, Herrero M. Focusing and nonfocusing modulation strategies for the improvement of on-line two-dimensional hydrophilic interaction chromatographyreversed phase profiling of complex food samples. Anal Chim Acta 2017;985:202–12. [57] Zhang H, Jiang J-M, Zheng D, Yuan M, Wang Z-Y, Zhang H-M, Zheng C-W, Xiao L-B, Xu H-X. A multidimensional analytical approach based on time-decoupled online comprehensive two-dimensional liquid chromatography coupled with ion mobility quadrupole time-of-flight mass spectrometry for the analysis of ginsenosides from white and red ginsengs. J Pharm Biomed Anal 2019;163:24–33. [58] Jakobsen SS, Christensen JH, Verdier S, Mallet CR, Nielsen NJ. Increasing flexibility in two-dimensional liquid chromatography by pulsed elution of the first dimension: a proof of concept. Anal Chem 2017;89:8723–30.

References

[59] Chen Y, Wu Y, Liu X, Li B, Hu D, Huang S, Ma M, Chen B. Pulsed elution modulation for on-line comprehensive two-dimensional liquid chromatography coupling reversed phase liquid chromatography and hydrophilic interaction chromatography. J Chromatogr A 2019;1583:98–107. [60] Groskreutz SR, Swenson MM, Secor LB, Stoll DR. Selective comprehensive multidimensional separation for resolution enhancement in high performance liquid chromatography. Part I: principles and instrumentation. J Chromatogr A 2012;1228:31–40. [61] Davis JM, Stoll DR. Likelihood of total resolution in liquid chromatography: evaluation of one-dimensional, comprehensive two-dimensional, and selective comprehensive twodimensional liquid chromatography. J Chromatogr A 2014;1360:128–42. [62] Larson ED, Groskreutz SR, Harmes DC, Gibbs-Hall IC, Trudo SP, Allen RC, Rutan SC, Stoll DR. Development of selective comprehensive two-dimensional liquid chromatography with parallel first-dimension sampling and second-dimension separationapplication to the quantitative analysis of furanocoumarins in apiaceous vegetables. Anal Bioanal Chem 2013;405:4639–53. [63] Stephan S, Hippler J, Kohler T, Deeb AA, Schmidt TC, Schmitz OJ. Contaminant screening of wastewater with HPLC-IM-qTOF-MS and LCLC-IM-qTOF-MS using a CCS database. Anal Bioanal Chem 2016;408:6545–55. [64] Petersson P, Haselmann K, Buckenmaier S. Multiple heart-cutting two dimensional liquid chromatography mass spectrometry: towards real time determination of related impurities of bio-pharmaceuticals in salt based separation methods. J Chromatogr A 2016;1468:95–101. [65] Stoll DR, Shoykhet K, Petersson P, Buckenmaier S. Active solvent modulation: a valvebased approach to improve separation compatibility in two-dimensional liquid chromatography. Anal Chem 2017;89:9260–7. [66] Tian H, Xu J, Guan Y. Comprehensive two-dimensional liquid chromatography (NPLCRPLC) with vacuum-evaporation interface. J Sep Sci 2008;31:1677–85. [67] Li J-F, Fang H, Yan X, Chang F-R, Wu Z, Wu Y-L, Qiu Y-K. On-line comprehensive two-dimensional normal-phase liquid chromatographyreversed-phase liquid chromatography for preparative isolation of toad venom. J Chromatogr A 2016;1456:169–75. [68] Verstraeten M, Pursch M, Eckerle P, Luong J, Desmet G. Thermal modulation for multidimensional liquid chromatography separations using low-thermal-mass liquid chromatography (LC). Anal Chem 2011;83:7053–60. [69] Fornells E, Barnett B, Bailey M, Hilder E, Shellie RA, Breadmore MC. Evaporative membrane modulation for comprehensive two-dimensional liquid chromatography. Anal Chim Acta 2018;1000:303–9. [70] Adamopoulou T, Deridder S, Desmet G, Schoenmakers PJ. Two-dimensional insertable separation tool (TWIST) for flow confinement in spatial separations. J Chromatogr A 2018;1577:120–3. [71] Wouters B, Davydova E, Wouters S, Vivo-Truyols G, Schoenmakers PJ, Eeltink S. Towards ultra-high peak capacities and peak-production rates using spatial threedimensional liquid chromatography. Lab Chip 2015;15:4415–22. [72] Themelis T, Amini A, De Vos J, Eeltink S. Towards spatial comprehensive threedimensional liquid chromatography: a tutorial review. Anal Chim Acta 2021;1148, 238157.

575

CHAPTER

Process concepts in preparative chromatography

22

Malte Kaspereit1 and Andreas Seidel-Morgenstern2,3 1

Friedrich Alexander University Erlangen-Nuremberg, Institute for Separation Science & Technology, Erlangen, Germany, 2Otto von Guericke University, Institute for Process Engineering, Magdeburg, Germany, 3Max Planck Institute for Dynamics of Complex Technical Systems, Magdeburg, Germany

22.1 Introduction Chromatographic separation processes using solid stationary and fluid mobile phases are frequently applied industrially with preparative purpose in order to isolate and purify specific target compounds [1]. Required purity criteria can often be met only if chromatographic techniques are applied, as for example in the important field of enantioseparations [2]. As in analytical chromatography, the basis for every successful preparative chromatographic separation is the proper choice of the chromatographic system, i.e., the combination of stationary and mobile phases. However, besides this selection, the way how the preparative separation process is carried out, i.e., the operating mode, plays an important role. It decides about the achievable productivity and yield, the specific desorbent consumption, and thus, the overall process economy. Traditionally, a specific separation process is first developed and realized exploiting the principle of isocratic batch elution. However, nowadays, there is a large arsenal of more flexible process options available that is increasingly applied to improve performance criteria. More straightforward modifications are here based on keeping the most essential features of the classical process, namely, the exploitation of a single column and its discontinuous character. Gradient elution processes exploit the fact that during the chromatographic process, a modulation of certain operating parameters (composition of the mobile phase, temperature, pressure, and flow rate) can improve the performance of conventional isocratic operation. Promising other additional degrees of freedom is exploited in concepts based on recycling insufficiently separated fractions back into the column. From a chemical engineering point of view, continuously operated separation processes are most attractive. Inspired by the success of applying other separation principles in a continuous mode (distillation, extraction, etc.), nowadays, concepts Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00018-7 Copyright # 2023 Elsevier Inc. All rights reserved.

577

CHAPTER 22 Process concepts in preparative chromatography

for continuous chromatographic separation are available. The most relevant approaches are the various variants of multi-column simulated moving bed (SMB) chromatography. This concept exploits several columns connected in series and mimics a continuous counter-current movement between the mobile and stationary. This facilitates a continuous separation. It should be mentioned that the increased application of more sophisticated operating regimes in preparative chromatography is also driven by our improved understanding of the dynamics of concentration profiles moving in chromatographic columns under nonlinear (overloaded) conditions [3–6].

22.2 Classical isocratic discontinuous elution chromatography The classical single-column principle of batch chromatography is well known from analytical chromatography, and Figure 22.1 is sufficient to explain the general principle. Traditionally, a specific separation processes is first developed and realized for preparative application by injecting in a repetitive periodic manner samples of the feed mixture and collecting the target products batch-wise. As indicated in the bottom of Figure 22.1, subsequent feed samples are typically dosed already before complete elution of the previous sample. Nevertheless, the stationary phase is often not efficiently exploited. Thus, in preparative chromatography, there is a strong interest in more productive and possibly continuous separation modes. There is large experience and good theoretical understanding of this classical and straightforward batch operating mode. Thus, it is typically considered as a Injection

Column

Detector

Fractionation

Feed

Product 1 ...

Product 2 Product n

Eluent 'tcycle

Oulet conc.

Inlet conc.

578

Time

'tcycle 1 2

'tcycle 1 2

3

3 Time

FIGURE 22.1 Top: Principle of classical discontinuously operated isocratic batch elution. Bottom: Illustration of injection policy and resulting chromatograms for three components and two consecutive injections.

22.2 Classical isocratic discontinuous elution chromatography

benchmark process, serving as a reference for evaluating the potential of alternatives. To perform such a comparative analysis, quantitative models are most suitable. A frequently applied standard model of a chromatographic column will be described below, together with simulation results that illustrate typical features of nonlinear chromatography. The model presented can be seen as a building block for deriving modified models that are capable of quantifying all other operating modes of preparative chromatography discussed in this chapter.

22.2.1 Mathematical modeling and typical effects There are a large number of models available to describe the migration processes of concentration fronts in chromatographic columns. A general classification is to distinguish between discrete equilibrium stage models and continuous models leading to systems of partial differential equations (PDE). The latter type of models allows for a more straightforward implementation of the various mass transfer processes occurring in a chromatographic column as demonstrated in the general rate model (GRM) [5,6]. Here we introduce briefly the frequently applied simple equilibrium dispersion model (EDM). It is based on the following assumptions: • • • •

the two phases are considered as quasi-homogeneous, i.e., adsorption equilibria are permanently established throughout the column, for relative small particles, intra-particle concentration gradients are negligible, in well-packed columns, radial concentration gradients can be neglected, isothermal conditions can be assumed due to relatively low thermal effects.

In the EDM, all effects causing band broadening are lumped into an apparent dispersion coefficient, Dapp. The corresponding mass balance for a component i in a mixture of K components is ∂ci 1  ε ∂qi ðc1 , c2 , …, cK Þ ∂ci ∂2 ci + ¼ Diapp 2 +u ∂t ∂z ∂z ε ∂t

i ¼ 1, …, K,

(22.1)

where ε stands for the total column porosity and u for the linear velocity of the mobile phase. Most essential in Eq. (22.1) are the equilibrium functions relating the loading of a component i, qi, to the fluid phase concentrations of all components, c1, c2, …, cK. In order to use the model for simulations, the typically nonlinear functions qi ¼ qi(c1, c2, …, cK) must be provided. Due to its simplicity and capability of fitting many experimental results, the following multi-component Langmuir equation is often applied to describe these functions based on component-specific parameters ai and bi [5,6]: qi ¼ 1+

ai ci K X j¼1

i ¼ 1, …, K bj cj

(22.2)

579

CHAPTER 22 Process concepts in preparative chromatography

For efficient columns, the apparent dispersion coefficient is closely related to the number of theoretical plates, N, by the following relation: Diapp ¼

uL 2N i

i ¼ 1, …, K

(22.3)

Hereby, L is the column length. To describe batch elution, typically, completely regenerated columns are considered to specify the initial conditions at t ¼ 0 ci ðz, t ¼ 0Þ ¼ 0 i ¼ 1, …, K

(22.4)

A rectangular injection profile is assumed as the boundary condition at the column inlet (z ¼ 0) 0  t < tinj : t  tinj :

ci ðz ¼ 0, tÞ ¼ ciinj

ðinjectionÞ i ¼ 1, …, K

ci ðz¼ 0, tÞ ðelutionÞ

(22.5)

i ¼ 1, …, K

(22.6)

(a) Linear condions (c inj = 1)

(b) Non-linear condions (c inj = 200) Concentraon

Other operating regimes described later will essentially differ with respect to these initial and boundary conditions as will be shown below. Since there are no general analytical solutions available for the set of PDEs given in Eq. (22.1), one of the numerous available numerical methods must be used [5,6]. Summarizing main results of a simulation study carried out solving Eq. (22.1), two figures illustrate typical features of nonlinear chromatography. In contrast to the small sample sizes that are processed for analysis, larger injection concentrations and/or volumes are relevant in a preparative scale. Under these conditions, the underlying distribution equilibria are not linear anymore and concentration-dependent migration speeds and competition effects arise. Figure 22.2A illustrates the

Concentraon

580

0,003

N 0,002

0,20 0,15

N 0,10

0,001 0,05 0,000

0,00 5

5,5

6

6,5

7

Time

0

2

4

6

8

Time

FIGURE 22.2 Elution profiles as function of the plate number N according to Eqs. (22.1)–(22.6) under linear (cinj ¼ 1) and nonlinear (cinj ¼ 200) conditions, respectively. Other parameters (in compatible units): ε ¼ 0.5, u ¼ 20, L ¼ 20, a ¼ 5, Injections: c(t,z ¼ 0) ¼cinj for 0 < t < 0.001. Plate numbers: N ¼ 500, 1000, 2000.

22.2 Classical isocratic discontinuous elution chromatography

significant impact of the plate number N (or the apparent dispersion coefficient Dapp) on the variance of chromatographic bands under linear conditions. Figure 22.2B reveals that under nonlinear conditions, i.e., for higher injection amounts, the plate numbers (or efficiencies) lose importance and thermodynamics start to influence more strongly band shapes and retention times. It is a well-known fact that the impact of column efficiency is typically less important under overloaded conditions which are typical for preparative separations. The important aspect of competitive adsorption equilibria relevant under nonlinear conditions is further exemplified in Figure 22.3, where simulated elution profiles are compared for injecting the same amount of a component alone or in a mixture with another component. The two effects that can be observed are called “displacement effect” (Figure 22.3A) and “tag-along effect” (Figure 22.3B). The essential differences between analytical and preparative elution chromatography were comprehensively analyzed and summarized by Guiochon et al. [5].

Concentraon

(a) Displacement effect (c 1inj = 20, c2inj = 180) 0,20 0,15 0,10 0,05 0,00

0

1

2

3

4

5

6

7

Time

Concentraon

(b) Tag-along effect (c 1inj = 180, c2inj = 20) 0,20 0,15 0,10 0,05 0,00

0

1

2

3

4

5

6

7

Time

FIGURE 22.3 Illustrations of the effects of competitive adsorption. Elution profiles according to Eqs. (22.1)–(22.6) for injections of the same amounts as a single component (overlay of dashed lines) and in a binary mixture (solid lines), respectively. (A) Illustration of the displacement effect for cinj1 ¼ 20, cinj2 ¼ 180. (B) Illustration of the tag-along effect for cinj1 ¼ 180, cinj2 ¼ 20. Other parameters (in compatible units): ε ¼ 0.5, u ¼ 20, L ¼ 20, a1 ¼ 4, b1 ¼ 4, a2 ¼ 5, b1 ¼ 5, N ¼ 1000. Injections for 0 < t < 0.001.

581

582

CHAPTER 22 Process concepts in preparative chromatography

22.3 Other discontinuous operating concepts In order to exploit the resolution potential of a single chromatographic column, several alternatives to conventional isocratic batch elution have been developed. One example is a concept based on changing periodically the flow direction. This socalled “flip-flop chromatography” [7] is attractive if the feed contains very slowly migrating components that would cause long cycle times in the conventional mode. Another powerful concept is to inject after the sample wide plugs of better adsorbable components that displace the sample components while the separation occurs. This concept, denoted as displacement chromatography, has the attractive potential to increase certain concentrations during the chromatographic process [8]. Below, we consider in more detail two very flexible and widely applicable options, namely, gradient elution and recycling chromatography.

22.3.1 Gradient chromatography Due to the fact that in an industrial scale, simple and reliable regimes are preferred, and separation processes with a preparative purpose are often performed using isocratic elution chromatography, as described in the previous section. During isocratic elution, all operating conditions are kept constant with the exception of the periodic changes of injecting at the column inlet alternating the feed or the mobile phase. Thus, analysis and design of the separation processes rely typically on nonlinear but fixed thermodynamic functions. However, often the retention behavior of the mixture components varies in a broad range, and it is attractive to modulate certain process conditions during the process, e.g., to speed up slowly migrating components [9,10]. Such a gradient operation allows to inject samples more frequently and to increase the overall process productivity [11–13]. In addition, there might be further favorable effects of gradients compared to isocratic operation related, e.g., to the potential of enhancing the product concentrations or stabilizing retention times. Parameters that influence the migration speeds of concentration fronts in chromatographic columns are, e.g., the temperature, the pressure, the flow rate and the composition of the mobile phase. The latter includes in liquid chromatography the frequently applied gradients of pH value, ionic strength, or polarity of the mobile phase. Due to the availability of accurate gradient pumps, the solvent composition can be accurately altered during the separation process. Usually, the elution strength is increased during the process which leads to gradual reductions of the retention times. Depending on the specific character of the separation problem, the shapes of the gradient and thus the component migration speeds can be adjusted to increase certain performance parameters. Figure 22.4 illustrates the consequences of a linear gradient on the course of an elution profile. Other conceivable gradient shapes are shown schematically in Figure 22.5. In order to quantify and optimize chromatographic processes using solvent strength gradients, it is necessary to describe the effect of the modifier on the course

Concentration

22.3 Other discontinuous operating concepts

2 linear gradient

isocrac 1

0 2

3

4

5

12

13

14

Time

FIGURE 22.4

Eluent composion [%]

Illustration of the effect of a linear gradient on the retention behavior of a ternary mixture. Solid lines—gradient conditions; dashed—isocratic operation.

100% "strong"

100% "weak"

Time

FIGURE 22.5 Examples of conceivable gradient shapes.

of the equilibrium functions in the model equations. This issue has been systematically studied under linear conditions for many years (e.g., [14,15]). The situation becomes more complex if the effect of the solvent composition has to be quantified for nonlinear isotherms. Currently, there is no reliable general theory available to predict accurately the corresponding equilibrium functions. For this reason, typically systematic experimental investigations at different modifier levels are carried out in order to determine the corresponding isotherms. Often the impact of the solvent composition or the concentration of a modifier on the parameters of an isotherm model, which is applicable to describe isocratic situations, is included empirically. For the

583

584

CHAPTER 22 Process concepts in preparative chromatography

Langmuir model given above (Eq. 22.2), this approach leads to the following isotherm model equation: qi ¼

  ai cmod ci K X   1+ bj cmod cj

(22.7)

j¼1

The impact of the dynamically changing concentrations of a modifier applied in the mobile phase, cmod, on the isotherms parameters ai and bi is frequently described by simple empirical functions, often in the form of power laws [16]. There are intense efforts to exploit the potential of applying gradients also in SMB processes (e.g., [17]) (see Section 22.4 below).

22.3.2 Recycling techniques Typically, the performance of preparative chromatographic separations is limited not only by the effects of non-linear thermodynamic equilibria as illustrated earlier but also by the band-broadening due to axial dispersion and slow mass transfer. These restrict the amounts that can be separated per cycle. Increasing the injection amounts leads to insufficiently separated peaks and, thus, decreased product recovery. One way to overcome these limitations would be to improve the column efficiency by increasing the column length and/or using a stationary phase with smaller particles. However, the resulting increased pressure drops limit the applicability of such attempts. An alternative to modifying the column is to apply recycling techniques. As illustrated in Figure 22.6, in recycling chromatography, the sample to be separated is transported several times through the column by connecting periodically the column outlet to its inlet. In some respect, this corresponds to “simulating” a longer column. Selected variants of recycling chromatography will be discussed below.

22.3.2.1 Closed-loop recycling chromatography In closed-loop recycling (CLR) chromatography [18–20], the column outlet is connected to the column inlet in order to re-inject the partially separated mixture into the column without “destroying” the elution profile. As illustrated in Figure 22.6B, this procedure is repeated until the components of interest are separated in one of the subsequent passages through the column. The critical point in closed-loop recycling chromatography is the increase in axial dispersion due to the additional hold-up volume generated by a recycle pump and additional piping. CLR chromatography is, therefore, in many cases combined with “peak shaving” [19]. The latter is shown in Figure 22.6C and denotes the removal of the sufficiently purified leading and trailing edges of the chromatogram. This leads to a reduction of the size of the re-injected fraction and, thus, to faster elution and possibly fewer required cycles. The main advantage of CLR techniques is the possibility of isolating even components of very low selectivity with good purity and yield, using simple standard

22.3 Other discontinuous operating concepts

(a) Injecon

Column

Detector

Fraconaon Product 1

Eluent

Product n

...

Feed

Recycle (c)

(d)

Oulet conc.

(b)

Time

Time

Time

FIGURE 22.6 Principle of recycling chromatography. (A) Schematic process setup. Solid lines mark continuous streams while dashed lines correspond to periodically active streams (B)–(D) illustrative chromatograms for different operating policies. (B) Closed-loop recycling (CLR), (C) CLR with peak shaving, (D) steady-state recycling (SSR). Dashed boxes mark the recycle fraction; vertical arrows denote the injection of fresh sample.

equipment. A second interesting aspect is that the consumption of mobile phase is not increased by this concept since only between the recycling intervals fresh eluent has to be introduced. One drawback of CLR chromatography is the dilution of the solutes in the column and the limited productivity caused by the fact that several cycles are required to perform the separation.

22.3.2.2 Steady-state recycling chromatography An interesting option to overcome this situation is to combine recycling chromatography with a periodic injection of fresh feed mixture. This concept, denoted as steady-state recycling (SSR) chromatography [21,22], combines the peak shaving strategy with the addition of a constant amount of fresh feed to each recycle fraction. This causes the process to attain a periodic steady state. An illustration is given in Figure 22.6D. There exist different operating concepts for SSR chromatography. The advantage of the closed-loop mode [22,23] is that the partially achieved separation of the recycle fraction is preserved. On the other hand, this concept is rather difficult to design and to optimize since in particular column length and/or the time point for reinjection need to be identified [23]. This holds also for more complex recycling strategies like segmented recycling [24], where different sub-fractions of the recycle are re-injected in an optimized sequence. In contrast to that, in the mixed-recycle mode [21,25], the recycle fraction is mixed before re-injection with the fresh feed in a

585

586

CHAPTER 22 Process concepts in preparative chromatography

reservoir in front of the column. While this mixing sacrifices the achieved partial separation, it facilitates a simple process design and optimization. Design methods exist for ideal, non-dispersive conditions, and Langmuir isotherms for pure products as well as limited purities [21,22]. The method was also extended for arbitrary purities under real conditions with dispersion [26]. Apart from improving the performance of a separation, an interesting aspect of recycling chromatography is the possibility to perform additional operations in the recycle line, for example, a solvent removal step [27] or a chemical reaction [28]. Regarding process performance, SSR chromatography can be seen as an intermediate between batch chromatography and the continuous multi-column SMB process to be discussed below. It combines lower complexity and equipment requirements compared to SMB chromatography with the potential of offering higher productivities and lower eluent consumption compared to batch chromatography.

22.4 Continuous simulated moving bed (SMB) chromatography In contrast to those described above, the operating concepts to be discussed in this section are capable of processing continuously a feed mixture and delivering continuously product streams. The simulated moving bed (SMB) process facilitates such operation. This operating concept applies a counter-current movement of the fluid phase and the solid phase located in several interconnected chromatographic columns.

22.4.1 Classical SMB operating concept Since its invention in the 1960s [29], simulated moving bed (SMB) chromatography has become a particularly powerful technology for preparative separations of binary mixtures. Its basis lies in a countercurrent movement of the solid (adsorbent) and mobile phases, respectively. This enhances the driving force for the separation and, thus, facilitates the utilization of the stationary phase. Properly designed SMB processes can achieve significantly higher productivity and lower eluent consumption than corresponding conventional single-column processes [30]. The principle is best explained using the hypothetical concept of the true moving bed (TMB) shown in Figure 22.7. In such TMB column, a continuous countercurrent of the solid and fluid phases is adjusted. A feed which contains the two components (or pseudo-components) A and B is introduced in the middle of the unit. If the flow rates in four characteristic zones are adjusted properly, a continuous and complete separation is possible. In this case, the weaker adsorbing component A travels with the fluid toward the raffinate port and the stronger retained B moves with the solid phase toward the extract. To utilize the two phases most economically, it is expedient to regenerate them in zones I and IV. This is achieved by introducing a desorbent stream at the bottom of zone I and adjusting a low flow rate in zone IV.

22.4 Continuous simulated moving bed (SMB) chromatography

conc. Zone IV

Raffinate (A)

A

Zone III

Feed (A + B)

B

Zone II

Extract (B)

Zone I Eluent fluid

solid

FIGURE 22.7 Principle of the true moving bed (TMB) chromatographic concept and typical internal concentration profiles for a successful binary separation. Feed (A + B)

Zone II

Raffinate (A)

Zone I

A III

Zone III

Liquid

IV

Port switching

conc.

II I

Extract (B)

B

Zone IV

Eluent

FIGURE 22.8 Principle of simulated moving bed (SMB) chromatography and typical internal concentration profiles for an 8-column unit in the middle of a shift period for a successful binary separation.

The obvious drawback of the TMB concept is the difficulty of transporting continuously the solid phase. Thus, in practice, this problem is avoided by providing the solid phase in a series of several connected packed columns. These are switched periodically by one position against the direction of the fluid flow, thus “simulating” the movement of the stationary phase. This simulated moving bed (SMB) process is illustrated in Figure 22.8. For larger column numbers, the SMB and TMB processes become equivalent. The SMB process was initially applied to specific separation problems in petrochemistry and in the sugar industry. In the last decade, SMB chromatography became a powerful tool for difficult separations in the pharmaceutical industry, as for example to separate enantiomers [31]. Currently, the technology finds growing application in the downstream processing of biomolecules [32].

587

588

CHAPTER 22 Process concepts in preparative chromatography

For design and evaluation of SMB processes, two complementary approaches exist. The first is based on the shortcut methods that give valuable initial estimates for the main design parameters, which are the four internal zone flow rates within the unit (see below). The second approach is to design and evaluate the process in more detail using a suitable (dynamic) process model that is based on single-column representations like the EDM explained in Section 22.2. The interconnection of the columns and their periodical switching are described by defining properly the initial and boundary conditions of each individual column (see Section 22.5). It is instructive to explain briefly the most successful shortcut design procedure outlined in [33]. It exploits the close analogy of the SMB to the TMB concept. Let us assume a TMB-based separation of a weaker adsorbing component A from a stronger retained solute B. Under ideal, dispersion-free conditions, the adsorption equilibria are established instantaneously throughout the unit. For simplicity, we assume linear equilibria qi ¼ ai ci , i ¼ A, B, aA < aB ,

(22.8)

i

with the parameters a denoted as Henry constants. As an example, consider the task of zone III of the TMB unit in Figure 22.7. In this zone, we demand for separation that component B travels with the solid phase while A moves with the fluid. This corresponds to requiring for the mass fluxes of the components transported by the fluid, n_ if , and by the solid phase, n_ is , respectively, that n_ Af > n_ As

and n_ Bf < n_ Bs

V_ f ,III ciIII

(22.9)

For these amounts hold ¼ and ¼ where V_ f ,III and V_ s are the volumetric flow rates of the liquid and the solid phases, respectively. Combining Eqs. (22.8), (22.9) provides n_ if

V_ f ,III cAIII > V_ s aA cAIII

V_ s qiIII ,

n_ is

and

V_ f ,III cBIII < V_ s aB cBIII

Dividing Eq. (22.10) by the concentrations cIII V_ f ,III > aA V_ s

and

i

(22.10)

and the solid flow rate V_ s leads to

V_ f ,III < aB : V_ s

(22.11)

These inequalities define the adjustable flow rate regions that lead to the separation of A and B in zone III as a function of their isotherm parameters aA and aB only. For designing the process, it is more convenient to apply dimensionless flow rates mj for each zone j: mj ¼

j V_ f , V_ s

j ¼ I…IV

(22.12)

These mj-values serve as the main design parameters. Defining corresponding tasks and inequalities to all four zones of the TMB process and requiring a non-negative feed flow rate, mIII  mII, leads to the following set of design constraints: mI  K B ,

(22.13a)

22.4 Continuous simulated moving bed (SMB) chromatography

K A  mII  mIII  K B ,

(22.13b)

mIV  K A :

(22.13c)

Here, in particular, the separation zones II and III are of interest. The inequality (22.13b) defines a triangular region for (mII, mIII) in which any chosen flow rate combination leads to a complete separation, provided the other inequalities are fulfilled. An example is shown in Figure 22.9. Operating inside of the triangle guarantees pure products, while outside, the purity of one or both outlets will be lower than 100%. The indicated optimal operating point lies on the vertex of the region. Furthermore, the figure contains examples of separation regions for the case of nonlinear Langmuir adsorption isotherms, Eq. (22.2) [33]. While the performance of the process in terms of productivity and eluent consumption improves with increasing the feed concentration, the region becomes smaller, which corresponds to an increased sensitivity to disturbances. An optimal feed concentration can be found making a compromise between robustness and performance.

3,5

m III

aB 3,0 cFi ↑

2,5

aA 2,0

1,5 1,5

2,0 aA

2,5 m II

3,0 aB

3,5

FIGURE 22.9 Design of SMB chromatography based on triangular regions of complete separation for the dimensionless flow rates in zones II and III of an ideal TMB process. Solid lines and filled symbol—separation region and optimal operating point for linear isotherms, Eq. (22.8). Dashed lines and open symbols—separation regions and optimal operating points for Langmuir isotherms, Eq. (22.2), for increasing feed concentrations cFi as determined according to [33]. Parameters: aA ¼ 2, bA ¼ 0.02 g/L, aB ¼ 3, bB ¼ 0.03 g/L, feed (inlet) concentrations cFA ¼ cFB ¼ 0, 1.0, 2.5, 5.0, 10.0, 20.0 g/L.

589

590

CHAPTER 22 Process concepts in preparative chromatography

Finally, the volumetric flow rates in the four zones and the switching time of the corresponding SMB process are found from the m-values using the following expression [33]: SMB V_ j ΔtS  V c ð1  εÞ mj ¼ , V c ð1  εÞ

j ¼ ðI…IV Þ,

(22.14)

where Vc denotes the volume of a single column. Extensions of this shortcut method have been established by Mazzotti and coworkers also for other isotherm types [34–38], and recently, a simplified method was given for favorable isotherms [39]. Furthermore, application is possible to design processes where the purity requirements are lower than 100%, as shown for linear [40] and Langmuir isotherms [41,42]. The operating parameters found by this shortcut procedure represent valuable initial estimates that can be applied as starting values for further model-based or experimental investigation and optimization of the process.

22.4.2 Improved SMB operating concepts Although the conventional SMB process described above is already a powerful technology, there exist a number of possibilities to further improve its performance or to broaden its scope of application. The periodic nature and its peculiar setup offer a multitude of options to interfere and to improve performance. Details can be found in several comprehensive reviews [43–45]. Here only a few relevant examples can be discussed. A rough classification of the major options suggested to enhance SMB processes is • • •



implementation of gradients of, for example, modulations of solvent composition, salt, temperature, or pH value; periodic manipulation of parameters such as flow rates, concentration, and column configuration; modification of the setup by, for example, enrichment steps, using fewer or more columns or zones, internal recycles, or combination with further separation or reaction steps; configurations with three outlets.

Process variants that combine one or more of the above modifications have also already been suggested. The gradient concept introduced in Section 22.3.1 for single-column chromatography can be applied also in SMB chromatography. This approach, originally suggested for separations using supercritical fluids [46], is capable of increasing productivity and product concentrations, as well as of reducing eluent consumption. In liquid chromatography, a gradient SMB process is easily accomplished by using mobile phases of different elution strengths for the feed and the desorbent, respectively [47]. Theoretical and experimental studies reveal that it is particularly attractive to apply a solvent as desorbent that has stronger elution strength than the solvent

Zone III

Extract (B)

Raffinate (A)

Zone IV

Eluent (strong solvent)

„weak solvent“

„strong solvent“

Concentraon of A and B

Zone II

Zone I

Feed (A + B) (weak solvent)

Concentraon of modifier

22.4 Continuous simulated moving bed (SMB) chromatography

A

B

I

II

III

IV

FIGURE 22.10 Gradient SMB chromatography using a two-step gradient exploiting a (non-adsorbing) modifier. Left—process set-up, right—illustration of the solvent strength profile developing in the apparatus (top) and resulting effect on the internal concentrations two components A and B (thin lines indicate profiles attained under conventional isocratic conditions).

containing the feed [48]. As illustrated in Figure 22.10, the resulting internal solvent strength profile leads to an enhanced desorption in zones I and II, while adsorption is favored in zones III and IV. An increasingly important application of gradient SMB chromatography is the separation of proteins by ion-exchange or hydrophobic interaction chromatography using salt gradients [49–51]. Other examples of successful application of gradient SMB are given in [52,53]. When designing a gradient SMB process, special care has to be taken on describing the dependency of the adsorption isotherm parameters on the modifier concentration (see Eq. 22.7). Furthermore, it has to be verified whether components of the solvent adsorb or not, since this is crucial for determining the internal solvent strength profiles. These information are required by the different suggested design methods, which range from extensions of the “triangle theory” introduced above [54] to applying more detailed simulation routines [55]. Another attractive option to develop enhanced SMB processes is to exploit directly their periodic nature by manipulating in an optimal manner the time course of certain operating parameters during each switching interval. Most of these concepts aim at optimizing the dynamic propagation behavior of the concentration bands within the unit, which is controlled by the non-linearity of the adsorption isotherms (see Section 22.2.1). There are various parameters that can be easily manipulated. Figure 22.11 shows a few selected examples. In the “Powerfeed” concept (Figure 22.11, left) [56,57], some or all internal flow rates are changed between sub-switching intervals. This leads to non-constant volumetric flowrates of the outlet streams. In the “Modicon” concept (Figure 22.11, middle) [58,59], the feed concentration is varied periodically. This directly affects the internal concentration-dependent migration velocities of the components. For systems with Langmuir isotherms, Eq. (22.2), it was found most beneficial to apply during each switching interval, first a low- and then a high-feed concentration. Figure 22.11

591

CHAPTER 22 Process concepts in preparative chromatography

Feed F1 (high conc.) Feed F2 (low conc.) R

X

E

Sub-switch 2 R

R

F

Sub-switch 1

Sub-switch 4

F

E

X

X

E

Feed conc.

VIII VII

ΔtS

Δ tS

Time

F2

F2

F1 ΔtS

Sub-switch

Sub-switch 3 Flow rate

592

F1 ΔtS

Time

1

2

ΔtS

3 4 1

2

ΔtS

3 4

Time

FIGURE 22.11 Selected improved concepts for SMB chromatography based on “super-periodic” variation of operating parameters. Top—schematic representations. F, R, X, and E denote feed, raffinate, extract, and eluent, respectively. Main manipulated parameters are marked by bold lines. Bottom—illustration of temporal profiles of the manipulated parameter(s) over two consecutive switching intervals. Left—“Powerfeed” operation with variation of flow rates (only flow profile for zones II and III shown). Middle—“ModiCon” process with modulation of feed concentration. Right—“Varicol” process based on asynchronous column shifting.

(right) shows the idea of an asynchronous column shifting regime, denoted as “Varicol” [60]. In contrast to conventional SMB processes, here the columns are switched not simultaneously but independently in an optimized fashion. A more recent variant, denoted as “Advanced Simulated Moving Bed (ASMB),” adopts a similar strategy by dividing the switching period of the classical SMB concept into five sub-intervals wherein the flows are active only in certain zones of the process [61]. Apart from the “super-periodic” processes discussed above, also rather simple changes of the process setup can create significant benefits. For example, Figure 22.12 (left) shows a simple three-zone, open-loop setup. Here, zone IV of the process is abandoned. This is of interest when a regeneration of the fluid phase is less important than preventing a possible breakthrough of weakly retained components, for example, in bioseparations or when using gradients. Also additional regeneration zones or cleaning in place (CIP) procedures can easily be added. Figure 22.12 (middle) illustrates a partial enrichment of the extract (“EE-SMB”) before its partial re-introduction into zone II [62,63]. The higher concentration established in this way causes a beneficial internal displacement effect (cf. Figure 22.3A). Figure 22.12 (right) shows the “Fraction-Feedback” concept (“FF-SMB”) [64]. Here, the insufficiently separated proportions of the raffinate stream are collected when they elute, and then recycled. This process has, thus, also a “super-periodic” character.

22.4 Continuous simulated moving bed (SMB) chromatography

Buffer tank F Zone II

F

F

R

R

R

Zone I

Zone III

X

X

E

E

Paral solvent removal

E

X

E

FIGURE 22.12 Selected improved concepts for SMB chromatography based on modification of the process setup. Left—simple three-zone open-loop setup for applications, where a regeneration of the fluid phase is abandoned due to specific requirements (see text). Middle—extract enrichment (EE) SMB with an increased recirculated extract concentration to enhance separation efficiency. Right—fraction-and-feedback (FF) SMB process that improves a partially successful separation by an internal recycle.

VIII IV

II

III

I

II I

A

III

IV A B C

A B C

C

VII

purge

VII

purge

VI B C

B

VIII

A

A B

V IV III II I

B

C

VI V

A

B

IV A B C

III II I

C

FIGURE 22.13 Selected options for performing ternary separation by SMB chromatography (according to [45,67]).

Finally, there exists a number of promising approaches to extent the application of SMB units beyond binary separations with two product streams, that is, to perform multi-component separations by SMB chromatography, see, e.g., [65–72]. Figure 22.13 shows three examples for the separation of ternary mixtures. The left of the figure shows a simple serial connection of two SMB processes. In the middle and right, additional zones are included that allow for an internal recycling of mixture streams, which corresponds to implementing a “two-in-one” SMB process. Regarding the latter concepts, one has to account for dilution effects which increase the volume of such recycle streams. This can necessitate purge streams as shown in the figure, or an additional solvent removal [67].

593

594

CHAPTER 22 Process concepts in preparative chromatography

FIGURE 22.14 (A) Schematic illustrations of a dividing wall distillation tower (left) and the analogous double-layer SMB configuration (right), which is capable of (pseudo-)ternary separations. Gray arrows indicate similarity of the corresponding constituents. Thick vertical line: dividing wall; A, B, and C: ternary mixture components; 3U (4U) and 3L (4L): divided upper and low layers, respectively. (B) Internal concentration profiles of the double-layer SMB unit at the end of port switching interval in steady-state. Arrows mark the positions of the feed inlet and the three outlets (extract, intermediate stream, and raffinate) for the next port-switching interval [71].

Further progress was achieved recently related to performing the so-called pseudo-ternary center cut separations, i.e., isolating a component traveling in the center of an elution train. Analogous to established dividing wall distillation (Figure 22.14 , left), which was developed for providing three or even more fractions, recently, a so-called double-layer SMB was developed [71,72], which is illustrated in Figure 22.14. A validated online optimization strategy is available to exploit the potential of this promising SMB process [73]. In the last years, a flexible alternative multi-column solvent gradient process (MCSGP) was introduced, which combines batch-wise carried our steps with continuous operation. The process is capable to efficiently isolate a component eluting somewhere within an elution train which contains a larger number of components [74]. Various other concepts were suggested for ternary and multi-component separations. For example, recently, a configuration was introduced which exploits just two separation columns [75] as illustrated in Figure 22.15. Such a two-column setup was also applied to capture a target component in one sub-step before recovering it in a regeneration step [76]. Besides the improved concepts discussed above, a number of further SMB concepts were proposed, like the “Partial feed” [77] or the “Selective withdrawal” [78]. Recently, a highly productive bypass SMB variant was suggested that can be efficiently applied if the purity requirements regarding the outlet stream can be relaxed [79]. Also the rational combination of chromatography with further alternative separation processes like crystallization [80,81] was shown to possess the potential to be beneficial.

22.5 Optimization and concept comparison

Interconnected substep Feed

1

2 Waste Batch substep

Feed at low flow

1 Waste

Eluon + CIP + equilibraon

2 Product Waste

FIGURE 22.15 Principle of two-column capture SMB processes according to [75]. Shown are the two substeps that comprise a single cycle. These are repeated after switching of the columns.

All the mentioned advanced SMB process concepts were demonstrated to be capable of achieving a significantly improved flexibility and performance in comparison to conventional SMB chromatography. A quantitative comparison of some of these concepts is given, for example, in [82]. However, despite their superior performance, the optimal design and implementation of such processes remains a challenging task. This can be solved using proper mathematical column models as the one introduced in Section 22.2 and by formulating an optimal control problem [56] or applying specialized optimization techniques [82,83]. A comprehensive review was recently provided [84].

22.5 Optimization and concept comparison As already mentioned above, model equations like Eq. (22.1) can be used to evaluate the performance of various configurations of arranging and operating chromatographic columns. In addition to classical elution, also recycling chromatography and the various variants of the SMB process can be described quantitatively. For example, to describe discontinuous closed-loop recycling, just the following condition at the column inlet has to be respected, while the recycle is active: ci ðz ¼ 0, tÞ ¼ ci ðz ¼ L, tÞ i ¼ 1, …, K

(22.15)

Since in the continuous SMB process, several columns j are connected in series, for all M columns (with the exception of the positions where feed or solvent are introduced) must hold the following boundary conditions: cj i ðz ¼ 0, tÞ ¼ cij1 ðz ¼ L, tÞ i ¼ 1, …, K, j ¼ 2, M

(22.16)

More details on formulating chromatographic process models can be found, for example in [5,6,85] and in a recently published book [86]. Recently, a very promising approach was presented allowing to compare batch and different continuous processes based on a unified theory [87].

595

596

CHAPTER 22 Process concepts in preparative chromatography

To evaluate the performance of chromatographic separation processes, frequently, the specific production rate PR and the recovery yield REC are applied as objective function: PR ¼

REC ¼

Amount of target component collected , Time  Amount of solid phase

Amount of target component collected Amount of the target component in the feed

Further, it might be important to consider a specific desorbent consumption DC: DC ¼

Amount of desorbent introduced Amount of product collected

The specific definitions of these performance criteria depend on the actual process concept. For details, the reader is referred to [88–90]. It can be also useful to consider combined objective functions, e.g., the product PR  REC or a goal function that considers different objectives in combination with weighting factors. Another approach is to apply multi-objective optimization, which provides more insight but is computationally more expensive [91]. Various reliable optimization techniques are available in order to find the process-specific free operating parameters. Most frequently, nonlinear programming methods are applied using, for example, sequential quadratic programming (SQP) methods (e.g., [41]). Increasingly used are also evolutionary algorithms [92]. An important and often neglected aspect is that when evaluating different process options, only optimized scenarios should be compared with each other. The progress in algorithms and computational power makes gradually a rational flowsheet design possible. New input can be expected from future applications of machine learning techniques to achieve faster process predictions. A recent example devoted to study multi-column pressure swing adsorption indicates the large potential [93]. Finally, a remark on the applicability of advanced process concepts should be given. An important rule is that a more sophisticated concept that exploits additional degrees of freedom should be applied only if it provides the chance for a substantial improvement. Otherwise, the simpler process concepts should be preferred. Besides identifying optimal operating parameters, it is also important to check the processes with respect to their robustness. Again, simple reliable processes should be favored. However, the knowledge and experience acquired in the last decades has clearly shown the potential of preparative chromatographic processes which go beyond the possibilities of standard elution. This holds true in particular for the SMB technology.

References

22.6 Conclusions Besides standard isocratic elution, there are various alternative process concepts in preparative chromatography available. The principles of the most important options were explained, including gradient elution, recycling chromatography, and simulated moving bed chromatography. In particular, the various variants of SMB chromatography offer the potential to develop efficient separation processes. It was pointed out that all the concepts mentioned can be quantitatively described exploiting basic models developed for the conventional standard process by adjusting the corresponding initial and boundary conditions and by installing correct connections in multi-column processes. This allows selecting and optimizing for a specific separation problem and the most suitable process concept. However, a necessary assumption for the application of this approach is the availability of validated parameters characterizing the chromatographic system under consideration.

Acknowledgments The essential contributions of numerous PhD students and our lab crew in Magdeburg are gratefully acknowledged. The authors are further grateful for the financial support of the European Union (Collaborative research projects IntEnant and CORE), the Deutsche Forschungsgemeinschaft (SFB 578 and SFB 1411, Project-ID 416229255), and Wissenschaftliche Ger€atebau Herbert Knauer GmbH, Berlin.

References [1] Ganetsos G, Barker PE. Preparative and production scale chromatography. New York: Marcel Dekker; 1993. [2] Cox GB, editor. Preparative enantioselective chromatography. Oxford: Blackwell Publishing; 2005. [3] Rhee H-K, Aris R, Amundson NR. First-order partial differential equations. Vol. 1 and 2. Englewood Cliffs: Prentice-Hall; 1986/1989. [4] Guiochon G. Preparative liquid chromatography. J Chromatogr A 2002;965:129–61. [5] Guiochon G, Felinger A, Shirazi DG, Katti AM. Fundamentals of preparative and nonlinear chromatography. 3nd ed. Amsterdam: Elsevier; 2020. [6] Schmidt-Traub H, editor. Preparative chromatography of fine chemicals and pharmaceutical agents. Weinheim: Wiley-VCH; 2005. [7] Colin H, Hilaireau P, Martin M. Flip-flop elution concept in preparative liquid chromatography. J Chromatogr A 1991;557:137–53. [8] Frenz J, Horvath CS. Displacement chromatography. In: Horvath CS, editor. Highperformance liquid chromatography-advances and perspectives, Vol. 5. New York: Academic Press; 1988.

597

598

CHAPTER 22 Process concepts in preparative chromatography

[9] Jandera P, Churacek J. Gradient elution in column liquid chromatography. Amsterdam: Elsevier; 1985. [10] Antia FG, Cs. Horvath. Gradient elution in non-linear preparative liquid chromatography. J Chromatogr 1989;484:1–27. [11] El Fallah Z, Guiochon G. Prediction of protein band profile in preparative reverse phase gradient elution. Biotechnol Bioeng 1992;39:877–85. [12] Gallant SR, Vunnum S, Cramer SM. Optimization of preparative ion-exchange chromatography of proteins: linear gradient separations. J Chromatogr A 1996;725:295–314. [13] Felinger A, Guiochon G. Optimizing experimental conditions in overloaded gradient elution chromatography. Biotechnol Prog 1996;12:638–44. [14] Wang A, Carr PW. Comparative study of the linear solvation energy relationship, linear solvent strength theory, and typical-conditions model for retention prediction in reversed-phase liquid chromatography. J Chromatogr A 2002;965:3–23. [15] Snyder LR, Dolan JW. High performance gradient elution. New York: Wiley; 2007. [16] Seidel-Morgenstern A. Preparative gradient ghromatography. Chem Eng Technol 2005;28:1265–73. [17] Antos D, Seidel-Morgenstern A. Continuous step gradient elution for preparative separations. Sep Sci Technol 2002;37:1469–87. [18] Bombaugh KJ, Dark WA, Levangie RF. High resolution steric chromatography. J Chromatogr Sci 1969;7:42–7. [19] Seidel-Morgenstern A, Guiochon G. Theoretical study of recycling in preparative chromatography. AIChE J 1993;39:809–18. [20] Heuer C, Seidel-Morgenstern A, Hugo P. Experimental investigation and modelling of closed-loop recycling in preparative chromatography. Chem Eng Sci 1995;50:1115–27. [21] Bailly M, Tondeur D. Recycle optimization in non-linear productive chromatography - I Mixing recycle with fresh feed. Chem Eng Sci 1982;37:1199–212. [22] Grill CM. Closed-loop recycling with periodic intra-profile injection: a new binary preparative chromatographic technique. J Chromatogr A 1988;796:101–13. [23] Schlinge D, Scherpian P, Schembecker G. Comparison of process concepts for preparative chromatography. Chem Eng Sci 2010;65:5373–81. [24] Charton F, Bailly M, Guiochon G. Recycling in preparative liquid chromatography. J Chromatogr A 1994;687:13–31. [25] Sainio T, Kaspereit M. Analysis of steady state recycling chromatography using equilibrium theory. Sep Purif Technol 2009;66:9–18. [26] Kaspereit M, Sainio T. Simplified design of steady-state recycling chromatography under ideal and nonideal conditions. Chem Eng Sci 2011;66:5428–38. [27] Siitonen J, M€antt€ari M, Seidel-Morgenstern A, Sainio T. Robustness of steady state recycling chromatography with an integrated solvent removal unit. J Chromatogr A 2015;1391:31–9. [28] Nimmig S, Kaspereit M. Continuous production of single enantiomers at high yields by coupling single column chromatography, racemization, and nanofiltration. Chem Eng Process Process Intensif 2013;67:89–98. [29] Broughton CB, Gerhold CG. US Patent 2 985 589; 1961. [30] Rodrigues A. Simulated moving bed technology: principles, design and process applications. Butterworth-Heinemann; 2015. [31] Subramanian G. Chiral separation techniques - a practical approach. Weinheim: WileyVCH; 2000.

References

[32] Mihlbachler K. Simulated moving bed chromatography - a promising alternative for the purification of biopharmaceuticals. In: Langer ES, editor. Advances in large-scale biopharmaceutical manufacturing and scale-up production. Washington: ASM Press; 2007. [33] Mazzotti M, Storti G, Morbidelli M. Optimal operation of simulated moving bed units for nonlinear chromatographic separations. J Chromatogr A 1997;769:3–24. [34] Mazzotti M, Storti G, Morbidelli M. Robust design of countercurrent adsorption separation processes: 2. Multicomponent systems. AIChE J 1994;40:1825–42. [35] Mazzotti M, Storti G, Morbidelli M. Robust design of countercurrent adsorption separation: 3. Nonstoichiometric systems. AIChE J 1996;42:2784–96. [36] Mazzotti M, Storti G, Morbidelli M. Robust design of countercurrent adsorption separation processes: 4. Desorbent in the feed. AIChE J 1997;43:64–72. [37] Migliorini C, Mazzotti M, Morbidelli M. Robust design of countercurrent adsorption separation processes: 5. Nonconstant selectivity. AIChE J 2000;46:1384–99. [38] Gentilini A, Migliorini C, Mazzotti M, Morbidelli M. Optimal operation of simulated moving-bed units for non-linear chromatographic separations: II. Bi-Langmuir isotherm. J Chromatogr A 1998;805:37–44. [39] Kaspereit M, Neupert B. Vereinfachte Auslegung der simulierten Gegenstromchromatographie mittels des Hodographenraums. Chem Ing Tech 2016;88 [in print]. [40] Rajendran A. Equilibrium theory-based design of simulated moving bed processes under reduced purity requirements: linear isotherms. J Chromatogr A 2008;1185:216–22. [41] Kaspereit M, Seidel-Morgenstern A, Kienle A. Design of simulated moving bed chromatography under reduced purity requirements. J Chromatogr A 2007;1162:2–13. [42] F€utterer M. Design of simulated moving bed Plants for reduced purities. Chem Eng Technol 2010;33:21–34. [43] Gomes PS, Minceva M, Rodrigues AE. Simulated moving bed technology: old and new. Adsorption 2006;12:375–92. [44] Seidel-Morgenstern A, Keßler LC, Kaspereit M. New developments in simulated moving bed chromatography. Chem Eng Technol 2008;31:826–37. [45] Kaspereit M. Advanced operating concepts for simulated moving bed Processes. In: Grushka E, Grinberg N, editors. Advances in chromatography. Boca Raton/Fla, USA: CRC Press, Taylor & Francis; 2009. p. 165–92. [46] Clavier J-Y, Nicoud R-M, Perrut M. A new efficient fractionation process: the simulated moving bed with supercritical eluent. In: von Rohr PR, Trepp C, editors. High Pressure Chemical Engineering. London: Elsevier; 1996. [47] Jensen TB, Reijns TGP, Billiet HAH, van der Wielen LAM. Novel simulated movingbed method for reduced solvent consumption. J Chromatogr A 2000;873:149–62. [48] Antos D, Seidel-Morgenstern A. Application of gradients in simulated moving bed processes. Chem Eng Sci 2001;56:6667–82. [49] Keßler LC, Gueorguieva L, Rinas U, Seidel-Morgenstern A. Step gradients in 3-zone simulated moving bed chromatography: application to the purification of antibodies and bone morphogenetic protein-2. J Chromatogr A 2007;1176:69–78. [50] Houwing J, Van Hateren SH, Billiet HAH, van der Wielen LAM. Effect of salt gradients on the separation of dilute mixtures of proteins by ion-exchange in simulated moving beds. J Chromatogr A 2002;952:85–98. [51] Wekenborg K, Susanto A, Schmidt-Traub H. Modelling and validated simulation of solvent-gradient simulated moving bed (SG-SMB) processes for protein separation. Comp Aided Chem Eng 2005;20:313–8.

599

600

CHAPTER 22 Process concepts in preparative chromatography

[52] Abel S, Mazzotti M, Morbidelli M. Solvent gradient operation of simulated moving beds. I. Linear isotherms. J Chromatogr A 2002;944:23–39. [53] Houwing J, Billiet HAH, van der Wielen LAM. Optimization of azeotropic protein separations in gradient and isocratic ion-exchange simulated moving bed chromatography. J Chromatogr A 2002;944:189–201. [54] Abel S, Mazzotti M, Morbidelli M. Solvent gradient operation of simulated moving beds. II. Langmuir isotherms. J Chromatogr A 2004;1026:47–55. [55] Beltscheva D, Hugo P, Seidel-Morgenstern A. Linear two-step gradient counter-current chromatography: analysis based on a recursive solution of an equilibrium stage model. J Chromatogr A 2003;989:31–45. [56] Kloppenburg E, Gilles ED. A new concept for operating simulated moving-bed processes. Chem Eng Technol 1999;22:813–7. [57] Zhang Z, Mazzotti M, Morbidelli M. PowerFeed operation of simulated moving bed units: changing flow-rates during the switching interval. J Chromatogr A 2003;1006:87–99. [58] Schramm H, Kaspereit M, Kienle A, Seidel-Morgenstern A. Simulated moving bed process with cyclic modulation of the feed concentration. J Chromatogr A 2003;1006:77–86. [59] Schramm H, Kienle A, Kaspereit M, Seidel-Morgenstern A. Improved operation of simulated moving bed processes through cyclic modulation of feed flow and feed concentration. Chem Eng Sci 2003;58:5217–27. [60] Ludemann-Hombourger O, Nicoud R-M, Bailly M. The “VariCol” process: a new multicolumn continuous chromatographic process. Sep Sci Technol 2000;35:1829–62. [61] Harada H, Suzuki K, Sato K, Okada K, Tsuruta M, Yajima T, Kawajiri Y. Process development for advanced simulated moving bed (ASMB) chromatography by parameter refinement using pilot plant experimental data. Sep Purif Technol 2022;281, 119932. [62] Bailly M, Nicoud RM, Adam P, Ludemann-Hombourger O. US Patent 2006124549; 2004. [63] Paredes G, Rhee H-K, Mazzotti M. Design of simulated-moving-bed chromatography with enriched extract operation (EE-SMB): Langmuir isotherms. Ind Eng Chem Res 2006;45:6289–301. [64] Keßler LC, Seidel-Morgenstern A. Improving performance of simulated moving bed chromatography by fractionation and feed-back of outlet streams. J Chromatogr A 2008;1207:55–71. [65] Wooley R, Ma Z, Wang N-HL. A nine-zone simulating moving bed for the recovery of glucose and xylose from biomass hydrolyzate. Ind Eng Chem Res 1998;37:3699–709. [66] Hashimoto K, Shirai Y, Adachi S. A simulated moving-bed Adsorber for the separation of tricomponents. J Chem Eng Jpn 1993;26:52–6. [67] Keßler LC, Seidel-Morgenstern A. Theoretical study of multicomponent continuous countercurrent chromatography based on connected 4-zone units. J Chromatogr A 2006;1126:323–37. [68] Chin CY, Wang N-HL. Simulated moving bed equipment designs. Sep Purif Rev 2004;33:77–155. [69] Hur JS, Wankat PC. New design of simulated moving bed (SMB) for ternary separations. Ind Eng Chem Res 2005;44:1906–13. [70] Kurup AS, Hidajat K, Ray AK. Comparative study of modified simulated moving bed systems at optimal conditions for the separation of ternary mixtures of xylene isomers. Ind Eng Chem Res 2006;45:6251–65.

References

[71] Lee JW. Expanding simulated moving bed chromatography into ternary separations in analogy to dividing wall column distillation. Ind Eng Chem Res 2020;59:9619–28. [72] Lee JW. Double-layer simulated moving bed chromatography for ternary separations: serialized layer configurations. Ind Eng Chem Res 2021;60:8911–26. [73] Lee JW, Kienle A, Seidel-Morgenstern A. On-line optimization of four-zone simulated moving bed chromatography using an equilibrium-dispersion model: I. Simulation study. Chem Eng Sci 225: 115810. II. Experimental validation. Chem Eng Sci 2020;226, 115808. [74] M€uller-Sp€ath T, Aumann L, Melter L, Morbidelli M. Chromatographic separation of three monoclonal antibody variants using multicolumn countercurrent solvent gradient purification (MCSPG). Biotechnol Bioeng 2008;100(6):1166–77. [75] Angarita M, M€uller-Sp€ath T, Baur D, Lievrouw T, Lissens G, Morbidelli M. Twincolumn CaptureSMB: a novel cyclic process for protein a affinity chromatography. J Chromatogr A 2015;1389:85–95. [76] Unger N, M€uller-Sp€ath T, Neustoecklin B, Aumann L, Bavand M, Morbidelli M. Affinity capture of F(ab‘)2 fragments using twin clolumn contercurrent chromatography. BioProcess International; 2015. p. 13. February. [77] Zang Y, Wankat PC. SMB operation strategy – partial feed. Ind Eng Chem Res 2002;41:2504–11. [78] Zang Y, Wankat PC. Three-zone simulated moving bed with partial feed and selective withdrawal. Ind Eng Chem Res 2002;41:5283–9. [79] Maruyama RT, Karnal P, Sainio T, Rajendran A. Design of bypass-simulated moving bed chromatography for reduced purity requirements. Chem Eng Sci 2019;205:401–13. [80] Kaspereit M, Gedicke K, Zahn V, Mahoney AW, Seidel-Morgenstern A. Shortcut method for evaluation and design of a hybrid process for enantioseparations. J Chromatogr A 2005;1092:43–54. [81] Amanullah M, Mazzotti M. Optimization of a hybrid chromatography-crystallization process for the separation of Tr€oger’s base enantiomers. J Chromatogr A 2006;1107:36–45. [82] Zhang Z, Mazzotti M, Morbidelli M. Continuous chromatographic Processes with a small number of columns: comparison of simulated moving bed with Varicol, PowerFeed, and ModiCon. Korean J Chem Eng 2004;21:454–64. [83] Kawajiri Y, Biegler LT. Large scale nonlinear optimization for asymmetric operation and design of simulated moving beds. J Chromatogr A 2006;1133:226–40. [84] Kawajiri Y. Model-based optimization strategies for chromatographic processes: a review. Adsorption 2021;27(1):1–26. [85] Guiochon G, Lin B. Modeling for preparative chromatography. Academic Press; 2003. [86] Nicoud R-M. Chromatographic processes: modeling, simulation, and design. Cambridge University Press; 2015. [87] Sainio T. Unified design of chromatographic processes with timed events: separation of ternary mixtures. Chem Eng Sci 2016;152:547–67. [88] Felinger A, Guiochon G. Comparing the optimum performance of the different modes of preparative liquid chromatography. J Chromatogr A 1998;796:59–74. [89] Heuer C, Kniep H, Falk T, Seidel-Morgenstern A. Comparison of various process engineering concepts of preparative chromatography. Chem Eng Technol 1998;21:469–77. [90] Seidel-Morgenstern A. Optimization and comparison of different modes of preparative chromatography. Analusis Mag 1998;26:M46–55.

601

602

CHAPTER 22 Process concepts in preparative chromatography

[91] Zhang Z, Hidajat K, Ray AK, Morbidelli M. Multiobjective optimization of SMB and Varicol process for chiral separation. AIChE J 2002;48:2800–16. [92] Zhang Z, Mazzotti M, Morbidelli M. Multiobjective optimization of simulated moving bed and Varicol processes using a genetic algorithm. J Chromatogr A 2003;989:95–108. [93] Subraveti SG, Li Z, Prasad V, Rajendran A. Machine learning-based multiobjective optimization of pressure swing adsorption. Ind Eng Chem Res 2019;58(44):20412–22.

CHAPTER

Modeling of preparative liquid chromatography

23

T. Fornstedt, P. Forssen, and J. Samuelsson Department of Engineering and Chemical Sciences, Karlstad University, Karlstad, Sweden

Preparative chromatography is the best generic method currently available for purifying small drugs and valuable chemical components at the > > i +F i +u ¼ Da , > > ∂t ∂x ∂x2 < ∂t 0  x  L, t  0, i ¼ 1, …, n, > > > C ð x0Þ ¼ C0,i , i > > : Ci ð0tÞ ¼ φi ðtÞ:

(23.1)

In the mass balance equation, Ci(x, t) is the mobile-phase concentration of component i at a distance x from the column inlet and at a time t after sample injection; qi is the stationary-phase concentration given by the adsorption isotherm, F is the volumetric phase ratio, u is the linear flow rate, Da is the apparent dispersion constant, and L is the column length. F can be expressed as follows: F¼

V S 1  εt ¼ , V0 εt

(23.2)

where VS and V0 are the stationary and mobile phase partial volumes and εt is the total porosity of the column. The apparent dispersion constant, Da, can be calculated from Da ¼

Lu , 2N

(23.3)

605

606

CHAPTER 23 Modeling of preparative liquid chromatography

where N is the number of theoretical plates. The last two lines in Eq. (23.1) are the initial and boundary conditions. The initial condition describes the mobile phase concentration, C0,i, at all positions x prior to the injection. Normally, C0,i ¼ 0, but as we see later, this is not always so. The boundary condition is the injection profile, φi, that is, the shape of the injected sample zone. LC separations can be simulated by numerically estimating the solution of Eq. (23.1) at the outlet Ci(L, t), that is, estimating the elution profiles. This column model can be solved using many different algorithms [36], the most common ones are the Rouchon finite-difference method [2] and the orthogonal collocation method [36–38].

23.3 Adsorption model Functions describing the relationship between the component concentrations in the mobile and stationary phases, at a specific and constant temperature (isothermal conditions), are called adsorption isotherms. Several adsorption isotherm models are available for describing single component as well as multicomponent systems at constant temperature.

23.3.1 Band shape dependence on adsorption In analytical LC, sample concentrations are normally very low and the corresponding adsorption isotherms are practically linear in this concentration range. This means that the adsorbed concentration is proportional to the concentration in the mobile phase. All molecules then migrate through the column, adsorb, and desorb independent of the other molecules, so each solute elutes as a Gaussian peak. The retention time of each peak depends on the initial slope of the corresponding adsorption isotherms. The peak shape deviates only slightly from the Gaussian ideal, but the chromatograms may become complex due to the multitude of solutes in the sample. In preparative LC, the concentrations are generally much higher as compared to analytical LC; therefore, the adsorption-isotherm curvature and saturation capacity have an enormous impact on the overloaded peak shapes. Molecules in highconcentration zones spend relatively more time in the mobile phase due to the difficulty of finding free adsorption sites. Because of this overload, the sample zone becomes asymmetrical, elongated, and strongly dependent on the shape of the adsorption isotherm. An adsorption isotherm can be classified according to the shape of the isotherm curves, see Figure 23.1. Most reported adsorption isotherms have a convex curvature, approaching a maximum adsorbed concentration, the saturation capacity. Such models are classified as Type I according to the IUPAC standard, see Figure 23.1A, left side. Type III adsorption isotherms are concave with an increasing slope at high concentrations, see Figure 23.1C, left side, whereas Type II isotherms are initially

23.3 Adsorption model

FIGURE 23.1 The most typical adsorption isotherms and the corresponding shapes of the elution profiles: (A) Type I, (B) Type II, and (C) Type III adsorption behavior.

convex but, after an inflection point, turn concave, see Figure 23.1B, left side. From Figure 23.1, it is further concluded that, if a Type I adsorption isotherm describes the adsorption process best, that is, a convex upward shape, then the overloaded eluted band has a sharp front and a diffusive rear, see Figure 23.1A, right side. The reason is that the higher eluted concentration strives for smaller retention times, because the higher the concentration, the smaller is the degree of adsorption. But, if a Type III adsorption isotherm describes the adsorption process best, that is, a concave upwards shape, then the overloaded eluted band has the opposite shape, that is, a diffusive front and a sharp rear, see Figure 23.1C, right side. The reason is that the higher eluted concentration strives for longer retention times, because the higher the concentration, the larger is the degree of adsorption. Type II adsorption isotherms are composed of both Type I and III and have vertical asymptotes that are unrealistic in LC, because this means the saturation capacity is unlimited, resulting in a most complex band shape, see Figure 23.1B, right side. Most reported liquid–solid adsorption processes are described with Type I adsorption isotherms. In a competitive situation, the adsorption of a component between the mobile and stationary phases depends not only on the local concentration of the component itself but also on all the other components. This ultimately results in complex chromatograms. Figure 23.2 shows the resulting preparative chromatogram after the injection of an equal mixture of two components, assuming Type I adsorption behavior. The first eluted component is displaced by the second eluted one with a mixed zone in between. We can also see that the second eluted component has a hump on its rear. This situation is much more advantageous for fractionation of component

607

608

CHAPTER 23 Modeling of preparative liquid chromatography

FIGURE 23.2 A typical binary elution profile, assuming Type I adsorption behavior.

1; however, it is of crucial importance to understand, predict, and control the competitive forces creating this situation. By using numerical simulation, we can predict the optimal experimental conditions for collecting pure amounts of component 1 or component 2.

23.3.2 Adsorption isotherms Adsorption isotherms describe the equilibrium distribution of solutes between the mobile and stationary phases, q(C), in a chromatography column. The nature of the interactions varies from system to system, so there are many adsorption isotherm models. Each model consists of a number of model parameters, which define the specific adsorption isotherm for the different components. If the adsorption isotherms can be measured and fitted to the appropriate model, a lot of information is obtained about system characteristics. Furthermore, it is then possible to perform computer simulations, such as by solving Eq. (23.1).

23.3.2.1 The Langmuir adsorption isotherm To understand the fundamentals of the adsorption mechanism at overloaded preparative conditions, it can be necessary to first study the adsorption of a single component in a chromatographic system. The Langmuir adsorption isotherm is a very simple Type I model. It assumes ideal solutions, homogeneous, and independent monolayer adsorption [39]: qi ðC1 , C2 , …, Cn Þ ¼

qs,i bi Ci ai Ci X X ¼ , 1+ bC 1+ bC j j j j j j

i, j ¼ 1, …, n:

(23.4)

In this expression, ai is the initial slope, bi is the equilibrium constant, and qs,i ¼ ai/bi is the saturation capacity of component i. It holds that ki ¼ Fai, so the

23.3 Adsorption model

retention times of the Gaussian peaks in analytical separations are given by the initial slope of the corresponding Langmuir adsorption isotherm. Figure 23.1A (left side) shows an example of a single-component Langmuir adsorption isotherm. Most separation problems of practical interest involve more than a single component and the Langmuir adsorption isotherm in Eq. (23.4) can be used to model the competition between the components. Here, the adsorbed concentration of any component i depends on the concentration of all components present. Because of this, all n mass-balance equations in Eq. (23.1) are coupled and cannot be treated independently. When the modifier dependence of the adsorption parameters is described by linear solvent strength theory [40,41], the competitive n-component Langmuir model for component i can be written [31] as follows: qi ðC1 , C2 , …, Cn , ϕÞ ¼

ai eSa,i ϕ Ci X , 1+ b eSb,j ϕ Cj j j

i, j ¼ 1, …, n,

(23.5)

where ϕ is the volume fraction of organic modifier in the mobile phase and S are empirical adsorption parameters that are determined experimentally. Eq. (23.5) was used here to simulate test data for our inverse method approach. Another common adsorption isotherm models that accounts for this experimental condition is the bi-Langmuir model [42], which is an empirical extension of the Langmuir model, with two Langmuir terms added to each other describing two different types of adsorption sites. The bi-Langmuir model applies to heterogeneous systems containing two separate types of adsorption sites. Examples are alkyl and silanol groups in C18 reversed-phase systems [43] and chiral selective and nonselective sites in chiral stationary phases [44]. The competitive n-component bi-Langmuir model, expanded to account for the modifier-dependence of the adsorption for component, i can be written [31]: qi ðC1 , C2 , …, Cn , ϕÞ ¼

aI,i eSaI ,i ϕ Ci aII,i eSaII ,i ϕ Ci X X + , SbI ,j ϕ 1+ b e Cj 1+ b eSbII ,j ϕ Cj j I,j j II,j

i, j ¼ 1, …, n, (23.6)

where subscripts I and II refers to the two adsorption sites. Eq. (23.6) has successfully been used to model the adsorption of a cyclohexanone/cycloheptanone mixture on a C18-column in gradient elution with methanol as the organic modifier [31]. More complex models of Type II and III exist, which accounts for lateral surface interactions, multilayer adsorption, adsorption energy distribution heterogeneity, are described elsewhere [2].

23.3.3 Determination of adsorption data There are several methods for determining the adsorption isotherm [2,5,45]. The most accurate technique today is frontal analysis [2,15], whereas the most recently developed method, the inverse method, is a better choice for process chromatography because it is fast [27–29].

609

610

CHAPTER 23 Modeling of preparative liquid chromatography

23.3.3.1 Frontal analysis Frontal analysis is usually carried out in a series of increasing concentration pulses [2,45]. The adsorption data for a single-component case is calculated by integrating the mass balance for those pulses: q¼C

VR  V0 , VS

(23.7)

where C and q are the solute concentrations in the mobile and stationary phase, and VR is the frontal breakthrough volume. FA can be used for any type of adsorption isotherm, is not affected by slow or concentration-dependent kinetics, and is therefore considered to be the most-accurate method for adsorption-isotherm determination. Due to these advantages, FA is more or less known as a reference method. For multicomponent cases, an intermediate plateau must be determined, which means that a fractionation and reinjection procedure must be followed for systems with more than two compounds. Unfortunately, it has been found that, for ternary mixtures, FA works only for high-efficiency separation systems [16]; otherwise, the erosion of the intermediate plateau is too pronounced. The major disadvantage with FA is that it is tedious and consumes a large amount of solvent and pure solute.

23.3.3.2 The inverse method With the IM, adsorption-isotherm parameters are determined from overloaded elution profiles (peak shapes at sample overload are treated in Section 23.3.1). The solute consumption and time requirements are modest compared to other methods. The adsorption isotherm cannot be obtained directly from this data (as opposed to FA data). Instead, the parameters are estimated by solving the inverse partial-differentialequation problem: Elution profiles are simulated iteratively, by solving Eq. (23.1) numerically, and the parameters are tuned by numerical optimization until the simulated and experimental profiles coincide in the least square sense. The IM is not as accurate as the FA method. However, in process chromatography, we are mainly interested in the column model’s ability, in combination with the determined adsorption isotherm, to predict elution profiles that later are going to be used for process optimization. If the determined adsorption-isotherm parameters or model is physicochemical correct or not is not a major concern in process chromatography, this makes IM a perfect candidate for adsorption-isotherm estimation.

23.4 Process optimization of preparative chromatography The goal of chromatographic processes optimization, in most cases, is to produce and manufacture a high-quality product as fast and cheaply as possible, and the optimization can be performed both empirically and numerically. The empirical process optimization approach requires extensive laboratory work to find the optimal conditions, which can be time-consuming.

23.4 Process optimization of preparative chromatography

23.4.1 Empirical optimization In the end, what matters to preparative chromatographers is not which model applies or the values of the model parameters. The most important aspect is how much sample (or process fluid) can be purified, if possible in a single injection otherwise in several, and how quickly a pure component can be produced. To characterize a system, one could calculate the maximal amount of substance that is possible to inject under the condition that the component bands, or profiles, are separated. This empirical quantity is called the loading capacity. Ideally, one should use saturated solutions of the components when doing these calculations. Here, often the flow rate is fixed at as high a value as allowed and only the injection volume is varied. For example, the injection volume is increased until the maximum cross-contamination of a component exceeds 1%, that is, the component bands are no longer separated. A better optimum can be reached by allowing greater cross-contamination, performing a series of experiments with different injection volumes, then plotting the measured yield and production rate. However, this requires that the individual component bands be measured or calculated from the total response. Notice that the component bands are allowed to overlap if the minimum yield constraint is set lower than 100%, that is, if the compound is readily available and cheap compared to the chromatographic process, highly overloaded overlapping bands [46,47] are preferable.

23.4.2 Numerical optimization The numerical approach requires considerably less laboratory work and the ability to generate fast simulations and analyze the results qualitatively and quantitatively [2]. The objective function in numerical process optimization, usually the productivity PR, is a function of the experimental conditions and depends on the adsorption mechanisms of the system, such as thermodynamics, mass-transfer rate, and dispersion. To simplify the problem, it is common to keep some process optimization parameters fixed in the objective function, that is, perform a “sub-optimization.” Then, it has to be decided which parameters to include in the optimization procedure and which ones should be kept fixed. Constraints are also usually present in the optimization problem, such as on the yield (Y) and purity (PU). Increasing the yield demand often lowers the productivity, and decisions about an acceptable limit must be made. The productivity (PR,i) is how much of component i is retrieved during one injection cycle per kg CSP, mCSP, and can be written as Z

tstop,i

FV PR,i ¼

Ci ðtÞdt

tstart,i

tcycle mCSP

,

(23.8)

where FV is the volumetric flow rate, tcycle is the cycle time, tstop,i and tstart,i are the end and start of the fraction collection for the i:th component. The yield, Yi, is defined as how much of the injected amount of component i is collected during one injection cycle. The purity, PU,i, is the amount of component i in the collected fraction as a percent of the total amount of all collected components.

611

612

CHAPTER 23 Modeling of preparative liquid chromatography

FIGURE 23.3 Flowchart describing the most important steps in numerical optimization of process chromatography.

23.4.2.1 General procedures Numerical optimization of batch-process chromatography can be represented by a workflow according to Figure 23.3. First, it is important to find a suitable experimental-separation system by thorough screening of the available stationary and mobile phases. Thereafter, the adsorption isotherms for the major components must be determined. The inverse method is one of the most convenient and rapid methods for the determination of competitive adsorption isotherms aimed at process chromatography. The heart of the method consists of defining and solving a column mass-balance model and the procedure of selecting an adequate adsorption-isotherm model, and initial parameter guesses. One then solves column mass-balance model and seek to maximize the overlap of simulated and experimental elution profiles to for a number of overloaded injections. Therefore, a set of proper elution profiles at varying loads must be provided. The inverse solver then produces a set of adsorption

23.4 Process optimization of preparative chromatography

isotherm parameters that best describes the system. Now, it is possible to use these parameters, together with measured van Deemter functions and process conditions of the large-scale separation system, to perform process optimization with a given objective function and constraints (see Figure 23.3) [2,48]. Decision variables in the objective function are typically injection volume, injection concentration, and flow rate. Constraints are typically the maximum allowed pressure and minimum purity and yield for the target component.

23.4.2.2 Numerical injection-volume optimization The empirical injection volume optimization just described can also be done by using computer simulations, if the adsorption-isotherm parameters and the number of theoretical plates, N, in Eq. (23.3) are measured for both components. This numerical approach requires no advanced optimization routines, for example, “gridding” can be used, i.e., dividing the injection volume range using a finite number of equidistant points and calculating the objective function and the constraints in each, the one with the maximum value of the objective function that also fulfills the constraints is the estimated optimal injection volume. In a pharmaceutical setting, the yield constraint is often set to 75% and the purity constraint to 99%.

23.4.2.3 Numerical full optimization Here, all relevant parameters are allowed to vary: injection volume, sample concentration, and flow rate. Full optimization is difficult to perform without computer simulations and requires more-advanced optimization algorithms. We often use a response-surface global-optimization algorithm for “costly” problems, such as TOMLAB [49], combined with a modified Nelder–Mead simplex algorithm [50] that supports inequality constraints. In the response-surface algorithm, a global optimum is sought; this is crucial, as an ordinary local optimization algorithm might get stuck in a local minimum far from the optimal solution.

23.4.3 Important operational conditions Several parameters are important to model correctly if a highly accurate model prediction is needed. In this section, we discuss the holdup volume, injection profiles, and the correct accounting for additives in the modeling procedure. In this section, we do not cover diluent-eluent mismatches such as due to pH effects [51,52], viscosity effects [53,54], or solvent strength effects [55,56].

23.4.3.1 Holdup volume Many articles have demonstrated the need to use the correct holdup volume (porosity) in the adsorption-isotherm determination [57–60]. All these studies were performed for single-component cases. However, from a process chromatographic point of view, it is more interesting to know how such an error affects the prediction of productivity. This was recently investigated by the determination of optimum experimental conditions using erroneous adsorption isotherms combined with a

613

CHAPTER 23 Modeling of preparative liquid chromatography

wrong holdup volume and applied in the true system to evaluate the objective functions and constraints: productivity, yield, and purity [61]. It was shown that, for underestimated holdup volumes, the purity requirements are fulfilled for only the second eluted component, whereas for overestimated holdup volumes, the process requirements are fulfilled for only the first eluted component. The decreased productivity is larger for overestimated holdup volumes than underestimated volumes.

23.4.3.2 Injection profiles Injection profiles are used as boundary conditions for solving the column model, and often, numerical process optimization is conducted using rectangular injection profiles instead of the “true” injection profile; see Eq. (23.1). The reason is that it is very time-consuming and tedious to determine the injection profiles for all the different operational conditions used in the numerical optimization. However, by assuming rectangular injection profiles, large errors are introduced. In Figure 23.4, an experimental injection profile is plotted together with the corresponding rectangular injection profile; the difference in shape is striking. The injection profile depends on the flow rate, injection volume, viscosity of the solvent, and the solute size [62]. The eroded injection profiles are mainly due to radial diffusion in the injection loop that transports solutes back-and-forth from fastermoving regions to the slower-moving regions in the in the parabolic flow. Since it is so tedious to determine all possible injection profiles in an optimization procedure,

1

0.8 Norm. resp.

614

0.6

0.4

0.2

0

0

0.5

1

1.5

Volume [mL]

FIGURE 23.4 An experimental injection profile (black line) of solute omeprazole injected in 600 μL into an eluent of pure methanol at flow rate 1 mL/min overlaid with the corresponding rectangular injection profile (gray line).

23.4 Process optimization of preparative chromatography

another approach is to determine, from a few measured injection profiles, a function that describes the flow-rate and injection-volume dependence of the injection. In [61,63], it was shown that the injection profile φi(t) in Eq. (23.1) can be described as a convolution of a Gaussian peak and an exponentially decaying pulse that has an initial constant part, the length of which is given by θ. Expressed in eluted volume, V, the convolution can be written as follows:      A 2V 0  2V + θ 2V 0 + 2V + θ pffiffiffi pffiffiffi erf CðV Þ ¼ + erf 2 2σ 2σ     2  (23.9) A 2V + 2V 0 + θ σ2 σ  2Vτ + 2V 0 τ + τθ pffiffiffi + 2 + ln erfc , + exp 2 τ 2τ 2στ

where A is the area of the injection profile (related to the amount injected); τ, σ, V0, θ are parameters; and erf and erfc are the error function and the complementary error function, respectively. Here, we have that V0, Vinj/σ, τ, and θ have a linear relationship with the injection volume, Vinj, for a constant flow rate. It is also possible to include the volumetric flow rate dependency by letting these linear relationship parameters depend on the flow rate.

23.4.3.3 Modeling additives Additives are mobile phase components that compete with the solute for the available column surface and are added to the eluent primarily to improve the peak shape and reduce the retention Additives are often used in the mobile phases of modern separation systems; this is especially the case for chiral separation systems. However, in almost all cases of numerical modeling of such preparative systems, the additives are neglected in the modeling. The reason is that the additive is often “invisible” to the detector, especially in such ranges, so that reliable measurements of their adsorption isotherms cannot be performed. However, this problem can be bypassed by using the inverse method. One then assumes that an “invisible” additive is present and the adsorption of it can be described by an adsorption isotherm function, such as the Langmuir function. Elution profiles are then measured for different additive levels and due to the effect on the visible eluted peaks, the adsorption isotherm parameters of the additive also can be estimated. For example, we used the inverse solver to characterize the adsorption behavior of the FMOC–allylglycine enantiomers on the quinidine-carbamate anion exchanger by estimating both the enantiomer and the additive (acetic acid) adsorption-isotherm parameters [10]. It was shown that a simulation based on adsorption-isotherm parameters estimated by the inverse method, neglecting the additive (acetic acid) in the mobile phase, fitted an experimental overloaded profile well. However, these adsorption-isotherm parameters failed to predict the experimental elution profile when using a mobile phase with a somewhat higher level of additive concentration (see Figure 3 in [10]). Then, we used the inverse solver and accounted for the additive concentration by using elution profiles from two additive levels and including the additive level in the

615

616

CHAPTER 23 Modeling of preparative liquid chromatography

(A)

(B)

FIGURE 23.5 Experimental (symbols) and predicted elution profiles using the adsorption parameters in Table 3 of [10] accounting for the additive: in (A) for a 350-μL injection using mobile phase 2 and in (B) for a 500-μL injection using mobile phase 3 in [10]. The solid line corresponds to the prediction based on Langmuir parameters, and the dotted line corresponds to the biLangmuir parameters. See Section 3 and Table 1 of [10] for experimental conditions. Figure was adapted from Figure 6 in ref. Arnell et al. [10].

model. With these parameters for the two enantiomers, as well as for the invisible additive, we successfully predicted the elution profiles for any additive levels inbetween. Figure 23.5 shows the experimental and predicted binary profiles for two intermediate mobile phases using two different but large injection volumes. Thereafter, we investigate how well the concept worked for process optimization of a real experimental system accounting for the additive [64]. In this context, it should be mentioned that, if the additive adsorption strength is larger than any of the injected components, very strange band shapes occur. Such profiles have been described in the literature since the 1990s [2] and the phenomenon has also been verified by computer simulations. The systems used at that time were often not of practical interested but were designed to provoke the strange effects. More recently, however, we found that the effects also take place in modern systems aimed at preparative chiral separations [65], and we can use the inverse solver approach to accurately simulate and predict cases where strong additives results in strange band shapes [66,67]. Recently, these effects on preparative as well as analytical band shapes—due to strong additives—were reviewed [68].

23.4.3.4 Modeling ion-pair reagent additives Ion-pair reagents (IPRs), on the other hand, are added to the eluent to increase the retention of charged compounds. The underlying mechanism has been discussed for many years, and two main models are as follows: (i) ion-pair formation between the solute and lipophilic IPR bonded to the stationary phase and (ii) ion-pair formation in the mobile-phase eluent, binding the complex to the non-polar stationary phase [69]. As implementing column models of either mechanism is complicated,

23.4 Process optimization of preparative chromatography

especially for compounds with many charges [70], IPRs are often neglected in the modeling [71]. Another reason for neglecting the IPR in a model is that it is often invisible to the detector, so reliable measurements of its adsorption isotherm are difficult to perform. ˚ sberg et al. studied three peptides under overloaded conditions by making A adsorption isotherm measurements in the presence of an IPR, i.e., trifluoroacetic acid (TFA), on an XBridge C18 column [70]. In their study, based on NMR experiments and the very weak adsorption of TFA to the stationary phase, ion pairs were assumed to form in the solution. In modeling, this was handled by (i) adding a reaction term for ion-pair formation in the solution to the right-hand side of Eq. (23.1) and (ii) having three mass balances, one each for the solute, the IPR, and the formed ion pair [70]. However, for multiple-charged compounds, the reaction terms are more complex, making it difficult to find good model parameters using IM. In a follow-up study, ˚ sberg et al. investigated peptide separation on a charged-surface hybrid C18 colA umn containing a small number of positively charged groups [71]. In this case, TFA adsorbed to the surface. Two models were investigated, one considering the ˚ sberg et al. found that the TFA comadsorption of TFA and the other neglecting it. A petition could be neglected without loss of predictive power of the model. Again, if the aim is just to model elution profiles, this simplification is probably acceptable. Recently, Lesko et al. studied the overloaded elution profiles of organic compounds containing a sulfonic acid functional group and using tetrabutylammonium bromide as an IPR [72]. In their study, it was also noted that in certain IPR concentration ranges, deformed U-shaped elution zones appeared, whereas in other ranges, more classical Langmuirian elution zones appeared. To model this, an electrostatic modified multilayer adsorption model was derived, consistent with dynamic ionexchange models. The model was able to predict all these strange elution profiles over a broad IPR concentration range. However, from a preparative modeling perspective, it was difficult to determine the parameters of this complicated adsorption model using IM, as in this case, the multilayer adsorption model contained seven constants as well as being an implicit function. Therefore, if one is operating the separation system in an IPR concentration range well away from that causing deformation to appear, simpler models will probably have acceptable predictive power.

23.4.3.5 Modeling gradient elution In gradient elution, the IM is often used for isocratic experiments and for gradient elution on constant isocratic plateaus [32]. However, one can directly form gradient data by estimating the adsorption estimate parameters of Eqs. (23.5), (23.6) directly from overloaded elution profiles in gradient-mode LC [30,31]. The latter method will allow the estimation of parameters for solutes such as peptides and proteins whose retention factors strongly depend on the mobile phase composition. For these solutes, using the IM on constant isocratic plateaus is not feasible at low plateau levels. In two studies, we validated and applied the inverse solver and also introduced a strategy for determining adsorption data for preparative gradient separations. The first study concerned the one-component case [30] and the second the competitive two-component

617

618

CHAPTER 23 Modeling of preparative liquid chromatography

case [31]. In the latter case, Eq. (23.6) was successfully used to model the adsorption of a cyclohexanone/cycloheptanone mixture on a C18 column in gradient elution mode with methanol as the organic modifier. The calculations were validated by comparing the simulated data with experimental data from gradient separation experiments other than those used in the extended IM (see Figure 23.6). Note that when using the IM with the bi-Langmuir adsorption model in Eq. (23.6), 8n parameters need to be estimated for an n-component case. In this context, we have shown that it is difficult to accurately estimate the parameters even for a two-component case and that simpler models, such as the Langmuir model in Eq. (23.5) with 4n parameters, sometimes had better overall predictive power [73]. Preparative separations in smaller amounts (i.e., 0.1–10 mg) will be most important for therapeutic oligonucleotides, a new class of important therapeutic modalities targeting genomic expression and considered next-generation drugs by the pharmaceutical community [74]. The purity of oligonucleotides post synthesis is highly dependent on their length, sequence, and other chemical modifications, but a rough estimate is that for a 20-unit long oligonucleotide synthesized at 98.5% coupling efficiency the purity is about 75% [75]. Preparative chromatographic techniques used for purifying oligonucleotides include both silica- and organic-resin-based ion-exchange LC and silica-based reversed-phase LC with or without IPRs [75,76]. Whatever the column phase system, however, gradient elution will be required, so new ways of optimizing gradient elution methods will be crucial in the future.

FIGURE 23.6 Comparison between predicted (solid lines) and experimental (dashed lines) elution profiles at (left figure) 1% and (right figure) 4%/min gradient slopes for a mixture of cyclohexanone (first peak) and cycloheptanone (second peak). Sample concentration is 0.3 M and injection volume is 400 μL. Figure was adapted from Figure 4 in ref. A˚sberg et al. [31].

23.5 Case example

23.5 Case example Enantiomeric separation of omeprazole has been extensively studied regarding both product analysis and preparation using several different chiral stationary phases. We recently made a full optimization of the preparative purification of R- and S-omeprazole using columns packed with amylose tris(3,5-dimethyl phenyl carbamate) on 10 μm particles. The resulting optimal chromatogram can be seen in Figure 23.7 below. The thick gray lines are the experimental and the thick black line is the simulated chromatogram, thin lines are simulated for R- and S-omeprazole, and symbols are fractions taken during the elution. As one can see, the model predicts the process very well with an error in the cut point of approximately only 5 s.

S

C [g/L]

6

Sum S R Cut

4

2

0

R

C [g/L]

4 3 2 1 0 3

4

5

6

t [min.] FIGURE 23.7 Overlay of experimental and simulated elution profiles for the optimal conditions on 10 μm Kromasil AmyCoat 25  0.46 cm column: (top figure) S- and (bottom figure) R-omeprazole. The solid lines are experimental UV signals representing the sum of the enantiomers, symbols are analyzed fractions, and the dashed horizontal lines represent the cut points.

619

620

CHAPTER 23 Modeling of preparative liquid chromatography

Acknowledgments This work was supported by the Swedish Knowledge Foundation via the project “Improved Methods for Process and Quality Controls using Digital Tools” (grant number 20210021) and by the Swedish Research Council (VR) in the project “Fundamental Studies on Molecular Interactions aimed at Preparative Separations and Bio-specific Measurements” (grant number 2015-04627). ˚ sberg for doing the gradient experiments We also would like to thank PhD Dennis A presented in this third edition.

References [1] Helfferich FG, Klein G, editors. Multicomponent chromatography: theory of interference. Dekker; 1970. 427 p. [2] Guiochon G, Shirazi DG, Felinger A, Katti AM. Fundamentals of preparative and nonlinear chromatography. 2nd ed. Boston, MA: Academic Press; 2006. xiv, 975 p. [3] FDA’S policy statement for the development of new stereoisomeric drugs. Chirality 1992;4(5):338–40. [4] Francotte ER. Enantioselective chromatography as a powerful alternative for the preparation of drug enantiomers. J Chromatogr A 2001;906(1–2):379–97. [5] Samuelsson J, Arnell R, Fornstedt T. Potential of adsorption isotherm measurements for closer elucidating of binding in chiral liquid chromatographic phase systems. J Sep Sci 2009;32(10):1491–506. [6] Marchetti N, Cavazzini A, Pasti L, Dondi F. Determination of adsorption isotherms by means of HPLC: adsorption mechanism elucidation and separation optimization. J Sep Sci 2009;32(5–6):727–41. [7] Asnin LD, Guiochon G. The adsorption of Naproxen enantiomers on the chiral stationary phase (R,R)-Whelk-O1 under reversed-phase conditions: the effect of mobile phase composition. J Chromatogr A 2010;1217(17):2871–8. [8] Zabkova M, Zabka M, Rodrigues AE. Separation of racemic chiral drugs using immobilized CHIRALPAK IA: methodology for preparative scale development. Sep Sci Technol 2009;44(2):275–303. [9] Mihlbachler K, De Jesu´s MA, Kaczmarski K, Sepaniak MJ, Seidel-Morgenstern A, Guiochon G. Adsorption behavior of the ()-Tr€ oger’s base enantiomers in the phase system of a silica-based packing coated with amylose tri(3,5-dimethyl carbamate) and 2propanol and molecular modeling interpretation. J Chromatogr A 2006;1113(1):148–61. [10] Arnell R, Forssen P, Fornstedt T, Sardella R, L€ammerhofer M, Lindner W. Adsorption behaviour of a quinidine carbamate-based chiral stationary phase: role of the additive. J Chromatogr A 2009;1216(16):3480–7. [11] Seidel-Morgenstern A, Kessler L, Kaspereit M. New developments in simulated moving bed chromatography. Chem Eng Technol 2008;31(6):826–37. [12] Pia˛tkowski W, Kramarz R, Poplewska I, Antos D. Deformation of the gradient shape as a result of preferential adsorption of solvents in mixed mobile phases. J Chromatogr A 2006;1127:187–99. € [13] Tiselius A, Claeson S. Adsorption analysis by interferometric observation. Ark FOR KEMI MINEALOGI OCH Geol 1942;15B(18):1–6.

References

[14] James DH, Phillips CSG. The chromatography of gases and vapours. Part III The determination of adsorption isotherms. J Chem Soc 1954;1066–70. [15] Lindholm J, Forssen P, Fornstedt T. Validation of the accuracy of the perturbation peak method for determination of single and binary adsorption isotherm parameters in LC. Anal Chem 2004;76(16):4856–65. [16] Lisec O, Hugo P, Seidel-Morgenstern A. Frontal analysis method to determine competitive adsorption isotherms. J Chromatogr A 2001;908(1–2):19–34. [17] Bl€umel C, Hugo P, Seidel-Morgenstern A. Quantification of single solute and competitive adsorption isotherms using a closed-loop perturbation method. J Chromatogr A 1999;865(1–2):51–71. [18] Forssen P, Lindholm J, Fornstedt T. Theoretical and experimental study of binary perturbation peaks with focus on peculiar retention behaviour and vanishing peaks in chiral liquid chromatography. J Chromatogr A 2003;991(1):31–45. [19] Lindholm J, Forssen P, Fornstedt T. Validation of the accuracy of the perturbation peak method for determination of multicomponent adsorption isotherm parameters in LC. Anal Chem 2004;76(18):5472–8. [20] Glueckauf E. Theory of chromatography. Part 10.—Formulæ for diffusion into spheres and their application to chromatography. Trans Faraday Soc 1955;51:1540–51. [21] Ravald L, Fornstedt T. Theoretical study of the accuracy of the elution by characteristic points method for bi-Langmuir isotherms. J Chromatogr A 2001;908(1–2):111–30. [22] Samuelsson J, Fornstedt T. Injection technique for generating accurate adsorption isotherm data using the elution by characteristic points method. Anal Chem 2008;80(20): 7887–93. [23] Samuelsson J, Undin T, T€orncrona A, Fornstedt T. Improvement in the generation of adsorption isotherm data in the elution by characteristic points method - the ECP-slope approach. J Chromatogr A 2010;1217(46):7215–21. [24] Dose E, Jacobson S, Guiochon G. Determination of isotherms from chromatographic peak shapes. Anal Chem 1991;63(8):833–9. [25] James F, Sepu´lveda M, Charton F, Quin˜ones I, Guiochon G. Determination of binary competitive equilibrium isotherms from the individual chromatographic band profiles. Chem Eng Sci 1999;54(11):1677–96. [26] Zhang L, Selker J, Annie Q, Velayudhan A. Numerical estimation of multicomponent adsorption isotherms in preparative chromatography: implications of experimental error. J Chromatogr A 2001;934(1):13–29. [27] Felinger A, Cavazzini A, Guiochon G. Numerical determination of the competitive isotherm of enantiomers. J Chromatogr A 2003;986(2):207–25. [28] Arnell R, Forssen P, Fornstedt T. Accurate and rapid estimation of adsorption isotherms in liquid chromatography using the inverse method on plateaus. J Chromatogr A 2005;1099(1–2):167–74. [29] Forssen P, Arnell R, Fornstedt T. An improved algorithm for solving inverse problems in liquid chromatography. Comput Chem Eng 2006;30(9):1381–91. ˚ sberg D, Lesko M, Enmark M, Samuelsson J, Kaczmarski K, Fornstedt T. Fast estima[30] A tion of adsorption isotherm parameters in gradient elution preparative liquid chromatography. I: the single component case. J Chromatogr A 2013;1299:64–70. ˚ sberg D, Lesko M, Enmark M, Samuelsson J, Kaczmarski K, Fornstedt T. Fast estima[31] A tion of adsorption isotherm parameters in gradient elution preparative liquid chromatography II: the competitive case. J Chromatogr A 2013;1314:70–6.

621

622

CHAPTER 23 Modeling of preparative liquid chromatography

[32] Marchetti N, Dondi F, Felinger A, Guerrini R, Salvadori S, Cavazzini A. Modeling of overloaded gradient elution of nociceptin/orphanin FQ in reversed-phase liquid chromatography. J Chromatogr A 2005;1079(1–2):162–72. [33] De Jong AWJ, Kraak JC, Poppe H, Nooitgedacht F. Isotherm linearity and sample capacity in liquid chromatography. J Chromatogr A 1980;193(2):181–95. [34] Petkovska M, Seidel-Morgenstern A. Nonlinear frequency response of a chromatographic column. Part I: application to estimation of adsorption isotherms with inflection points. Chem Eng Commun 2005;192(10–12):1300–33. [35] Zhong G, Fornstedt T, Guiochon G. Profiles of large-size system peaks and vacancy bands in liquid chromatography I. Analytical solution of the ideal model. J Chromatogr A 1996;734(1):63–74. [36] Guiochon G, Lin B. Modeling for preparative chromatography. 1st ed. Amsterdam ; Boston: Academic Press; 2003. 342 p. [37] Villadsen J, Michelsen ML. Solution of differential equation models by polynomial approximation. Englewood Cliffs, NJ: Prentice-Hall; 1978. 446 p. [38] Kaczmarski K, Mazzotti M, Storti G, Morbidelli M. Modeling fixed-bed adsorption columns through orthogonal collocations on moving finite elements. Comput Chem Eng 1997;21:641–60. [39] Langmuir I. The constitution and fundamental properties of solids and liquids. Part I. solids. J Am Chem Soc 1916;38(11):2221–95. [40] Snyder LR, Dolan JW, Gant JR. Gradient elution in high-performance liquid chromatography : I. theoretical basis for reversed-phase systems. J Chromatogr 1979;165(1):3–30. [41] Dolan JW, Gant JR, Snyder LR. Gradient elution in high-performance liquid chromatography : II. Practical application to reversed-phase systems. J Chromatogr 1979;165(1): 31–58. [42] Graham D. The characterization of physical adsorption systems. I. The equilibrium function and standard free energy of adsorption. J Phys Chem 1953;57(7):665–9. [43] Samuelsson J, Franz A, Stanley BJ, Fornstedt T. Thermodynamic characterization of separations on alkaline-stable silica-based C18 columns: why basic solutes may have better capacity and peak performance at higher pH. J Chromatogr A 2007;1163(1–2):177–89. [44] Enmark M, Samuelsson J, Undin T, Fornstedt T. Characterization of an unusual adsorption behavior of racemic methyl-mandelate on a tris-(3,5-dimethylphenyl) carbamoyl cellulose chiral stationary phase. J Chromatogr A 2011;1218(38):6688–96. [45] Seidel-Morgenstern A. Experimental determination of single solute and competitive adsorption isotherms. J Chromatogr A 2004;1037(1–2):255–72. [46] Jacobson SC, Guiochon G. Experimental study of the production rate of pure enantiomers from racemic mixtures. J Chromatogr A 1992;590(1):119–26. [47] Ziomek G, Antos D, Tobiska L, Seidel-Morgenstern A. Comparison of possible arrangements of five identical columns in preparative chromatography. J Chromatogr A 2006;1116(1):179–88. [48] Degerman M, Westerberg K, Nilsson B. A model-based approach to determine the design space of preparative chromatography. Chem Eng Technol 2009;32(8):1195–202. [49] Bj€orkman M, Holmstr€om K. Global optimization of costly nonconvex functions using radial basis functions. Optim Eng 2000;1(4):373–97. [50] Lagarias J, Reeds J, Wright M, Wright P. Convergence properties of the Nelder-Mead simplex method in low dimensions. SIAM J Optim 1998;9(1):112–47. [51] Samuelsson J, Forssen P, Fornstedt T. Sample conditions to avoid pH distortion in RP-LC. J Sep Sci 2013;36(23):3769–75.

References

[52] Streel B, Ceccato A, Chiap P, Hubert P, Crommen J. Injection-generated solvent and pH gradients for sample enrichment on injection of large volumes in microcolumn liquid chromatography. Biomed Chromatogr 1995;9(6):254–6. [53] Shalliker RA, Samuelsson J, Fornstedt T. Sample introduction for high performance separations. TrAC Trends Anal Chem 2016;81:34–41. [54] Samuelsson J, Shalliker RA, Fornstedt T. Viscosity contrast effects in analytical scale chromatography - evidence and impact. Microchem J 2017;130:102–7. [55] Jandera P, Guiochon G. Effect of the sample solvent on band profiles in preparative liquid chromatography using non-aqueous reversed-phase high-performance liquid chromatography. J Chromatogr A 1991;588(1–2):1–14. ˚ sberg D, Shalliker A, Samuelsson J, Fornstedt T. A closer study of peak [56] Enmark M, A distortions in supercritical fluid chromatography as generated by the injection. J Chromatogr A 2015;1400:131–9. [57] Sajonz P. Influence of the column hold-up time measurement accuracy on the prediction of chromatographic band profiles. J Chromatogr A 2004;1050(2):129–35. [58] Gritti F, Guiochon G. Systematic errors in the measurement of adsorption isotherms by frontal analysis: impact of the choice of column hold-up volume, range and density of the data points. J Chromatogr A 2005;1097(1–2):98–115. [59] Samuelsson J, Sajonz P, Fornstedt T. Impact of an error in the column hold-up time for correct adsorption isotherm determination in chromatography: I. Even a small error can lead to a misunderstanding of the retention mechanism. J Chromatogr A 2008;1189 (1–2):19–31. [60] Samuelsson J, Zang J, Murunga A, Fornstedt T, Sajonz P. Impact of an error in the column hold-up time for correct adsorption isotherm determination in chromatography: II. Can a wrong column porosity lead to a correct prediction of overloaded elution profiles? J Chromatogr A 2008;1194(2):205–12. [61] Samuelsson J, Enmark M, Forssen P, Fornstedt T. Highlighting important parameters often neglected in numerical optimization of preparative chromatography. Chem Eng Technol 2012;35(1):149–56. [62] Samuelsson J, Edstr€om L, Forssen P, Fornstedt T. Injection profiles in liquid chromatography. I. A fundamental investigation. J Chromatogr A 2010;1217(26):4306–12. [63] Forssen P, Edstr€om L, Samuelsson J, Fornstedt T. Injection profiles in liquid chromatography II: predicting accurate injection-profiles for computer-assisted preparative optimizations. J Chromatogr A 2011;1218(34):5794–800. [64] Forssen P, Edstr€om L, L€ammerhofer M, Samuelsson J, Karlsson A, Lindner W, et al. Optimization strategies accounting for the additive in preparative chiral liquid chromatography. J Chromatogr A 2012;1269:279–86. [65] Arnell R, Forssen P, Fornstedt T. Tuneable peak deformations in chiral liquid chromatography. Anal Chem 2007;79(15):5838–47. [66] Forssen P, Arnell R, Kaspereit M, Seidel-Morgenstern A, Fornstedt T. Effects of a strongly adsorbed additive on process performance in chiral preparative chromatography. J Chromatogr A 2008;1212(1–2):89–97. [67] Forssen P, Arnell R, Fornstedt T. A quest for the optimal additive in chiral preparative chromatography. J Chromatogr A 2009;1216(23):4719–27. [68] Fornstedt T, Forssen P, Westerlund D. System peaks and their impact in liquid chromatography. TrAC Trends Anal Chem 2016;81:42–50. [69] Ion CT, Chromatography P. Crit Rev Anal Chem 2008;38(3):161–213.

623

624

CHAPTER 23 Modeling of preparative liquid chromatography

˚ sberg D, Langborg Weinmann A, Leek T, Lewis RJ, Klarqvist M, Lesko M, et al. [70] A The importance of ion-pairing in peptide purification by reversed-phase liquid chromatography. J Chromatogr A 2017;1496:80–91. ˚ sberg D, Lesko M, Leek T, Samuelsson J, Kaczmarski K, Fornstedt T. Estimation of [71] A nonlinear adsorption isotherms in gradient elution RP-LC of peptides in the presence of an adsorbing additive. Chromatographia 2017;80(6):961–6. [72] Lesko M, Samuelsson J, Kaczmarski K, Fornstedt T. Experimental and theoretical investigation of high-concentration elution bands in ion-pair chromatography. J Chromatogr A 2021;1656, 462541. ˚ sberg D, Enmark M, Samuelsson J, Fornstedt T, Kaczmarski K. Choice of [73] Lesko M, A model for estimation of adsorption isotherm parameters in gradient elution preparative liquid chromatography. Chromatographia 2015;78(19–20):1293–7. [74] Roberts TC, Langer R, Wood MJA. Advances in oligonucleotide drug delivery. Nat Rev Drug Discov 2020;19:673–94. [75] Paredes V, Aduda KL, Ackley H. Cramer. In: Chackalamannil S, Rotella D, Ward SE, editors. Comprehensive medicinal chemistry III: Manufacturing of oligonucleotides. Oxford: Elsevier; 2017. p. 233–79. € [76] Enmark M, Bagge J, Samuelsson J, Thunberg L, Ornskov E, Leek H, Lime F, Fornstedt T. Analytical and preparative separation of phosphorothioated oligonucleotides: columns and ion-pair reagents. Anal Bioanal Chem 2020;412:299–309.

CHAPTER

Capillary electrochromatography

24

Susanne K. Wiedmer and Marja-Liisa Riekkola Department of Chemistry, University of Helsinki, Helsinki, Finland

24.1 Introduction The early work by Arne Tiselius in the 1930s laid a ground for capillary electromigration techniques that have developed into efficient liquid-phase separation techniques suitable for the separation of a large variety of compounds ranging in size from small ions to large biomolecules. The widespread field of applications of capillary electromigration techniques is mainly based on the availability of different subtechniques. Of these, capillary electrochromatography (CEC), which can be also carried out under nonaqueous media, has been under continuous development over the years. The methodology is an excellent alternative to conventional liquid chromatography. However, although among the different capillary electromigration techniques, only 5% of the published research on characterization and quality control of pharmaceuticals between 1994 and 2014 was carried out by CEC [1], great activities in the development of new stationary phases will most probably guarantee an increase in the popularity of CEC, especially in the industry [2,3]. The starting point for the development of CEC can be trailed back to the early works by Strain [4] in 1939 and Lecoq in 1944, when the successful exploitation of an electroosmotic flow (EOF) in chromatography was introduced. Harold H. Strain showed in his work that the resolution of a mixture of water-soluble compounds could be enhanced by applying an electric potential to both ends of the column. However, V. Pretorius in 1974 and B.J. Hopkins and J.D. Schieke were the first to demonstrate the ability to exploit the EOF to push analyte zones through a chromatographic column [5]. The electrophoresis community had to wait for additional seven years until Jorgenson and Lukacs published their work in 1981 on the CEC separation of some aromatic compounds (9-methylantracene and perylene) in a 170-μm capillary column packed with 10 μm of octadecylsilane particles [6]. Although this work can be seen as groundbreaking, it was just studies carried out by J.H. Know and I.H. Grant in the late 80s and early 90s that accelerated the interest in CEC [7,8]. Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00017-5 Copyright # 2023 Elsevier Inc. All rights reserved.

625

626

CHAPTER 24 Capillary electrochromatography

In this chapter, the basic idea of CEC is described, as well as the working principle, and the instrumentation is briefly introduced with the main focus on the stationary phases and the detection techniques utilized. Finally, some general applications on CEC in miniaturized systems are demonstrated.

24.2 Principles of capillary electrochromatography CEC is a hybrid technique that combines the chromatographic and electrophoretic separation mechanisms. Compared with liquid chromatographic methods, much higher separation efficiencies in terms of plate numbers per meter can be obtained, which is one of the main benefits of the technique. The reason for this is the successful exploitation of EOF generated in the capillary when a high voltage is applied. The plug-like flow profile generated in the capillary is beneficial for creating sharp peak profiles as compared with the parabolic ones obtained in pressure-assisted liquid chromatographic techniques. Although in practice, this means that there is no need for an extra pump to transfer the injected sample toward the detector, EOF rates are limited, and therefore, the flow can be assisted by means of pressure. The pressureassisted CEC approach leads to slightly distorted flow profiles, but this disadvantage is compensated by the possibility to control flow rates within a much wider range, leading to shorter analysis times. For the CEC separation of neutral compounds, the separation mechanism is solely based on interactions with the stationary phase, and the electrophoretic part of the separation mechanism is only linked to the creation of the EOF. For charged compounds in addition to the electrophoretic mobility of the compound, the EOF mobility and the chromatographic interactions with the stationary phase will affect the separation. The importance of the EOF mobility cannot be overlooked in CEC separations. In open capillaries, the EOF mobility (μeo, open) is dependent on the thin double layer on the surface of the capillary and is expressed as μeo,open ¼

ε ε0 ζ w , η

(24.1)

where ε is the dielectric constant of the medium, ε0 is the permittivity of vacuum, ζ w is the zeta potential of the capillary wall, and η is the viscosity of the mobile phase. In practice, EOF mobility can be calculated from the migration time of an uncharged marker compound using the following equation: μeo,open ¼

Ldet Ltot : teo,open V

(24.2)

Here, Ldet is the distance from the inlet end of the capillary to the detection point, Ltot is the total length of the capillary, teo,open is the migration time of the EOF marker compound, and V is the applied voltage.

24.3 Instrumentation

In packed capillary columns, the EOF through the column is affected by the type of the stationary phase employed. The EOF mobility in the interstices of a packed column is defined as μeo,open ¼

ε ε0 ζs , η

(24.3)

where ζs is the zeta potential at the surface of the packing material. The equation is also valid for the EOF mobility through the pores (if the packing particles are porous) of the packing particles when the particle diameter is large enough (typically larger ˚ for low electrolyte concentrations) as compared with the thickness of the than 300 A double layer. The actual interstitial mobility of a neutral EOF marker through the packed segment of the capillary can be defined as μeo,packed ¼

L2e t0,packed Vpacked

¼

τ2 L2packed , t0,packed Vpacked

(24.4)

where Le is the equivalent length of the packed segment, τ stands for tortuosity of the packing, and τ is dependent on the column architecture and is defined as τ ¼ Le/Lpacked. The processes that take place in the interstitial space are the mass transfer processes of convection and diffusion. These are the factors that have the strongest impact on the efficiency (H) in CEC, and they can be described with the simplified van Deemter equation: H¼A+

B + Cu, u

(24.5)

where A, B, and C are constants for specific running conditions used in the CEC system. The u term stands for the EOF velocity in the system. The A term (also called the eddy diffusion term) relates to the contribution of nonlaminar flow profiles in the packed column, without any extra-column effects on the plate height. The B term represents band broadening resulting from longitudinal diffusion of the sample component and will have a significant effect on the plate height only at very low flow velocities. The B term is negligibly small under common CEC operating conditions. The C term is related to the mass transfer resistance by the sample components in the retention process, based on the compounds’ distribution between the mobile and the stationary phase. For CEC columns packed with porous stationary-phase particles, the C term strongly depends on the kinetics of both the intraparticle mass transfer and the film diffusion.

24.3 Instrumentation The CEC instrumentation, as illustrated in Figure 24.1, includes the following parts: (a) a separation capillary and a thermostat system; (b) a high-voltage power supply; (c) electrodes immersed in the vials containing background electrolyte (BGE)

627

628

CHAPTER 24 Capillary electrochromatography

FIGURE 24.1 Schematic illustration of an instrumentation used for capillary electrochromatography.

solution; (d) a detector; and (e) a data-handling system. The most crucial part of the instrumentation is the separation capillary. By careful selection of the capillary material, stationary phase, and capillary column dimensions, separation selectivity can be controlled. Different stationary phases are described in more detail in the following section. The temperature at which the separation takes place can be adjusted by carefully controlling the temperature around the capillary. The temperature is either adjusted by liquid or more commonly by air flow. Proper temperature control in capillary electromigration techniques is the most important because during the electrophoretic separation process, a part of the electrical energy is released as Joule heat. Since thermal heat is not released totally to the environment, part of the energy is absorbed by the liquid inside the capillary, resulting in an increased temperature in the capillary, leading to decreased viscosity of the background electrolyte (BGE) solution and, through enhanced current of the system, increased heat. Eventually, the temperature increase might result in the formation of vapor bubbles due to boiling of the BGE solution, and in breakdown of the current (and loss of separation). Without appropriate heat dissipation, the plug-like flow profile will be lost, resulting in a decrease in the separation efficiency and resolution. The thermal energy (Qv; W cm3) in the capillary is dependent on the applied field (E; V/cm), the equivalent conductivity (Λ; m2 Ω1 mol1) of the BGE solution, and the buffer concentration (c; mol L1), and it can be calculated by Eq. (24.6): Qv ¼ E2 Λ c

(24.6)

24.3 Instrumentation

The heat inside the capillary can reach values of hundreds of watts per cubic centimeter, and therefore, active cooling is important, especially in electrophoretic separations using conventional field strengths higher than 200 V/cm. In general, analyte mobilities increase approximately 2–3% per degree Celsius. Analyte band broadening effects are crucial in electrophoretic separations, and under common operation conditions, the heat release per unit volume is around 1500 times larger in electro-driven systems than in pressure-driven systems. High-voltage power supply in commercial instruments is limited to +/ 30 kV. However, with home-built instruments even higher than 120 kV separation potentials have been used, see, for example [9–12]. Such ultrahigh-voltage capillary electrophoresis devices allow faster separations and improved separation efficiencies and resolution. Platinum electrodes that are immersed into the BGE vials are either close to or surrounding the ends of the separation capillary. In most commercial instruments, on-capillary column detection is carried out by UV–vis spectroscopy. A small section of the protecting polyimide layer is gently removed, and UV–vis or fluorescence detection takes place through the capillary. However, for other types of detection, the capillary is inserted into a special cartridge, or some other technical arrangements are required for on-line coupling of the capillary column, for example, to mass spectrometry, laser-induced fluorescence, or nuclear magnetic resonance spectrometry. The hyphenation of CEC to mass spectrometry is introduced later on in this section. Data handling is typically conducted on-line, and all commercial CE systems have integrated easy-to-use automated programmable software systems.

24.3.1 Injection The typical sample volumes injected into the narrow bore capillary columns are in the nanoliter range. Samples are introduced hydrodynamically by applying an external pressure of typically 8–12 bar and/or electrokinetically by applying a voltage of +/ 1 to 30 kV for a certain time (often between 10 and 200 s). The electrokinetic injection mode is the most common type of injection in CEC. When the sample is electrokinetically injected, both the electroosmotic flow and electrophoretic mobility of the sample (in case of charged analytes) will influence the amount of injected sample. Since the nature of the analyte (charge/size) will affect its electrophoretic mobility, selective analyte injection is obtained using the electrokinetic mode. The nature of the sample matrix has an influence on the amount of injected sample if electrokinetic injection is applied; however, hydrodynamic injection is less influenced by the type of sample. For both injection modes, the injected quantity (Qi) can be calculated from the length of the injected plug (Li), the analyte concentration (ci) in the injected sample, the capillary diameter (dc), the linear velocity of the injection plug (vi), and the sample injection time (ti) according to Eq. (24.7): Qi ¼ Li ci

π dc 2 π dc 2 ¼ vi ti ci : 4 4

(24.7)

629

630

CHAPTER 24 Capillary electrochromatography

24.3.2 Stationary phases Capillary columns in CEC are generally divided into packed columns and opentubular columns. The packed columns are further divided into particle-packed columns and monolithic packings [13]. Much attention has been put on the development of novel CEC stationary phases either for specific applications or for the purpose of separating a specific group of analytes. The common trend among these CEC columns is that they all are of a capillary format and contain stationary phases that withstand high electric field strengths utilized during the electrophoretic separation process. A schematic overview of the three different types of capillary columns used in CEC are shown in Figure 24.2.

24.3.2.1 Packed columns: Particle-packed columns Packed columns in CEC are either based on particle packings or on in situ formed monolithic packings. The packing materials should have charged functional groups in order to provide an EOF and to enable electrostatic interactions with the analyzed compounds. Particle packings have by far been the most used ones. In principle, they are prepared as conventional LC columns, usually employing 50- to 100-μm-inside diameter fused silica capillaries. Typical packing materials are CEC Hypersil C18

FIGURE 24.2 A schematic presentation of the different column types used in CEC. (A) Open-tubular CEC, (B) monolithic packed CEC, and (C) particle-packed CEC. Printed with permission from Xue et al. [13]. Copyright John Wiley and Sons.

24.3 Instrumentation

and Spherisorb octadecyl-silica phases, which belong to reversed-phase materials. The large number of free silanol groups on the silica surface after packing assures a sufficient EOF, which, in many cases, is highly beneficial. The most common packing procedure is slurry packing, where a suspension of the packing material is introduced into a capillary with a mechanical frit at the end of the capillary. The particles, which typically have diameters of 3–5 μm and narrow size distribution, are suspended in an organic solvent, and in case of a reversed-phase packing material, the solvent is typically acetone or hexane. After the slurry has been filled into the capillary, the organic solvent is removed by rinsing with an aqueous solution, and the retaining frits are sintered by heating the packing material. In some cases, different frit materials can be used rather than the chromatographic material and then a plug of the frit material is first introduced into the capillary. It is important to note that the detection window is prepared some millimeters away from the outlet frit in order to have an open section without any stationary phase for detection. Other types of frits are available; for example, tapers have been fabricated on the fused silica capillary column in order to retain the particulate stationary-phase material. Both internally and externally tapered columns have been fabricated, but the main disadvantage of these frits is the demanding process for preparing the frits. In addition, the externally tapered columns suffer low stability due to their very fragile nature. The internally tapered frit columns are less fragile, but the decreased inner diameter might sometimes result in blockage of the capillary. A schematic of a typical slurry packing procedure is shown in Figure 24.3 [14]. However, in addition to slurry, other packing methods are available, and these include the use of carbon dioxide for packing, electrokinetic packing, packing by centripetal forces, and packing by gravity. Although the slurry packing method is the most commonly used method for preparing particle-packed CEC columns, the theoretical plates (efficiencies) of model compounds have often been higher when other techniques described before have been used. However, these packing methods need special skills and experience, and the challenges in packing capillaries and fabricating frits can be seen as the Achilles’ heel of the methodology. Poor frit fabrication will lead to bubble formation in the capillary and to decreased separation efficiencies. There are several types of stationary phases available for particle-packed columns, but even though octadecylsilane (ODS; silica-C18)-based columns are the most frequently used packed columns, octylsilane (silica-C8) and phenyl silica columns, in addition to a large number of different chiral and ion-exchange phases, have been quite commonly employed. Depending on the functionalization of the stationary-phase materials, the separation mechanism will be based on reversedphase, ion-exchange, or mixed-mode interactions. Both porous and nonporous ODS particles have been employed as reversed-phase packing materials. The mobile phases are common buffered aqueous solutions including organic modifiers such as acetonitrile, methanol, and tetrahydrofuran. The correlation between the logarithmic retention factor of neutral compounds and the percentage of organic modifier has been linear for eluents containing

631

632

CHAPTER 24 Capillary electrochromatography

(B)

(C)

Heating element

Flush

(D)

Temporary frit 2nd

1st

Pressure (H2O)

(E)

Silica material

(A)

Inlet frit

(F)

Heating element

Outlet frit

Detection window

FIGURE 24.3 Schematic of the preparation of a particle-packed CEC capillary column using the slurry packing method. The fabrication process includes introduction of the silica material into the capillary by tapping (A), resulting in a short plug of the frit material is in the capillary (B) and sintering the material by heating (C). The excess of silica material is rinsed out of the capillary after the first frit has been formed (D), and the packed capillary is rinsed with water under pressure to form the other frit by heating (E). Finally, the capillary is ready (F) with inlet and outlet frits keeping the packing material in place, and with a detection window after the outlet frit. Reprinted with permission from Colo´n et al. [14]. Copyright Elsevier Science Publishers.

acetonitrile and methanol. This trend is also typical in LC separations; however, we should keep in mind that the changes in the eluent viscosity affect the EOF mobility in CEC. Packed ODS columns offer excellent platforms for the separation of a wide range of drugs and pharmaceuticals, biomolecules, and natural products. Moreover, several studies have focused on enantioseparations using capillaries packed with chiral stationary phases [15]. The ion-exchange stationary phases are divided into strong cation-exchange (SCX) and strong anion-exchange (SAX) phases. The SCXs, which typically contain sulphonic acid groups, are negatively charged at pH values between 2 and 9. The separation mechanism is mainly based on electrostatic interactions between the charged analytes to be separated and the functional groups of the stationary phase, due to the rather low hydrophobicity of typical SCX stationary phases. The SAX stationary phases usually include quaternary amino groups, which result in a reversed EOF mobility due to the positively charged surface. The separation of negatively charged compounds is based on an ion-exchange mechanism. Applications include the separation of inorganic ions, lanthanides, amino acids, and proteins.

24.3 Instrumentation

24.3.2.2 Packed columns: In situ formed monolithic columns An alternative to particle-packed capillary columns are monolithic columns. These can be seen as continuous bed columns of either silica- or polymeric-based materials. As with the particle-packed columns, the separation of analytes in monolithic CEC is based on the interactions with the solid-phase monolithic material and the difference in the electrophoretic mobilities of the analytes. Here again, the driving force affecting analyte elution from the capillary in CEC is the EOF created by the electric field in the capillary column. The EOF is often supported by pressure (pCEC) applied from the inlet side of the capillary in order to increase the flow of the mobile phase and also to prevent bubble formation inside the capillary column. Even though there has been a great increase in the development of novel monolithic stationary phases for LC and CEC, monoliths cannot be considered replacements of traditionally particle-packed columns, but more as complementary alternatives. However, as recent publications have demonstrated, polymeric monoliths are step-by-step replacing particle-packed columns in large biomolecule separations. The main advantages of monolithic packed columns as compared with particle-packed columns include their easy fabrication, flexibility in surface modification, high loading capacity, good permeability and peak capacity, and no need for separate frits. Another important aspect is that monolithic columns offer great selectivity toward many kinds of compounds, even with mobile phases compatible with mass spectrometric detection. Regarding monolithic packings for CEC, acrylamide- and methacrylate-based polymeric materials seem to be the most popular ones. These materials offer great possibilities for tailor-made modifications based on the requirements of analysis. Silicabased monolithic capillaries are less used. Enantioseparations by monolithic CEC capillaries have also attracted much attention, and typical chiral stationary phases are based on proteins, macrocyclic antibiotics, polysaccharide derivatives, and cyclodextrins as chiral selectors [15]. To maintain the protein conformation, care must be taken when immobilizing them into the monolithic material. Various types of immobilization strategies have been developed, including sol–gel encapsulation, physical adsorption, and covalent bonding. Due to a large variety of functionalities on macrocyclic antibiotic chiral stationary phases, separation mechanisms can be based on hydrogen bonding, inclusion, dipole stacking, steric interaction, electrostatic, and π-π interactions. Polysaccharide derivatives have been rather popular chiral selectors as well, mainly due to their broad enantioselectivity, excellent efficiency, stability, and easy derivatization of the functional hydroxyl groups. Cyclodextrins have been frequently used as chiral selectors, both in free solution in capillary electrophoresis and liquid chromatography and bonded, immobilized, or linked to stationary phases. Cyclodextrins are cyclic oligosaccharides with six, seven, or eight D-glucose units bound via α-1,4-glycosidic linkages. These are named α-, β, and γ-cyclodextrin, respectively. The cyclodextrins form inclusion complexes with analytes, and if underivatized, hydroxyl groups on the surface of cyclodextrins provide additional hydrogen bonding interactions. Often the surface hydroxyl groups are modified, resulting in other types of interaction mechanisms such as dipole–dipole and π-π-bonding effects. The size of the cavity of the cyclodextrin

633

634

CHAPTER 24 Capillary electrochromatography

molecule is dependent on the type of cyclodextrin used, that is, α-, β-, or γ-cyclodextrin, of which native or derivatized β-cyclodextrin is most commonly employed in packed CEC.

24.3.2.3 Open-tubular columns The first successful open-tubular CEC separation was published in 1982 [16], in which an octadecylsilane (C18/ODS)-coated 30-μm-internal diameter capillary was used for the separation of some benzene derivatives and polyaromatic hydrocarbons. In open-tubular CEC, the inner wall of the capillary column is coated with the stationary phase, and the middle of the column is filled with the mobile phase. Therefore, the phase ratio, that is, the ratio of the volume of the stationary phase to the volume of the mobile phase, is much lower than that in the case of columns packed with particles or with the monolithic material. This is one of the main drawbacks of the methodology because this will greatly reduce the retention time window and selectivity. However, on the other hand, the biggest advantage of open-tubular CEC is that the technique does not suffer air bubble formation sometimes occurring in packed columns, or around the frits used for keeping the stationary phase in place in packed capillary columns. The biggest challenge in open-tubular CEC has been to prepare capillary stationary phases that homogeneously coat the inner wall of the capillary without any leakage. Several approaches have been developed, and these are divided into (a) chemical modification of the capillary wall, (b) noncovalent coatings, and (c) layer-by-layer coatings [17]. In the first type of coatings, the stationary phase is covalently bound to the fused silica capillary through chemical interactions. Chemical modification can be performed by in situ preparation approaches, either by in situ polymerization or by filling the capillary column with preformed stationary-phase materials. Another option is to carry out postpolymerization steps in order to attach the stationary phase covalently to the capillary wall. Nanomaterials such as nanoparticles comprising gold nanoparticles or polymeric particles can also be chemically attached either by in situ preparation methods or through postmodification approaches. Graphenebased nanomaterials have attracted much attention during the last decade [18]. Graphene, which is an allotrope of elemental carbon, is a planar two-dimensional arrangement of monolayers of carbon atom. Graphene should not be mixed up with the term graphite, which is the three-dimensional structure of carbon atoms. For chromatographic applications, graphene is typically in the form of graphene oxide, which is obtained by oxidation of graphite using a strong water-free mixture of concentrated sulfuric acid, nitric acid, and potassium permanganate. Graphene-based (or more correctly, graphene oxide-based) nanomaterials can be chemically attached to silica capillaries, and such stationary phases can also be exploited in open-tubular CEC separations. They are typically prepared through in situ condensation reactions or by postmodification approaches. Noncovalent stationary-phase coatings are typically easier to prepare, but they suffer poor stability. The approach includes physical adsorption of charged stationary-phase material through electrostatic or hydrophilic interactions. To note

24.3 Instrumentation

is that the stationary phase is not covalently attached to the fused silica wall and can therefore be easily replaced by another coating. Noncovalent stationary-phase coatings are often divided into static and dynamic coatings. In static coatings, the interactions between the stationary-phase material and fused silica capillary are taken place through strong electrostatic or hydrophilic interactions. For example, static polydopamine coatings have been prepared through selfpolymerization of dopamine with ammonium persulfate as the oxidant. Semipermanent coatings are another type of noncovalent coatings, which typically need regeneration of the stationary-phase material before each run. If higher stability is achieved, the rinsing interval can be decreased, and the columns allow many repeatable runs without serious leakage of the stationary-phase material. There are many polymer-based stationary phases of this type, as well as biomolecule-based stationary phases comprising, for example, of cellulose, lipoproteins, liposomes, or proteins. In dynamic coatings, the stationary-phase material such as surfactants or polyelectrolytes is added to the mobile phase, and a dynamic adsorption equilibrium is established with the capillary wall during the electrophoretic run. The advantage of the dynamic coating process is that a fresh coating is always available and that the background electrolyte modifier can be easily replaced by another substance. However, the main disadvantage is that this type of methodology cannot be typically hyphenated using a mass spectrometer due to the nonvolatile characteristics of the BGE solution. The third type of open-tubular CEC coating is based on layer-by-layer assemblies in which the stationary-phase material is sequentially attached to the fused silica capillary inner surface, and typically, the layers are linked together either through a chemical linkage or through electrostatic interactions. One specific motivation for developing layer-by-layer coatings is that separation selectivity can be increased by making the stationary phase thicker by employing several layers of material or by enhancing the phase ratio using porous polymers. Another benefit is the stability of the coating that can be often improved, as compared with single-layer coatings. In Figure 24.4, a specific type of layer-by-layer assembly for open-tubular CEC is presented [19]. The separation mechanisms in open-tubular CEC are based on reversed-phase, ion-exchange, hydrophilic, mixed-mode, or chiral interactions. They are very similar to those exploited in liquid chromatography; however, the electrophoretic mobility of the analytes will then also contribute to the separation. In the reversed-phase mode, interactions are mainly based on hydrophobicity and aromatic π-π interactions. Ion-exchange interactions take place between analytes of opposite surface charge to that of the stationary phase. Regarding hydrophilic interactions in opentubular CEC, the interactions are much like those described elsewhere in the book (see Hydrophilic interaction liquid chromatography) and are especially important for highly polar analytes. Mixed-mode separations take place in columns with stationary phases containing both reversed-phase and ion-exchange functional groups. Such open-tubular CEC columns are typically prepared from block–copolymer

635

CHAPTER 24 Capillary electrochromatography

(a)

(b)

Bare

PDA/Au NP@capillary

DOPA

AuNPs

200

thiols –S

HAuCl4

1 2

CH3

PDA

160

Monolayer

repeat

4

5

1,2,3 AuNPs

-S-(CH2)10SH

3

D

-S-(CH2)10SH

mAU

636

PDA

-S-(CH2)10S-

120 4

C

5

80

B

thiols

1,2,3,4,5 40

-S-(CH2)10S-

-S-(CH2)10S-

–S

A

CH3 0

PDA/Au NP/thiol@capillary

Multi-layer

0

3

6

9

12

15

18

Time (min)

FIGURE 24.4 (A) Illustration of the preparation of a multilayer coating based on polydopamine/gold nanoparticles/thiols (PDA/Au NPs/thiols) for open-tubular CEC separation. (B) Effect of the number of stationary-phase layers on the separation of alkylbenzenes; (1) benzene; (2) methylbenzene; (3) ethylbenzene; (4) propylbenzene; (5) n-butylbenzene. In capillary A, there is one layer; in B, two layers; in C, four layers; and in D, six layers of PDA/Au NPs/thiols. Reprinted with permission from Li et al. [19]. Copyright Royal Society of Chemistry.

mixtures. Over the years, there has been a great interest in developing chiral stationary phases for open-tubular CEC, and one of the most common stationary phases are those based on cyclodextrins that recognize and separate chiral compounds via the host–guest interactions that occur between the compound and the chiral molecule. The cavity size of cyclodextrin and the charge of the cyclodextrin-based stationary phase are the most crucial factors affecting the enantioseparation.

24.3.3 Detection UV–vis is still predominantly used in on-column arrangements, even though a wide range of other detection possibilities, such as fluorescence, conductivity, infrared, nuclear magnetic resonance, and electrochemical detection, are available. When packed columns are used, on-line detection is preferably performed in the unpacked segment of the capillary. The main drawback is the short optical path length, which results in reduced detection sensitivity with respect to the analyte concentration. In addition, UV–vis detection allows analyte identification only by simple comparison of UV spectra of the analyzed sample and standard analyte. Such an approach is useful and applicable only for a limited number of compounds containing chromophore groups; therefore, simple UV–vis detection cannot be exploited as the single detector in many applications. In order to improve the detection sensitivity, specially designed high-sensitivity detection cells, like the bubble detection cell or the z-shaped detection cell, can be utilized. Another option is to use on-line

24.3 Instrumentation

preconcentration techniques, where the sample analytes are concentrated inside the capillary by various stacking techniques.

24.3.3.1 Mass spectrometry The main reason for hyphenating CEC to MS is the possibility to improve the sensitivity and selectivity of detection that the methodology offers [20]. The first publication on the coupling of CEC to a mass spectrometer was published in 1991 [21], and ever since the technique has slowly started to gain the popularity, many different ionization techniques are available, but electrospray ionization (ESI) is by far the most commonly used ionization technique in combination with CEC and other capillary electromigration techniques. The widespread availability of ESI-MS instruments in laboratories contributes to the gradual replacement of UV-vis detectors by MS. However, the type of MS used is generally laboratory-specific without any emphasis on the analyzed compounds. CEC-MS interfaces. In the first CEC-MS application in 1991, a continuous flow fast atom bombardment (FAB) source was utilized [21]. In FAB, a beam of highenergy atoms bombards a sample surface to create ions. With this soft ionization technique, a wide range of different types of polar and charged compounds within the range of 300–6000 Da could be identified. The sample that is dissolved in a matrix will be bombarded by high-energy atoms (typically Ar or Xe), resulting in a mixture of cluster ions, adduct ions, analyte ions, matrix ions, matrix modifiers, and impurities. The two most commonly used matrices are glycerol and mnitrobenzyl alcohol. The main disadvantage of the technique is the high chemical background and challenges in maintaining a stable electrical current. In the beginning of the 1990s, other ionization sources started to appear, and in 2002, John B. Fenn and Koichi Tanaka shared the prestigious Nobel Prize in chemistry for their lifelong development of soft desorption ionization methods for MS analyses of biological macromolecules [22]. As already mentioned, MS detection with ESI is the most frequently used interface in capillary electromigration methods. In ESI, gas-phase ions are created from the sample solution by the use of a strong electric field between the tip of the spraying capillary and the orifice of the MS instrument. A Taylor cone is formed, and through constant evaporation of the solvent surrounding the analytes, smaller and smaller droplets are formed, finally resulting in the release of gas-phase analyte ions into the MS analyzer. ESI is characterized as a soft ionization technique as little analyte fragmentation takes place; however, structural information on compounds can be accomplished through the use of tandem ESI-MS or high-resolution ESI-MS. The main advantage of ESI is that multicharged compounds can be generated, which means that high-molar mass compounds can be detected using mass analyzers with a restricted m/z range [23,24]. The low flow rate (less than 1 μL/min; typically in the nL/min range) in CEC capillaries is generally insufficient for providing a stable electrospray under typical ESI conditions, which requires flow rates in the range of 1–10 μL/min. Therefore, the nebulization process in ESI can be accelerated by the use of a sheath gas or a sheath liquid. The three most typical ESI interfaces are the coaxial sheath flow interface, the

637

638

CHAPTER 24 Capillary electrochromatography

FIGURE 24.5 Illustration of CE-MS interfaces to an ESI source. (A) Coaxial sheath flow interface; (B) sheathless interface; (C) liquid junction interface. Printed with permission from Gelpı´ [26]. Copyright John Wiley and Sons.

liquid junction interface, and the sheathless interface [25,26]. These are illustrated in Figure 24.5. In the coaxial sheath flow interface, the mobile phase from the CEC capillary is mixed with a sheath liquid and sheath gas [27]. The benefit of this approach is that the effect of the MS unfriendly compounds in the CEC mobile phase is diminished since the characteristics of the sheath liquid will dominate. However, on the other hand, this mixing process might also result in decreased concentration sensitivity of the CEC-separated compounds. In the liquid junction interface, there is a narrow gap (typically 10–25 μm) between the CEC separation capillary and the ESI emitter [28]. This liquid compartment can be used to connect the voltage and to introduce additional liquids for improving the spraying process. The method is especially attractive if the CEC mobile phase contains compounds, such as nonvolatile substances, that hinder the ESI spraying process or contaminate the MS analyzer. Another important benefit is the possibility to connect capillaries of different diameters for CEC and MS spraying. However, a disadvantage of the liquid junction interface is the difficulty in precisely aligning the CEC capillary with the spraying capillary, leading to analyte band broadening. In the sheathless interface, the liquid entering the MS is assisted by a sheath gas [29]. A very sharp capillary tip with an internal diameter of a few micrometer and with an electrically conducting surface coating is often employed. The conducting surface is either fixed at the end of the capillary column or the CEC column is

24.3 Instrumentation

integrated with a conductive nanospray tip. The major disadvantage of the conductive capillary coating approach is the limited lifetime of the conductive coating, which results in the need for replacing the whole CEC capillary column. Instead of the conducting coating or the use of a nanospray tip, the electrical contact can also be achieved by utilizing a conductive electrode (a wire), which is typically made of platinum, inside the capillary column. The nanoelectrospray interface is the interface of choice when nano- and capillary columns are to be conjoined with MS [30]. The popularity of nanoelectrospray, since its invention in the mid-1990s, has gradually increased due to its benefits such as higher sensitivity, lower background noise in the mass spectrum, and compatibility with nL/min flow rates. Atmospheric pressure chemical ionization (APCI), another atmospheric pressure ionization technique, has not been much employed for CEC-MS hyphenation [31]. With APCI liquids of rather low flow rates can be nebulized, and therefore, direct combination with CEC is possible. In CEC-APCI/MS compounds from the CEC, capillary is directly introduced into a pneumatic nebulizer. Once in the exit of nebulizer, the compounds are converted into a mist with the aid of a high-speed nitrogen nebulizing gas, and desolvation is further assisted by heating. The CEC mobile phase and the sample in the gas flow are vaporized by the heat transferred to the spray droplets in the desolvation chamber and leave the chamber in the form of a mixture of the compounds of interest and hot gas. Subsequently, both positive and negative ions can be obtained with the help of a corona discharge needle. With APCI, polar to semipolar compounds with molar masses typically less than 1500 Da can be ionized. Therefore, APCI can be seen as a complementary ionization method to ESI regarding the type of analytes that can be ionized. The main advantages of APCI over ESI are good ion beam stability and higher tolerance against matrix effects. Only a few works employing matrix-assisted laser desorption ionization (MALDI) as an ionization technique in conjunction with CEC has been published, see, for example [32]. MALDI is a soft ionization method especially suitable for proteins and biomolecules of a molar mass higher than ca 500 kDa. In MALDI, the sample solution is mixed with a matrix, deposited on a MALDI plate, and the solvent is evaporated [33]. The dry sample is irradiated in vacuum with a high-energy laser such as a nitrogen laser (337 nm) or an Nd:YAG laser (355 nm and 266 nm). The MALDI plume generated by laser ablation contains mainly the primary ions of the MALDI matrix, and through charge transfer, the sample analytes will be ionized. The reason for the low popularity of on-line connecting CEC with MALDI/MS is most probably due to the laborious process connected with the necessity of decoupling the high-voltage field applied in CEC. The connected mass analyzers vary quite much, and almost all types of MS instruments have been on-line connected to CEC. The most relevant criteria of mass analyzers when connected on-line to CEC are selectivity, sensitivity, and acquisition speed. The very low amounts of sample ions entering the MS from the CEC capillary put great demands on the sensitivity of the mass analyzer. In addition, the need for a high acquisition speed is due to the narrow analyte zones formed during the

639

640

CHAPTER 24 Capillary electrochromatography

electrophoretic separation. Quadrupole-based MS analyzers have been frequently utilized, as well as ion trap and time-of-flight mass analyzers. For even increased resolution and determination of highly accurate molar masses, sophisticated and precise MS analyzers based on Fourier transform ion cyclotron resonance (FTICR) and Orbitrap analyzers are especially useful, but not yet so much employed with CEC. The high resolving power and accuracy of Orbitrap analyzers have made them particularly useful, especially in the field of proteomics.

24.4 Miniaturized systems Miniaturization of analytical instrumentation is, in many fields, considered to be one of the most promising approaches to meet the challenges related to high-throughput screening. In general, miniaturization of analytical instruments via microfabrication dates back to the advanced development of silicon microfabrication processes in the early 1980s. This resulted in an explosive growth of semiconductor technology, which, in practice, meant that all main electrical components such as diodes, integrated circuits, solar cells, and transistors could be fabricated and produced at such low cost that computers, mobile phones, and other electrical supplies became available to average users. This technological process also led to the development of miniaturized chip-based separation techniques, which has increased much during the last decade. The first microchip-based CEC system was published in 1994, just 20 years after the first capillary electrochromatographic system, was demonstrated. In this work by J.M. Ramsey et al., an open-channel electrochromatographic system was fabricated on a glass chip with a cross section of 66 μm in width and 5.6 μm in height [34]. The channels on the glass microchip were prepared by standard photolithographic techniques and chemical wet etching. The chromatographic surface contained a reversedphase octadecylsilane material, and detection was performed by fluorescence. A crucial challenge with microchip channels is related to uniform packing of the channels with stationary-phase particles. Various approaches for packing microchannels with stationary-phase particles have been developed. For example, an external magnetic field for keeping noncovalently attached magnetic nanoparticles fixed to the wall in an open-tubular CEC approach has been demonstrated, as shown in Figure 24.6 [35]. The separation channel was 50 μm wide and 18 μm deep, and the system contained molecularly imprinted polymers made from Fe3O4 nanoparticles as the supporting substrate and dopamine as the functional monomer. Due to challenges encountered with uniformly packed stationary-phase microchannels, porous polymer monoliths have been much used in chip-based capillary electrochromatography. The porosity of polymer monoliths can easily be modified by changing the porogen content to permit lower back-pressures than those in packed microchannels. Thermal polymerization or UV-initiated polymerization is the most widely employed methodology for preparing monolithic beds in microchannels. One specific advantage of UV-initiated polymerization is the possibility to position the

24.4 Miniaturized systems

FIGURE 24.6 Illustration of the stabilization of imprinted Fe3O4–polydopamine (PDA) nanoparticles (NPs) by the aid of a magnetic field in a polydimethylsiloxane (PDMS)-based microchip. Reprinted with permission from Wang et al. [35]. Copyright Elsevier Science Publishers.

monolith in a specified part of the microdevice. Figure 24.7 shows a schematic of an integrated microchip CEC system with on-line chemiluminescence (CL) detection [36]. Microchannels with different channel lengths (in cm scale), diameters (ranging from μm to nm scale), and microstructures (like pillars within microchannels) can be created using microfabrication technology. For example, microfabricated pillar arrays in microchip electrochromatography allow low back-pressures and highly efficient separation of compounds. However, their important limitation is limited specific surface. The pillar structures can be fabricated, for example, by the procedure described in Figure 24.8 [37].

FIGURE 24.7 Schematic presentation of an integrated microchip CEC system with on-line chemiluminescence (CL) detection. (A) Device; (B) channel design (dimensions are given in mm) and SEM micrographs of the photopolymerized monolith; (C) configuration of the spiral detection channel. V, vacuum gauge; T, timer; B, buffer reservoir; C, copper solution; S, sample reservoir; SW, sample waste reservoir; H, hydrogen peroxide solution reservoir; L, luminol solution reservoir; W, waste reservoir; PMT, photomultiplier tube. Reprinted with permission from Wang et al. [36]. Copyright Elsevier Science Publishers.

641

642

CHAPTER 24 Capillary electrochromatography

(B)

(C)

(E) (A)

(D)

FIGURE 24.8 Fabrication starts with a silicon wafer substrate (A) on which photolithographic patterning is performed (B), followed by deep reactive ion etching (C) to create the high-aspect ratio pillars. A thin layer of silicon oxide (100 nm) is deposited on the wafer surface using plasmaenhanced chemical vapor deposition (D). Figure E shows a scanning electron microscopy picture of a typical array. Reprinted with permission from Kirchner et al. [37]. Copyright (2013) American Chemical Society.

The microchip technology is undergoing continuous development, and much research has been put on the integration of sample preparation and sample enrichment with efficient separation and on-line detection. Such miniaturized or micrototal chemical analysis systems (μTAS) will most presumably replace in the future time-consuming multistep analysis approaches, but new studies are still needed before these devices meet the demands of totally replacing conventional methodologies. The μTAS applications include protein analysis, and various cell analysis systems including cell sorting and capturing of circulating tumor cells, organisms on a chip, environmental analysis, and drug screening [38]. There are still major challenges with the automated operation of μTAS systems, and much is to be done especially in the fields of detection development and sample preparation to achieve the sensitivity needed. Future trends also include the development of handheld and bedside biomarker detection instrumentations for biomedical applications, with specific focus on sensitive immunoaffinity methodologies.

24.5 Applications The CEC applications have mainly dealt with the separation of low-molar mass organic compounds and small biomolecules. In many cases, the compounds have

24.5 Applications

been used as model compounds in the evaluation of the performance of the CEC columns, without focusing on novel applications. However, also larger biomolecules such as proteins have been separated [39]. The applications can be divided into the separation of neutral compounds, organic acids and bases, inorganic anions, and inorganic heavy metal cations. The type of compounds that have been separated can be roughly classified into biomolecules like amino acids and peptides [40], drugs like nonsteroidal anti-inflammatory drugs, and other types of compounds such as enzyme inhibitors, neurotransmitters, and herbicides [41]. Specific applications include impurity detection and the separation of enantiomers by chiral CEC. Detection of low concentrations of possible impurities is an important topic especially in the pharmaceutical industry, and pharmaceuticals are typical analytes in enantioseparations [15,42]. In the beginning of year 2000, enantioseparations were mostly carried out using particle-based stationary phases, which originally were fabricated for liquid chromatography applications. Packed capillary columns with silica-based particles including core–shell and nonporous silica have been the most commonly employed in enantioseparations, and polysaccharide and cyclodextrin-based chiral selectors have been the most frequently used particle-based chiral stationary phases. Lately, however, monoliths and open-tubular CEC have attracted more and more interest due to the better performance of the monoliths than columns packed with 3to 5-μm particles. Monolithic chiral stationary phases based on polysaccharides, cyclodextrins, and macrocyclic antibiotics have been successfully employed for the separation of a large number of various pharmaceuticals and other chiral compounds. The monoliths are usually based on silica, zirconia, silica/zirconia hybrids, although polymeric monoliths have been used as well. Monoliths with immobilized protein-based selectors such as bovine serum albumin have also been utilized to some extent. Also, molecularly imprinted polymers have been used for chiral separation of enantiomers (Figure 24.9). Chiral molecular imprinted particles have been used with monoliths, both in open-tubular CEC and in particle-packed capillaries [43,44]. In open-tubular CEC, polysaccharide or cyclodextrin-based selectors have been most frequently used. In addition to the molecular imprinted polymers mentioned, open-tubular affinity capillary electrochromatography systems with an immobilized protein have been utilized [45]. The combination of chiral and nonchiral separation of low-molar mass drugs using a molecular imprinted template has offered successful separation of S-enantiomer of a mixture of 11 drugs [46]. The open-tubular molecular imprinted particle capillary column has also enabled the separation of many enanantiomers, including nonchiral separation of some alkylbenzenes and other low-molar mass polar analytes.

643

644

CHAPTER 24 Capillary electrochromatography

FIGURE 24.9 Schematic illustration of molecular imprinting in chiral analysis using a noncovalent approach. Reprinted with permission from Iacob et al. [43]. Copyright John Wiley and Sons.

References [1] Tamizi E, Jouyban A. The potential of the capillary electrophoresis techniques for quality control of biopharmaceuticals—a review. Electrophoresis 2015;36(6):831–58. [2] Mao Z, Chen Z. Advances in capillary electro-chromatography. J Pharm Anal 2019;9 (4):227–37. [3] Kartsova LA, Makeeva DV, Bessonova EA. Current status of capillary electrophoresis. J Anal Chem 2020;75(12):1497–513. [4] Strain HH. On the combination of electrophoretic and chromatographic adsorption methods. J Am Chem Soc 1939;61(5):1292–3. [5] Pretorius V, Hopkins BJ, Schieke JD. Electro-osmosis: a new concept for high-speed liquid chromatography. J Chromatogr A 1974;99:23–30. [6] Jorgenson JW, Lukacs KD. High-resolution separations based on electrophoresis and electroosmosis. J Chromatogr A 1981;218:209–16. [7] Knox JH, Grant IH. Miniaturisation in pressure and electroendosmotically driven liquid chromatography: some theoretical considerations. Chromatographia 1987;24(1):135–43.

References

[8] Knox JH, Grant IH. Electrochromatography in packed tubes using 1.5 to 50 μm silica gels and ODS bonded silica gels. Chromatographia 1991;32(7):317–28. [9] Henley WH, Jorgenson JW. Ultra-high voltage capillary electrophoresis >300 kV: recent advances in instrumentation and analyte detection. J Chromatogr A 2012;1261:171–8. [10] Palonen S, Jussila M, Riekkola M-L. Effect of initial voltage ramp on separation efficiency in non-aqueous capillary electrophoresis with ethanol as background electrolyte solvent. J Chromatogr A 2005;1068(1):107–14. [11] Hutterer KM, Jorgenson JW. Ultrahigh-voltage capillary zone electrophoresis. Anal Chem 1999;71(7):1293–7. [12] Palonen S, Jussila M, Porras SP, Hy€otyl€ainen T, Riekkola M-L. Extremely high electric field strengths in non-aqueous capillary electrophoresis. J Chromatogr A 2001;916(1–2): 89–99. [13] Xue Y, Gu X, Wang Y, Yan C. Recent advances on capillary columns, detectors, and two-dimensional separations in capillary electrochromatography. Electrophoresis 2015;36(1):124–34. [14] Colo´n LA, Maloney TD, Fermier AM. Packing columns for capillary electrochromatography. J Chromatogr A 2000;887(1–2):43–53. [15] Fanali S, Chankvetadze B. History, advancement, bottlenecks, and future of chiral capillary electrochromatography. J Chromatogr A 2021;1637, 461832. [16] Tsuda T, Nomura K, Nakagawa G. Open-tubular microcapillary liquid chromatography with electro-osmosis flow using a UV detector. J Chromatogr A 1982;248(2):241–7. [17] Tarongoy FM, Haddad PR, Boysen RI, Hearn MTW, Quirino JP. Open tubular-capillary electrochromatography: developments and applications from 2013 to 2015. Electrophoresis 2016;37(1):66–85. [18] Singh V, Joung D, Zhai L, Das S, Khondaker SI, Seal S. Graphene based materials: past, present and future. Prog Mater Sci 2011;56(8):1178–271. [19] Li X-J, Fu Q-F, Zhang Q-H, Jiang X-M, Yang F-Q, Wei W-L, et al. Layer-by-layer selfassembly of polydopamine/gold nanoparticle/thiol coating as the stationary phase for open tubular capillary electrochromatography. Anal Methods 2015;7(19):8227–34. [20] Barcelo´-Barrachina E, Moyano E, Galceran MT. State-of-the-art of the hyphenation of capillary electrochromatography with mass spectrometry. Electrophoresis 2004;25(13): 1927–48. [21] Verheij ER, Tjaden UR, Niessen WMA, Van der Greef J. Pseudo-electrochromatography-mass spectrometry: a new alternative. J Chromatogr A 1991;554(1):339–49. [22] Fenn J, Mann M, Meng C, Wong S, Whitehouse C. Electrospray ionization for mass spectrometry of large biomolecules. Science 1989;246(4926):64–71. [23] Cole RB. In: Cole RB, editor. Electrospray and MALDI mass spectrometry: Fundamentals, instrumentation, practicalities, and biological applications. New York: Wiley; 1997. [24] Chawdhury ASMMR, Shamsi SA, Miller A, Liu A. Capillary electrochromatographymass spectrometry of kynurenine pathway metabolites. J Chromatogr A 2021;1651, 462294. [25] Klampfl CW. Review coupling of capillary electrochromatography to mass spectrometry. J Chromatogr A 2004;1044(1–2):131–44. [26] Gelpı´ E. Interfaces for coupled liquid-phase separation/mass spectrometry techniques. An update on recent developments. J Mass Spectrom 2002;37(3):241–53. [27] Smith RD, Olivares JA, Nguyen NT, Udseth HR. Capillary zone electrophoresis-mass spectrometry using an electrospray ionization interface. Anal Chem 1988;60(5):436–41.

645

646

CHAPTER 24 Capillary electrochromatography

[28] Lee ED, M€uck W, Henion JD, Covey TR. Liquid junction coupling for capillary zone electrophoresis/ion spray mass spectrometry. Biol Mass Spectrom 1989;18(9):844–50. [29] Olivares JA, Nguyen NT, Yonker CR, Smith RD. On-line mass spectrometric detection for capillary zone electrophoresis. Anal Chem 1987;59(8):1230–2. [30] Warriner RN, Craze AS, Games DE, Lane SJ. Capillary electrochromatography/mass spectrometry—a comparison of the sensitivity of nanospray and microspray ionization techniques. Rapid Commun Mass Spectrom 1998;12(17):1143–9. [31] Tanaka Y, Otsuka K, Terabe S. Evaluation of an atmospheric pressure chemical ionization interface for capillary electrophoresis-mass spectrometry. J Pharm Biomed Anal 2003;30(6):1889–95. [32] Tegeler TJ, Mechref Y, Boraas K, Reilly JP, Novotny MV. Microdeposition device interfacing capillary electrochromatography and microcolumn liquid chromatography with matrix-assisted laser desorption/ionization mass spectrometry. Anal Chem 2004;76 (22):6698–706. [33] Karas M, Kr€uger R. Ion formation in MALDI: the cluster ionization mechanism. Chem Rev 2003;103(2):427–40. [34] Jacobson SC, Hergenroeder R, Koutny LB, Ramsey JM. Open channel electrochromatography on a microchip. Anal Chem 1994;66(14):2369–73. [35] Wang X-N, Liang R-P, Meng X-Y, Qiu J-D. One-step synthesis of mussel-inspired molecularly imprinted magnetic polymer as stationary phase for chip-based open tubular capillary electrochromatography enantioseparation. J Chromatogr A 2014;1362:301–8. [36] Wang X, Dong S, Li F. A miniaturized analysis system composed of monolith-based microchip electrochromatography coupled with on-line chemiluminescence detection for amino acids. Sens Actuators B 2016;225:529–34. [37] Kirchner TB, Hatab NA, Lavrik NV, Sepaniak MJ. Highly ordered silicon pillar arrays as platforms for planar chromatography. Anal Chem 2013;85(24):11802–8. [38] Culbertson CT, Mickleburgh TG, Stewart-James SA, Sellens KA, Pressnall M. Micro total analysis systems: fundamental advances and biological applications. Anal Chem 2014;86(1):95–118. [39] Hajba L, Guttman A. Recent advances in capillary electrochromatography of proteins and carbohydrates in the biopharmaceutical and biomedical field. Crit Rev Anal Chem 2021;51(3):289–98. [40] Kasˇicka V. Recent developments in capillary and microchip electroseparations of peptides (2019–mid 2021). Electrophoresis 2022;43:82–108. [41] Przybylska A, Gackowski M, Koba M. Application of capillary electrophoresis to the analysis of bioactive compounds in herbal raw materials. Molecules 2021;26(8):2135. [42] Declerck S, Vander Heyden Y, Mangelings D. Enantioseparations of pharmaceuticals with capillary electrochromatography: a review. J Pharm Biomed Anal 2016;130:81–99. [43] Iacob B-C, Bodoki E, Oprean R. Recent advances in capillary electrochromatography using molecularly imprinted polymers. Electrophoresis 2014;35(19):2722–32. [44] Lobato A, Pereira EA, Gonc¸alves LM. Combining capillary electromigration with molecular imprinting techniques towards an optimal separation and determination. Talanta 2021;221, 121546. [45] Wang M, Liu Y, Liu Y, Xia Z. MOFs and PDA-supported immobilization of BSA in open tubular affinity capillary electrochromatography: prediction and study on drugprotein interactions. Talanta 2022;237, 122959. [46] Zaidi SA, Han KM, Hwang DG, Cheong WJ. Preparation of open tubular molecule imprinted polymer capillary columns with various templates by a generalized procedure and their chiral and non-chiral separation performance in CEC. Electrophoresis 2010; 31(6):1019–28.

CHAPTER

Miniaturization and microchips

25

Jozef Sˇesta´k, Filip Dusˇa, Anna Ty´cˇova´, Jan Prikryl, and Frantisˇek Foret Institute of Analytical Chemistry of the Czech Academy of Sciences, Brno, Czechia

25.1 Introduction The past twenty years have witnessed a new wave in the development of analytical instrumentation. Fuelled mainly by the needs of rapidly developing life sciences, new generations of DNA sequencers, mass spectrometers, and liquid chromatographs are being developed in the waves just a few years apart. Fortunately, the unique challenges presented by biologists, the pharma industry, or medical research can be quickly addressed by the rapid developments in the material sciences, IT technology, and new theoretical approaches available for the improvements of chromatographic separations. The following text focuses on reviewing the benefits of miniaturization and microfluidics for the development of the new liquid chromatography instrumentation.

25.2 Compact solvent delivery systems The solvent delivery system is an essential part of every HPLC instrument. It must blend two or more solvents with accurate compositional control and deliver them at a stable flow rate even when the modern chromatography column generates back pressure up to 1000–1500 bar. Flow rate stability and composition accuracy is crucial prerequisite for reproducible results of HPLC analysis. In piston pumps, sapphire or ceramic piston [1] and polymer-filled PTFE seal [2] have become a gold standard of high-pressure pumping liquids. The piston is typically motor-driven (constant flow mode) or powered by compressed gas (constant pressure mode, e.g., Smartline Pneumatic Series from Knauer). The fluid is sucked into the pump head on piston stroke while reverse piston motion compresses the liquid and generates flow toward and through the injector, column, and detector. Continuous pumping requires a reciprocating pump with a dual-piston design (one piston is displacing, the second one is acquiring liquid) and active or passive check valves, Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00020-5 Copyright # 2023 Elsevier Inc. All rights reserved.

647

648

CHAPTER 25 Miniaturization and microchips

FIGURE 25.1 Principle of HPLC reciprocating pumping. (A) A single-piston pump, (B) a quaternary gradient HPLC pump with proportional valve and single dual-piston pump, (C) a binary gradient HPLC pump with two dual-piston pumps.

which prevent the backflow and pressure drops during the pumping cycle. Diagrams of HPLC reciprocating pumping are shown in Figure 25.1. The analytical HPLC pumps are further equipped with degasser, pressure sensors, flow meters, pulse damper, mixer, seal wash, and leak management system [3]. Hence, the analytical HPLC pumps are complex and bulky devices. Thus, commercially available compact reciprocating pumps consist of a single or two pistons (discontinuous or continuous flow, respectively), sealing, check valves, and only a few other essential parts. Teledyne ISCO M1 Class single piston (14  7.6  26.7 cm, 1.6 kg, up to 140 bar and 10 mL/min), Eldex 1LM single piston (10  23  24 cm, 5.3 kg, up to 413 bar, 0.002–2.5 mL/min, Figure 25.2A), or Knauer AZURA P 4.1S dual piston (12  13  22 cm, 2.3 kg, up to 400 bar, 0.1–8 mL/min) belong to this category of products, all operating in isocratic mode. Syringe pumps represent an alternative category of simple (to use and clean) and compact pumps. They are equipped with a syringe made of glass (up to 50 bar with 100 μL syringe) or stainless steel (up to 517 bar with 3 mL CETONI stainless steel syringe and modular CETONI Nemesys high-pressure pump, Figure 25.2B). Some syringe pumps rely on sapphire piston and sealing used for reciprocating pumps as discussed above. While piston stroke in reciprocating pumps is about a few millimeters, the syringe pumps include a stepper motor or a high resolution servo motor (e. g. CETONI syringe pumps) screw-driven piston whose stroke is significantly longer (e.g., 60 mm with 100 μL glass syringe, or 59.64 mm with 3 mL CETONI stainless steel syringe). The operation of the syringe pump is discontinuous, and analysis time is scaled to the single-piston displacement (e.g., 140 μL in ThermoScientific EasynLC pump, 35 μL in VICI TrueNano pump). While single syringe pump can be used for isocratic mode of chromatography only, gradient runs require two pumps. A compact dual syringe pump was developed by the team of Milton L. Lee [4] and is now sold by VICI [5].

25.2 Compact solvent delivery systems

FIGURE 25.2 Examples of compact high-pressure pumps. (A) A single-piston reciprocating pump Eldex 1LM, (B) A single module of CETONI Nemesys high-pressure syringe pump with stainless steel syringe. Pictures reprinted with permission.

FIGURE 25.3 (A) The principle of electroosmotic pumping; the electroosmotic flow (EOF) of aqueous solution generates a pressure gradient between the inlet and the outlet of the channel with negatively charged inner surface. (B) Example of the series connection of two electroosmotic pumping subunits.

Electroosmotic pumps are experimental devices that utilize electroosmosis and electroosmotic flow (Figure 25.3A) for pressure and flow generation [6]. No moving parts and generation of constant and pulse-free flow are considered as the major benefits. Hence, they are ideal for incorporating in microchips [7]. Pump characteristics are influenced by the pumping element, which can be an open channel [8], packed bed [6], or monolithic column [9,10]. One end of the pumping element is connected with a reservoir with liquid to be pumped while the other end is decoupled and serves as the flow outlet. The generated flow is influenced by the applied voltage, liquid composition, and column back pressure. Series connection of several pumping units

649

650

CHAPTER 25 Miniaturization and microchips

enhances pumping power (Figure 25.3B), and the latter design is even capable of gradient generation and delivering flow at a rate up to 0.5 μL/min and pressure up to 17 MPa [8] or 39 MPa [10].

25.3 Aspects of sample injection in miniaturized HPLC In an ideal case, the separation column or channel is the main contributor to the observed efficiency and resolution. In isocratic separation, the column efficiency can be expressed as theoretical plate count N (Eq. 25.1).  t 2 r N ¼ 16 w

(25.1)

If N is known, then Eq. (25.1) allows for the calculation of peak width w [min] at the baseline at any value of retention time tr [min]. Retention time can be further expressed as tr ¼ k t0 + 1 where k is the retention factor and t0 is the retention time of an unretained compound. Finally, the calculated peak width w ¼ 4σ [min] can be multiplied by flow rate [μL/min] to obtain peak volume [μL] and peak variance σ 2col [μL2]. Values calculated for typical conditions operated in miniaturized HPLC are listed in Table 25.1. σ 2 ¼ σ 2inj + σ 2col

(25.2)

For the purpose of this section, we assume that the observed peak variance is the sum of injection variance σ 2inj and column variance σ 2col. Eq. (25.2) together with the considerations mentioned above allow for the calculation of efficiency loss due to injection volume (Figure 25.4). Table 25.1 Calculated peak volume and variance. tr [min]

1

2

3

4

5

6

k

0

1

2

3

4

5

Column

Peak volume [nL]

0.15  100 mm @ 1 μL/min

37

73

110

146

183

219

0.3  100 mm @ 5 μL/min

182

365

548

730

912

1095

Peak variance [nL2] 0.15  100 mm @ 1 μL/min

83

333

750

1333

2083

3000

0.3  100 mm @ 5 μL/min

2083

8333

18,750

33,333

52,083

75,000

Column efficiency N ¼ 12,000 theoretical plates. Isocratic conditions, t0 ¼ 1 min.

25.4 Microchips

FIGURE 25.4 Calculated efficiency as a function of retention factor and injection volume. Column: 0.15  100 mm, 1 μL/min or 0.3  100 mm, 5 μL/min. Theoretical plate count N ¼ 12,000, t0 ¼ 1 min.

Figure 25.4 illustrates that the efficiency observed for early eluting compounds (k < 2) decreases with increasing injection volume. When high efficiency is required for early eluting compounds (e.g., because of the resolution of the critical pair), the sample volume injected on 0.15  100 and 0.3  100 mm column should not exceed 10 and 50 nL, respectively. Such a volume is typically injected by the injector valve with an internal sample loop (e.g., Cheminert Injectors produced by VICI, internal loops 4–50 nL) or by partial displacement of the external loop (loop is switched into the main flow path for a very short time). Figure 25.4 further shows that strongly retained compounds sustain the separation efficiency even at larger injection volumes. If the elution strength of the sample solvent is lower than that of the mobile phase, the retention factor of the analyte is temporarily increased, and a larger volume can be injected without a significant decrease of observed efficiency. This so-called on-column focusing improves the limits of detection and can be performed directly at the head of the separation column or channel or at the trap column installed instead of the external sample loop.

25.4 Microchips 25.4.1 Microchip materials and microfabrication technologies For the fabrication of microfluidic devices, the researchers currently have an enormous selection of materials. The nature and the final function determine the way of the device fabrication, i.e., subtractive (removal of material) or additive (layer-bylayer building) families. This section will pay attention to the machining of hard and soft polymers and silicon-based materials, all being traditional building blocks of microfluidic LC (mLC) devices.

651

652

CHAPTER 25 Miniaturization and microchips

25.4.1.1 Hard polymers Hard polymers gained popularity due to their simplicity of processing. Cyclic olefin copolymer [11], polystyrene [12,13], polycarbonate [14], and polyimide [15] belong to the most frequently used materials due to their optical transparency and chemical resistance. Their processing is commonly based on subtractive methods. Additionally, thermoplastic materials can be machined via hot-embossing and 3D printing. Micromachining (e.g., micro-cutting, micro-milling, or micro-drilling) is the most traditionally used technique for processing firm polymers, mechanically removing unwanted parts of the material. While the manual control of micromachines produces typically auxiliary components (e.g., chassis or holders), the coupling to computer numerically controlled (CNC) units significantly improves the performance and turns them into a precise tool, e.g., for the fabrication of open channels [13]. Although such a fabrication procedure is very rapid, the typical width of the channels is in hundreds of μm, often carrying microscratches from the tools. Thus, the microfluidic platforms from firm polymers serve mostly for prototyping mLC devices. An excellent example is the work of Wouters et al., which prototyped a device for spatial microfluidic LCxLC (Figure 25.5A) [11]. Laser machining represents another popular tool for polymer manufacturing, removing the material via a process of ablation. As a non-contact technology, the final product is scratch-free; however, the residuals of decomposed polymers can create roughness on the exposed surface [18]. The fine and precise cut of the laser beam is considered the most significant advantage for the structures down to 5 μm [18,19]. The technology of sequential laser ablation is used, for example, by Agilent for the mass production of polyimide HPLC-chip/MS [15]. Therein, the ablation

FIGURE 25.5 (A) Photograph of a microfluidic device for spatial LC  LC analysis fabricated by CNC microdrilling. (B) SEM images of embossed micropillars. (C) The photocurable mixture used for the direct 3D printing of monolithic ion-exchange sorbent. (A) Reprinted with permission from Wouters et al. [11]. (B) Reprinted with permission from Kourmpetis et al. [16]. (C) Reprinted with permission from Simon and Dimartino [17].

25.4 Microchips

process forms essential features of the device (the microfluidic channels, ports, chambers, and columns) and cuts the chip’s overall shape, including the ES tip. Hot embossing has a capacity for rapid manufacturing of large quantities of identical devices from thermoplastics. The polymer is heated above its glass transition temperature (Tg) making it pliable. The desired structure is pressed into the polymer via a stamp (typically made of metal or glass). Applied pressure, heating temperature, embossing time, and demolding temperature have to be carefully optimized to avoid deformation of the embossed structures [20]. This is particularly important for the fabrication of high/dense structures. Kourmpetis et al. precisely controlled all the crucial parameters to fabricate an on-chip chromatographic column with a highdensity array of cylindrical micropillars (Figure 25.5B) [16]. The group of Oleschuk showed another upside of this manufacturing, using the embossing for on-chip integration of external components [21]. Their simple protocol involved the embossing of a fused silica capillary into a polymeric plate. The emitter and inlet for the mobile phase were embedded in the channel, having the same outer diameter as the embossed capillary, without any fitting. While the previous techniques remove parts of the bulk material, 3D printing belongs to additive manufacturing, building solid objects layer-by-layer. In the last decade, 3D printing evolved into a mature technology capable of processing various materials; however, polymeric substances still prevail. The different materials and applications require different operation principles, and several successful technologies have been developed. The fused deposition modeling (FDM), stereolithography (SLA), selective laser sintering (SLS), and multijet modeling (MJM) belong to the most common [22]. Although the physicochemical nature of their processes differs, the general principle is similar—the incoming material is precisely situated on the substrate and solidifies. In the case of polymers, thermoplastics (bulk or powder) or photosensitive resins can act as the incoming material. Thus, the technology of 3D printing represents a universal tool for processing a wide range of polymeric substances. In the miniaturized LC technology, several promising achievements were obtained in the direct printing of monolithic stationary phase (Figure 25.5C) [17], valves [23], and pumping systems [24]. However, their implementation into a functional LC microfluidic device has not been reported yet. The new developments in 3D printing signalize that this technology is one of the promising tools, which will constantly push achievements in microdevice constructions beyond current limits [22].

25.4.1.2 Soft polymers Soft polymers are typically treated via a collection of techniques called soft lithography, providing self-standing structures or structures helpful in transferring a pattern (e.g., stamps, molds, and photomasks). In the family of soft polymers, polydimethylsiloxane (PDMS) is the backbone of many microfluidic systems. Its solidification is based on cross-linking of monomeric units via a curing agent. The structure and mechanical properties of PDMS

653

654

CHAPTER 25 Miniaturization and microchips

elastomers are highly influenced by their reaction ratio [25]. A higher percentage of the curing agent increases the crosslinking and thus affects the resulting mechanical properties of PDMS elastomer. Temperature is another important factor to control (commonly 60°C) since it affects the rate of the polymerization reaction [26,27]. For the patterning of PDMS, the liquid mixture of monomer and curing agent is poured into a mold having the desired design. After the solidification process, the PDMS block is carefully peeled off, being an elastomeric negative replica of the mold [13,28]. While being a dominant material in many microfluidic systems, its broader use in the field of mLC is hindered by three main reasons: (1) many molecules tend to absorb onto PDMS surface, (2) swelling of PDMS occurs by numerous organic compounds, (3) the elasticity of the material is not compatible with high pressures used in HPLC systems. As the deformation at relatively low hydrodynamic pressure is generally taken as a disadvantage, Kabiri et al. used elasticity of PDMS-based device for frit-free packing of stationary phase [28]. The PDMS channel was packed with silica beads consisting of magnetic nanoparticles, being trapped in the separation channel via magnetic forces. The loading pressure expanded the walls of the channel, so once the flow of the liquid was stopped, the PDMS walls pushed the stationary phase inward, making a very tight packing of beads. The work of Itsiha et al. demonstrates a more typical example of PDMS’s role in chromatographic microdevices. While the chromatographic column was located in a firm polystyrene substrate, the attached PDMS component served only as an electrochemical flow cell [13]. Photoresists are a fundamental group of photoactive resins formed into patterns via photolithography. The exposure of a resin to UV light is the key element of photolithography. While the negative photoresists became after the exposure to UV light in the washing solution (developer) insoluble, the behavior of positive photoresists is the opposite—they turn into a soluble form and leave the substrate. The UV-initiated solidification of negative photoresist is widely employed in the technology of 3D printing (stereolithography), being shortly discussed previously [22]. The on-substrate structures of photoresists are especially useful in wet etching processes, acting as a protective layer of the substrate against etchant (see the next section) or as a mold for PDMS patterning [29]. Additionally, they can be in some specific cases used as an auxiliary material for mLC chip constructions. For instance, the negative photoresist SU-8 has been used for the fabrication of nES emitters [30], micropillar array [31], or a microchannel anchoring the monolithic stationary phase [32].

25.4.1.3 Silicon-based materials Silicon is the most common material used in microtechnology due to its wellcharacterized properties and sophisticated fabrication techniques developed for microelectromechanical systems (MEMSs). Although these techniques were developed for fabrication in silicon, many of the methods were adopted for the fabrication of microfluidic devices from the silicon dioxides (i.e., fused silica, glass, quartz)

25.4 Microchips

[33,34], which are by far more popular due to superior optical transparency as well as due to surface chemistry with good EOF characteristics (for EOF pumping, see Section 25.2) [13]. The silicon-based materials are typically exposed to chemical etching performed in solution (wet etching) or plasma (dry etching). The wet etching has a dominant position in the majority of microfabrication processes. While many etching protocols were published, the etching of silicon-based materials by buffered hydrofluoric acid belongs to the most common procedure [35,36]. Prior to the etching step, the substrate is typically coated with a thin photoresist layer, patterned via photolithography. Only the photoresist-free surface is etched away in the etching bath. Dimensions of the photoresist mask, etching time, and temperature are crucial parameters controlling the width and depth of the obtained channel [35]. However, the isotropic character of etching removes the material equally in all the directions resulting in round channels (Figure 25.6C). Additionally, the edges of the channels are often irregular especially if long etching times or increased etching temperature is used (Figure 25.6A) [37]. While strong bases etch silicon monocrystals anisotropically, the anisotropic etching via wet chemistry is challenging to achieve in silicon dioxide. Reactive ion etching (RIE), so-called dry etching, can address this difficulty [39]. A fluorine-containing gas, e.g., CHF3 [38], XeF2 [40], SF6 [41], and C4F8 [42], is lead into a reactive chamber and converted into chemically reactive plasma. The plasma bombards the surface of the substrate and removes its bare parts, producing volatile

FIGURE 25.6 (A) A typical profile (left) and roughness (right) of the channel prepared via isotropic wet etching. (B) A micropillar frit fabricated via deep reactive ion etching. (C) Profiles of channels after isotropic (up) and anisotropic (down) etching. (A) Reprinted from Kara´sek et al. [37]. (B) Reprinted with permission from Haapala et al. [38].

655

656

CHAPTER 25 Miniaturization and microchips

SiFn products. The process of ion etching suffers from the low speed of the etching process and a lack of selectivity—removing both the substrate and the mask [43]. Despite these facts, the RIE is considered a highly controlled process. Alternating the etching and thermal-passivation cycles (deep reactive ion etching, DRIE) is used for the fabrication of very fine nano-/micropillars suitable (after the chemical modification of their surface) as a stationary phase or a frit (Figure 25.6B) [38,40,41,44]. Sainiemi et al. combined wet and dry etching for the fabrication of a microfluidic LC with MS detection. While the dry etching was used for the fabrication of a micropillar array on a silicon wafer (stationary phase), the edge of bonded glass cover plate was shaped into a sharp tip via isotropic wet etching (electrospray emitter) [45].

25.4.2 Separation channels and beds Generally said, the μLC chip features similar separation beds as the larger-scale LC columns. Particles are the most widespread stationary phase, and it was the first solution for commercially developed LC chips [46]. The common stationary phases developed during the past several decades for regular HPLC and UPLC offer high separation mechanisms and particle sizes variability. However, particle-based microchips have to include one very important feature a frit-like structure to contain the particles inside the separation channel. While making a frit in situ inside the microchannel can be difficult, there are several ways to tackle this issue [46]. Typical solutions include a single particle to block the column output, a microstructure (e.g., set of micropillars, narrow channels) effectively blocking the further movement of the particles (Figure 25.6B), or a short porous monolith structure closing the microchannel [47]. In a fritless column, the keystone effect causes particle congestion during the slurry packing at the end of a tapered channel [47]. Tapering the separation channel to three times the size of the used particles is usually sufficient. Elevated structures built inside the separation channel (i.e., weirs) can also be used to contain particles [48]. For high efficiency of particle columns, it is required to have high-pressure resistant frits or frit-like structures to endure the packing process and future separation. This mainly applies to commercial LC chips made of hard materials to withstand high pressures [49]. The commercial system Ekspert nanoLC 400 cHiPLC from Sciex (formerly Eksigent) uses either 3 or 5 μm particles while Waters TRIZAIC™ nanotile and the ionKey/MS systems offer particles with the diameter as low as 1.7 μm. Before the Agilent HPLC-chip was discontinued, chips containing 3.5 to 5 μm ID particles were offered with a variety of stationary phase selectivity. More versatile PicoChip is produced by New Objective, which is compatible with most of the mass spectrometers and is offered with a wide variety of particles. The mentioned commercial chip stationary phases are summarized in Table 25.2. Apart from particles, there are further carriers of stationary phase that can benefit from the fabrication process of LC chips or that are more suitable for in situ fabrication. Namely, those are monolithic, open tubular stationary phases and stationary phases made of microstructures. Due to their high permeability, monoliths

25.4 Microchips

Table 25.2 Examples of stationary phases in commercial chips. Commercial name

Extraction column

Analytical column

Mass spectrometer

Agilent

HPLC-Chip/ MS

40–500 nL channel, packed with ZORBAX 300SB-C18, 5-μm i.d. particles

Compatible with Agilent mass spectrometer/ nanospray source

Waters

TRIZAIC™ nano tile

180 μm  20 mm, packed with C18, 5-μm particles

75  50-μm cross-section and a length of 45 mm ZORBAX 300SB-C18 3.5-μm Elution at 100 nL/min 85 μm  100 mm, packed with HSS T3, 1.8-μm particles

ionKey/MS



85 μm  100 mm, packed with: C18 BEH and CSH, 1.7 μm particles

Sciex (Eksigent’s)

cHiPLC®Nanoflex

New Objective

PicoChip

0.5 mm  200 μm, and packed with ChromXP C18-CL 3-μm and ChromXP C4-CL –

15 cm  75 μm, and packed with ChromXP C18CL 3-μm and ChromXP C4-CL 50–150 μm ID  15 μm, packed with diverse particles

Producer

Compatible with Xevo series mass spectrometers from Waters Compatible with Xevo series mass spectrometers from Waters Compatible with any mass spectrometer/ nanospray source Compatible with any mass spectrometer/ nanospray source

Table reprinted and adapted with permission from Vargas Medina et al. [50].

significantly reduce column back pressure, their formation/polymerization can be precisely controlled, and they do not require frits. On the other hand, swelling or shrinking of monoliths upon contact with some organic mobile phases can be a problem due to possible retraction of the monolith from the separation channel walls [51]. Microstructures made directly during the chip fabrication are another alternative intrinsic to the chip format. Most often they are composed of arrays of geometrical shapes such as pillars dubbed as “collocated monolithic support structures” or shortly COMOSS [52] and μPACs (micropillar array columns) (Figure 25.7) [53,54]. Recently, PharmaFluidics company presented the second generation of the μPACs where pillar widths and interpillar distances were decreased from 5 and 2.5 μm to 2.5 and 1.25 μm, respectively. Such improvement resulted in a more than 10-fold increase in the detection sensitivity for proteomics samples [55]. Fluid distribution plays a significant role in the minimization of dispersion of injected samples and

657

658

CHAPTER 25 Miniaturization and microchips

FIGURE 25.7 (A) Schematic overview, and (B), (C) SEM images of the porous shell micro-pillar array column (μPAC). Reprinted with permission from Callewaert et al. [53].

separated analytes [47,49,54]. Channel turns, and geometry of the chip has to be adequately designed to minimize dispersion [47,56]. Popular techniques of additive manufacturing (3D printing) are also utilized in the design of LC chips and columns (Figure 25.5C) [57–59]. 3D printing enables complete control of printed stationary phase, which enables fabrication of perfectly organized micro-structures and could lead to fully integrated LC in the future [59]. Open narrow LC chip channels have high permeability and can be easily fabricated by coating the channel walls with a suitable stationary phase [59]. However, such columns suffer from limited capacity, and the narrow channels tend to clog easily [49].

25.4.3 Analyte injection Reliable injection of adequate sample volume is a prerequisite for efficient separation. Due to very low volumes of separation columns in LC chips, the sample volume that can be injected is limited (see Section 25.3). The injection in LC chips is usually performed by in-chip structures serving either as sample loops, injection crosssections, or as trap columns [60,61]. On-column focusing is suitable for diluted samples since it does not require additional chip structures and sample components concentrate at the column inlet using a mobile phase of low eluting power. The low-molecular matrix compounds from the

25.4 Microchips

sample (e.g., salts, buffers, saccharides) are rinsed out during sampling, minimizing their influence on the separation. Alternatively, a trap column with stationary phase different from a separation column can be embedded prior to the separation column and sample can be released using mobile phase modulation [62]. Another option is to utilize separate sample loops at the expense of incorporating a switching valve. LC chips provide an excellent opportunity to design such structures as a part of the chip. Such an approach limits the maximum injected volume and prevents column overloading. One of the main benefits is fast sample injection. Moreover, switching between two or more trap columns can transfer fractions between subsequent separation dimensions [62]. Electrokinetic sample injection is used widely for its minimal instrumental requirements and easy integration into the chip structure. Sample discrimination can occur in electrokinetic injection based on the net charge of the sampled molecules [61,63]. On the other side, sample enrichment can be utilized with electrokinetic injection (e.g., nanochannel preconcentration, field-amplified stacking, moving boundary stacking, isotachophoresis, IEF, pH junction preconcentration, and electric field gradient focusing) [63]. However, some of those techniques have specific requirements on the composition of sample solution (e.g., buffers, high salt concentrations, carrier ampholytes), which could negatively affect the efficiency of chip LC performance.

25.4.4 Two-dimensional LC While separation efficiency is gradually improving by better column technologies, there is still not enough resolving power for very complex mixtures originating, especially from proteomics and other “omics” dealing with the identification of thousands of analytes. The answer can be a multi-dimensional separation where the peak capacity of orthogonal separations is multiplied [64,65]. A comprehensive 2D separation usually requires a long time to analyze a sample due to the sequential processing of fractions from the first dimension. For this reason, heart-cutting 2D analysis is often adopted. In this case, only selected fractions from the first dimension are analyzed to save the analysis time (Figure 25.8) [48]. Depending on the particular separation method in each dimension, the mobile phase has to be compatible to prevent distortion or breakthrough of peaks from the first dimension [66]. LC chips can benefit from the parallel analysis of fractions from the previous orthogonal dimensions [67,68]. Spatial multi-dimensional LC enables highly parallel analysis as all the fractions from one dimension are sampled from their actual position in the column to the next dimension (Figure 25.9) [66,67]. One of the main issues in spatial multi-dimensional LC is flow confinement and the establishment of an even flow rate within all channels of the following dimensions. The design of such channel structures is not simple and often requires the use of fluid dynamics simulations [68]. Application of external/internal valves, constriction of channel size, or application of low permeable monolithic filling can be used to modulate the flow to desired directions at required rates. Moreover, stationary phases

659

FIGURE 25.8 (1) Schematic overview of the valving and fluidics of the 2D chip-HPLC/MS setup in elution mode. (2) Multiple heart-cutting 2D chip-HPLC/MS analysis of a tryptic digest of BSA. Reprinted with permission from [48].

25.4 Microchips

FIGURE 25.9 (A) Schematic view of a system for spatial comprehensive three-dimensional liquid chromatography and (B) first prototype microfluidic chip for spatial 3D-LC. Reprinted with permission from Themelis et al. [69] originally adapted from Beaver and Guiochon [70], Wouters et al. [71].

containing more modes of separation or thermo/pH/photo-responsive materials can be utilized to produce different selectivity. Careful selection of the stationary phases and operating conditions/mobile phase compositions enables the high resolving power of such spatial multi-dimensional LC chips. Nevertheless, there is still a considerable amount of development waiting for detection at the parallel outputs, and many technical obstacles have to be solved before this technology becomes applicable in practice.

25.4.5 Microfluidic LC coupled to mass spectrometry Liquid chromatography coupled to mass spectrometry (LC–MS) is currently one of the workhorses in modern analytical laboratories [72]. Electrospray (ES) is a prime ionization technique used for the transfer of the LC effluent to the mass spectrometer [73]. The low flow rates of the miniaturized LC systems bring several advantages: (1) simplifying the ES interface (less or no drying gas required), (2) more efficient ionization process with lower ion suppression, and (3) improved transmission of ions into the MS [74,75]. A stable ES plume demands an emitter of high quality. This is given by its physical parameters (symmetry and sharpness) being particularly important at low flow rates [19,76–79]. The chemical nature of the emitter should consider the material of a chip as well as the composition of a mobile phase. Thus, a device with a hydrophilic emitter (e.g., glass) operating with a hydrophilic mobile phase provides improved ES stability if hydrophobically treated [77,80].

661

662

CHAPTER 25 Miniaturization and microchips

There are various approaches for the fabrication of the emitter. One of the popular strategies is based on the fabrication of the separation chip and the emitter independently. Such an external emitter is attached to the edge of the chip. The integration requires an exact alignment to minimize dead volume associated with channel–capillary junctions. While this is challenging for glass/silicon chip devices, the nearly zero-dead volume conditions seem to be more easily achievable in polymer-based devices [21,81]. The externally attached emitter is typically constructed from a fused silica capillary sharpened to a tip [38] or a capillary of tiny external diameter [82,83]. An interesting approach was reported by the group of Oleschuk, which attached microstructured optical fiber with 54 channels of 4 μm in diameter [21]. The fiber acted as a multichannel emitter as well as a retaining frit, thus simplifying the overall chip design. Microfabrication techniques can address the challenge of effective ES ionization coupling by direct integration of an emitter, e.g., by precise cleaving (dicing saw or laser cutting) of the edge of the chip, sometimes in combination with further sharpening (Figure 25.10A) [77,84,86]. Freire et al. introduced spraying from the corner of a thin chip, thus eliminating the cutting procedure [78,80,85] (Figure 25.10B). Photolithography and chemical etching can provide full control of the dimensions of the monolithically integrated emitter with highly repeatable results (Figure 25.10C) [31,40,45,79] with the potential for mass production. The generation of electrospray plume requires a sufficient electric field between the emitter tip and the MS orifice. The electrical connection at the emitter can be achieved via an auxiliary liquid mixed with the effluent shortly before the ionization process or via a solid electrode being in direct physical contact with the effluent. The composition of the auxiliary liquid, called sheath liquid, frequently stabilizes the

FIGURE 25.10 (A) mLC-MS device with monolithically integrated emitter fabricated by dicing and grinding of the chip edge [84]. (B) Spraying from the corner of a thin chip [85]. (C) ES emitter obtained via microfabrication [79]. Reprinted with permission.

25.5 Electrochemical and optical detection

FIGURE 25.11 Microfluidic LC device designed for: (A) atmospheric pressure ionization (APPI), and (B) onchip MALDI. (A) Reprinted with permission from Haapala et al. [38]. (B) Reprinted from Lazar and Kabulski [88].

electrospray plume. Thus, it is the method of choice if the analysis involves substantial changes in physicochemical properties of the mobile phase (e.g., solvent plug or gradient elution) [45,84]. The sub-μL/min flow rates in microfluidic devices make electrospray ionization more robust, allowing the mLC-MS to operate in a sheathless mode [40,87]. While the prototyping usually relies on the externally attached platinum wires, more sophisticated constructions use on-chip electrodes, commonly fabricated via e-beam/vacuum evaporation or sputtering [12,86]. Electrospray ionization is a dominant ionization source for coupling of mLC to MS, and only a few alternatives have been reported. Haapala et al. developed mLC-MS device operating with atmospheric pressure photoionization (APPI), suitable for nonpolar analytes (Figure 25.11A) [38]. The separated zones were mixed with nebulizer gas in the vaporizer channel (up to 400°C) and sprayed out as a confined jet. The early efforts for coupling LC with MALDI MS were reviewed in 2002 [89]. Today, the MALDI MS is seldom used, mostly offline [90]. Lazar and Kabulski introduced a microfluidic platform for chromatographic separation and on-chip MALDI detection (Figure 25.11B) [88]. Their technology relied on the orthogonal electro-driven transfer of analytes from the separation column via hundreds of shallow microchannels into an array of micro reservoirs. Once the sample was collected, the solvent was evaporated, and the matrix was added.

25.5 Electrochemical and optical detection in miniaturized and chip-based setups Nowadays, two main approaches to LC miniaturization include nano-LC, which uses columns with internal diameters less than 100 μm (nanocolumns), and microfluidic LC, which uses a separation channel incorporated in a structure of microchip (μLC chip). When compared to conventional HPLC, the volume of the peak eluted from nanocolumn or μLC chip is significantly decreased. Therefore, the detection cell has

663

664

CHAPTER 25 Miniaturization and microchips

to be adequately small, typically only few nanoliters. Moreover, the detection system applied in LC has to be resistant to mobile phase composition and temperature changes and capable of a higher detector sampling rate when a rapid analysis is performed [91–93]. Mass spectrometry is a powerful, sensitive, and selective technique that has already been established as a gold standard of detection in LC. However, some alternative detection techniques may benefit from lower hardware demands, higher sensitivity, and/or specificity. Optical methods such as UV–visible absorption (5.3), refraction (5.4), fluorescence [94] (5.2), and chemiluminescence [95] detection are frequently used for sensing purposes. While all the optical detections deal with limited detection path length resulting in limited sensitivity, some recently described systems based on atomic absorption spectrometry [96], tip-enhanced Raman spectrometry [97], differential interference contrast thermal lens microscopy [98,99], coherent anti-Stokes Raman scattering [100], or photothermal detection in nanocapillary [101] may provide improved detection limits for specific analytes.

25.5.1 Electrochemical detection Electrochemical detection monitors the changes in an electrical signal due to the electrochemical reactions on the surface of an electrode. Conductivity [102,103], amperometry [104–108], and potentiometry are the most common electrochemical detection approaches in microfluidic devices. Their popularity originates from the possibility of integrating microelectrodes inside the separation channel during the fabrication process of μLC chip. Thanks to that, electrochemical detection is considered most facile and simplest to miniaturize among all detection techniques in microchip devices in general. For some LC techniques such as ion chromatography, usage of potentiometric sensing is considered a natural choice due to significant changes in the eluent conductivity. Many small molecules can be directly detected with electrochemical signals without the necessity of derivatization [109,110]. However, in general, electrochemical detectors suffer from poor selectivity. Additionally, the lifetime of electrodes is usually limited.

25.5.2 Fluorescence detection Fluorescence spectroscopy detectors [101,111–114] do not suffer from sensitivity issues and offer acceptable linearity range and compatibility to gradient elution. The microchip substrate needs to be transparent for excitation and emitted light. The detector is typically placed at the outlet of the separation channel and usually consists of a light source, optical filters, and a light detector. With high-intensity laser excitation, fluorescence belongs to the most sensitive detection methods, with single-molecule detection limits [115]. Besides laser, a lamp or LED-based fluorescence represents a cost-effective option while sacrificing narrow excitation band and/or sensitivity [116–122].

25.5 Electrochemical and optical detection

The application of fluorescence detectors is limited to a small group of analytes possessing native fluorescence (e.g., dyes). Other compounds require fluorescence labeling, a chemical reaction of target analyte with a fluorescent reagent. Labeling belongs to well-established strategies allowing reliable detection of specific analytes such as oligonucleotides or proteins [123]. Even though the sensing of native fluorescence of some biomolecules is possible, it requires the use of a UV laser (500 nm) has been tested in glass and polymer-based microchips [126,127]. As a result, label-free fluorescence is possible on glass and polymeric microchips [128,129].

25.5.3 Absorption detection Absorption detection is the most common detection method in conventional liquid chromatography. However, a μLC detector requires a significant reduction of volume to preserve the separation resolution. The limited channel dimensions significantly impact the path length of the detector cell (typically only several tens of micrometers or less) causing loss of sensitivity following Beer–Lambert–Bouguer law. As a result, different approaches have been used to increase the path length, such as bubble cells, U- or Z-shape detection cells, multireflection cells, and embedded waveguides [130–135]. Moreover, the UV absorption of the microchip substrate makes UV detection more difficult. For example, the glass and plastic materials used for microchip construction may have strong UV absorbance, while quartz and PDMS materials are advantageous alternatives given their superior optical transparency [136]. Hybrid microchips have been explored where the optical merits of different materials are incorporated [137,138]. Despite the continuous improvements, UV–vis detection plays a relatively minor role in miniaturized LC [139–142] and thus is rather applied in nanocolumn LC than in microfluidic LC.

25.5.4 Refraction detection Refractive-index-based sensors offer nondestructive and universal detection for liquid chromatographic separations; however, low sensitivity, limited dynamic range, and sensitivity to thermal perturbations limit the utility of refractive-index detection. As such, refractive index detectors are used solely for isocratic separations, e.g., for the analysis of sugars. Lately, several approaches have been suggested to overcome these limitations, including integrated sensors based on silicon photonic microring resonators [10], Fabry–Perot resonator [143], or dual-beam systems [91,144].

665

666

CHAPTER 25 Miniaturization and microchips

25.6 Chip-based HPLC instruments The early instrumentation brought to the market by some of the HPLC instrumentation manufacturers offers docking station for the chip devices installed in a Plug & Play cartridge. The docking station serves as an interface between the nanoliter flow pump and detector, typically MS. The main advantage of Plug & Play design is straightforward and quick column replacement. In 2005, Agilent Technologies introduced the first HPLC-Chip/MS instrument. The chip was fabricated from multiple polyimide film layers and integrated enrichment and analytical columns, injection channel, electrospray tip, and the necessary connections [145]. However, the production of this device was discontinued in 2018. cHiPLC by Sciex (formerly by Eksigent) [146] is a docking station for three cartridges with microfluidic chips acting as a sample loop, trap column, and separation column. Cylindrical channels are made of fused silica, and columns use a unique weir structure instead of conventional frits. Ten-port two-position valve serves as an interface between individual chips with a dead volume of less than 1 nL. Workflows, that can be performed, include direct injection, trap-elute, two-column switching, and the use of two columns in series. Columns and traps are offered in 75, 300, and 500 μm inner diameter (ID) sizes, and columns lengths vary from 5 to 15 cm. The ionKey from Waters (Figure 25.12) [147] is a docking station for one chip cartridge, and it is designed to be mounted directly on the inlet of ESI-MS and coupled with ACQUITY® UPLC M-Class system. The chip is made of ceramic

FIGURE 25.12 Waters ionKey docking station and the chip device installed in a Plug & play cartridge. Reprinted with permission.

25.7 Portable HPLC

material and contains channel (150 or 300 μm ID, length 50 or 100 mm) packed with sub-2-μm silica-based or hybrid particles. It may contain one additional empty flow channel. The cartridge includes a metal spraying tip for ESI-MS.

25.7 Portable HPLC Miniaturization of technologies typically leads to battery-powered portable devices. First attempts to make liquid chromatographs portable can be traced to the early 80s for analysis of pesticides [148] and primary aromatic amines in coal [149]. Since then, the topic was discussed rarely until the last decade (e. g. [150–152]), together with the first commercial instruments entering the market. A detailed evolution of the field can be found elsewhere [153,154]. The fact that portable LC instruments would hardly be possible without progress in other fields, including electronics, optics, and material science, is indisputable. Light-emitting diodes are being constantly pushed to shorter wavelengths (currently down to 235 nm), and progress in battery technology leads to decreased dimensions and weight and increased capacity. While current technologies allow the production of portable HPLC instruments, in-field analyses are often hindered by the necessity of sample treatment, which typically requires other laboratory equipment (shakers, centrifuges, ultrasonic baths, etc.) and can be time-consuming. The wide application of portable HPLC outside the laboratory thus requires further progress in the development of new, simple, and rapid sample treatment techniques. Another important aspect is the development of portable mass spectrometers, which would significantly increase the applicability and accelerate the further development of portable HPLC instruments. Several current commercial portable HPLC systems are shown in Figure 25.13. Specifications are listed in Table 25.3, and short comments are given below.

FIGURE 25.13 Commercial portable HPLC. (A) Axcend Focus LC, (B) Smart LifeLC, and (C) Light Lab 3. Pictures reprinted with permission.

667

668

CHAPTER 25 Miniaturization and microchips

Table 25.3 Specification of commercial portable LC instruments. Dimensions (cm  cm  cm) Weight (kg) Max. pressure (bar) Flow rate range (μL/min) Gradient capabilities Column I. D. (mm) Column length (mm) Column heater Detection wavelength (nm) Detection flow cell Detector noise (μAU) Retention time RSD % Battery Operating time (hours) Price (USD)

Axcend Focus LC

Smart LifeLC

Light Lab 3

20.1  23.1  32.0

41  32  17

35.8  21  17.5

8.3 689 0.5–10 Yes 0.15 to 0.30

11.3 207 1–5000 Yes 4.6

6.2 7 1–6000 No Proprietary wide-bore

50–250 Up to 80°C 235, 255, and/or 275 On column 7.6–13 Up to 1% 12 V 10.5 Ah LiFePO4 10 47,500

35–200 On demand 255 or 280 or 415 10 mm, 15 μL 30

No Multipleproprietary

12 V

14.8 V 4.4 Ah Li-ion

9 12,000

8 13,800

The Axcend Focus LC originated from research conducted at Brigham Young University by Lee et al. with engineering support from VICI Valco [155,156] and has been manufactured and marketed by Axcend since 2018. It consists of two piston-based syringe pumps and custom-made stop-flow injector. Capillary column and single- or dual-wavelength LED-based UV-absorption detector are mounted together in a replaceable cartridge. Application notes of the manufacturer demonstrate separation of pharmaceuticals, cannabinoids, aromatic compounds, drugs of abuse, and selected proteins. Smart LifeLC portable HPLC is produced by PolyLC since January 2018 and includes two reciprocating piston pumps (automatic piston wash), a manual injection valve (external sample loop), an analytical column, and a single-wavelength LEDbased UV-absorption detector. This system has been applied for analyzing the major hemoglobin variants (including diabetes marker A1c), cannabinoids, and vitamins [157]. Light Lab 3 by Orange Photonics is a HPLC-based portable cannabinoids analyzer. It consists of an isocratic low-pressure pump, manual injector, and custommade wide-bore reverse-phase column. Detection and quantitation are based on UV spectrometry and multi-dimensional non-linear regression. Screening for the presence of up to 12 cannabinoids takes 10.5 min.

References

Acknowledgments The authors would like to thank the Czech Science Foundation for the provided financial support (Grant No. 20-14069Y) and the Institute of Analytical Chemistry, Czech Academy of Sciences in Brno, Czech Republic for the institutional support (RVO:68081715).

References [1] Analytical Scientific Instruments. HPLC/UHPLC Pump Replacement Parts., 2021, https://www.hplc-asi.com/content/files/Replacement%20Parts%20Brochure% 20Website%202021%20-%20no%20price.pdf. [Accessed 20 December 2021]. [2] Omniseal_Solutions. Omniseal Polymers - Spring-energized seals., 2021, https://www. omniseal-solutions.com/sites/hps-mac3-seals-seals/files/2021-09/Omniseal-SpringEnergized-Seal-Handbook-EU_0.pdf. [Accessed 20 December 2021]. [3] Broeckhoven K, Dong MW, Shoykhet K. Modern HPLC pumps: perspectives, principles, and practices. LCGC Eur 2019;32:410–9. [4] Zhao X, Xie X, Sharma S, Tolley LT, Plistil A, Barnett HE, et al. Compact ultrahighpressure nanoflow capillary liquid chromatograph. Anal Chem 2017;89:807–12. [5] VICI. True Nano UHPLC, https://www.vici.com/uhplc/TrueNano-HPLC.php. [Accessed 31 January 2022]. [6] Chen LX, Ma JP, Guan YF. Study of an electroosmotic pump for liquid delivery and its application in capillary column liquid chromatography. J Chromatogr A 2004;1028: 219–26. [7] Borowsky JF, Giordano BC, Lu Q, Terray A, Collins GE. Electroosmotic flow-based pump for liquid chromatography on a planar microchip. Anal Chem 2008;80:8287–92. [8] Wang W, Gu C, Lynch KB, Lu JJ, Zhang Z, Pu Q, et al. High-pressure open-channel onchip electroosmotic pump for nanoflow high performance liquid chromatography. Anal Chem 2014;86:1958–64. [9] Wu Y, Zhang W, Chen Y, Chen Z. Electroosmotic pump-supported molecularly imprinted monolithic column for capillary chromatographic separation of nitrophenol isomers. Electrophoresis 2015;36:2881–7. [10] Zhou L, Lu JJ, Gu C, Liu S. Binary electroosmotic-pump nanoflow gradient generator for miniaturized high-performance liquid chromatography. Anal Chem 2014;86: 12214–9. [11] Wouters B, De Vos J, Desmet G, Terryn H, Schoenmakers PJ, Eeltink S. Design of a microfluidic device for comprehensive spatial two-dimensional liquid chromatography. J Sep Sci 2015;38:1123–9. [12] Ishida A, Natsume M, Kamidate T. Microchip reversed-phase liquid chromatography with packed column and electrochemical flow cell using polystyrene/poly(dimethylsiloxane). J Chromatogr A 2008;1213:209–17. [13] Ishida A, Fujii M, Fujimoto T, Sasaki S, Yanagisawa I, Tani H, et al. A portable liquid chromatograph with a battery-operated compact electroosmotic pump and a microfluidic chip device with a reversed phase packed column. Anal Sci 2015;31:1163–9. [14] Novotny J, Ostatna V, Foret F. Electrochemical analysis of glycoprotein samples prepared on a pneumatically-controlled microfluidic device. Electroanalysis 2019;31: 1994–2000.

669

670

CHAPTER 25 Miniaturization and microchips

[15] Yin H, Killeen K, Brennen R, Sobek D, Werlich M, Van De Goor T. Microfluidic chip for peptide analysis with an integrated HPLC column, sample enrichment column, and nanoelectrospray tip. Anal Chem 2005;77:527–33. [16] Kourmpetis I, Kastania AS, Ellinas K, Tsougeni K, Baca M, De Malsche W, et al. Gradient-temperature hot-embossing for dense micropillar array fabrication on thick cyclo-olefin polymeric plates: an example of a microfluidic chromatography column fabrication. Micro Nano Eng 2019;5, 100042. [17] Simon U, Dimartino S. Direct 3D printing of monolithic ion exchange adsorbers. J Chromatogr A 2019;1587:119–28. [18] Choudhury IA, Shirley S. Laser cutting of polymeric materials: an experimental investigation. Opt Laser Technol 2010;42:503–8. [19] Lu Y, Liu F, Lion N, Girault HH. Dual-channel electrospray microchip. J Am Soc Mass Spectrom 2013;24:454–7. [20] Deshmukh SS, Goswami A. Hot Embossing of polymers – a review. Mater Today Proc 2019;26:405–14. [21] Mats L, Gibson GTT, Oleschuk RD. Plastic LC/MS microchip with an embedded microstructured fibre having the dual role of a frit and a nanoelectrospray emitter. Microfluid Nanofluidics 2014;16:73–81. [22] Weisgrab G, Ovsianikov A, Costa PF. Functional 3D printing for microfluidic chips. Adv Mater Technol 2019;4:1900275. [23] Chan HN, Shu Y, Xiong B, Chen Y, Chen Y, Tian Q, et al. Simple, cost-effective 3D printed microfluidic components for disposable, point-of-care colorimetric analysis. ACS Sensors 2016;1:227–34. [24] Begolo S, Zhukov DV, Selck DA, Li L, Ismagilov RF. The pumping lid: investigating multi-material 3D printing for equipment-free, programmable generation of positive and negative pressures for microfluidic applications. Lab Chip 2014;14:4616–28. [25] Lee WS, Yeo KS, Andriyana A, Shee YG, Mahamd Adikan FR. Effect of cyclic compression and curing agent concentration on the stabilization of mechanical properties of PDMS elastomer. Mater Des 2016;96:470–5. [26] Kiran Raj M, Chakraborty S. PDMS microfluidics: A mini review. J Appl Polym Sci 2020;137:48958. [27] Jeong OC, Konishi S. Controlling the size of replicable polydimethylsiloxane (PDMS) molds/stamps using a stepwise thermal shrinkage process. Microelectron Eng 2011;88:2286–9. [28] Kabiri S, Kurkuri MD, Kumeria T, Losic D. Frit-free PDMS microfluidic device for chromatographic separation and on-chip detection. RSC Adv 2014;4:15276–80. [29] Levy MH, Goswami S, Plawsky J, Cramer SM. Parameters governing the formation of photopolymerized silica sol-gel monoliths in PDMS microfluidic chips. Chromatographia 2013;76:993–1002. [30] Arscott S. SU-8 as a material for lab-on-a-chip-based mass spectrometry. Lab Chip 2014;14:3668–89. [31] Nissil€a T, Sainiemi L, Franssila S, Ketola RA. Fully polymeric integrated microreactor/ electrospray ionization chip for on-chip digestion and mass spectrometric analysis. Sens Actuators B 2009;143:414–20. [32] Carlier J, Chuda K, Arscott S, Thomy V, Verbeke B, Coqueret X, et al. High pressureresistant SU-8 microchannels for monolithic porous structure integration. J Micromech Microeng 2006;16:2211–9.

References

[33] Thurmann S, Mauritz L, Heck C, Belder D. High-performance liquid chromatography on glass chips using precisely defined porous polymer monoliths as particle retaining elements. J Chromatogr A 2014;1370:33–9. [34] Ericson C, Holm J, Ericson T, Hjerte S. Electroosmosis- and pressure-driven chromatography in chips using continuous beds. Anal Chem 2000;72:81–7. [35] Pal P, Swarnalatha V, Rao AVN, Pandey AK, Tanaka H, Sato K. High speed silicon wet anisotropic etching for applications in bulk micromachining: a review. Micro Nano Syst Lett 2021;9. [36] Prem P, Veerla S, Avvaru VNR, Ashok KP, Hiroshi T, Kazuo S. A review of wet chemical etching of glasses in hydrofluoric acid based solution for thin film silicon solar cell application. Micro Nano Syst Lett 2021;9. [37] Kara´sek P, Grym J, Roth M, Planeta J, Foret F. Etching of glass microchips with supercritical water. Lab Chip 2015;15:311–8. [38] Haapala M, Saarela V, Po´l J, Kolari K, Kotiaho T, Franssila S, et al. Integrated liquid chromatography-heated nebulizer microchip for mass spectrometry. Anal Chim Acta 2010;662:163–9. [39] Park JH, Lee NE, Lee J, Park JS, Park HD. Deep dry etching of borosilicate glass using SF6 and SF 6/Ar inductively coupled plasmas. Microelectron Eng 2005;82:119–28. [40] Xie J, Miao Y, Shih J, Tai YC, Lee TD. Microfluidic platform for liquid chromatography-tandem mass spectrometry analyses of complex peptide mixtures. Anal Chem 2005;77:6947–53. [41] Durucan O, Wu K, Viehrig M, Rindzevicius T, Boisen A. Nanopillar-assisted SERS chromatography. ACS Sensors 2018;3:2592–8. [42] Herth E, Edmond S, Bouville D, Cercus JL, Bayle F, Cambril E. Micro-/nanopillars for micro- and nanotechnologies using inductively coupled plasmas. Phys Status Solidi Appl Mater Sci 2019;216. [43] Osipov AA, Iankevich GA, Alexandrov SE. Monocrystalline quartz ICP etching: road to high-temperature dry etching. Plasma Chem Plasma Process 2020;40:423–31. [44] Taylor LC, Lavrik NV, Sepaniak MJ. High-aspect-ratio, silicon oxide-enclosed pillar structures in microfluidic liquid chromatography. Anal Chem 2010;82:9549–56. [45] Sainiemi L, Nissil€a T, Kostiainen R, Franssila S, Ketola RA. A microfabricated micropillar liquid chromatographic chip monolithically integrated with an electrospray ionization tip. Lab Chip 2012;12:325–32. [46] Kecskemeti A, Gaspar A. Particle-based liquid chromatographic separations in microfluidic devices - a review. Anal Chim Acta 2018;1021:1–19. [47] Haghighi F, Talebpour Z, Nezhad AS. Towards fully integrated liquid chromatography on a chip: evolution and evaluation. TrAC Trends Anal Chem 2018;105:302–37. [48] Piendl SK, Geissler D, Weigelt L, Belder D. Multiple heart-cutting two-dimensional chip-HPLC combined with deep-UV fluorescence and mass spectrometric detection. Anal Chem 2020;92:3795–803. [49] Vargas Medina DA, Maciel EVS, Lanc¸as FM. Miniaturization of liquid chromatography coupled to mass spectrometry. 3. Achievements on chip-based LC–MS devices. TrAC - Trends Anal Chem 2020;131, 116003. [50] Vargas Medina DA, Maciel EVS, Lanc¸as FM, Soares Maciel EV, Lancas FM, Maciel EVS, et al. Miniaturization of liquid chromatography coupled to mass spectrometry. 3. Achievements on chip-based LC–MS devices. TrAC - Trends Anal Chem 2020;131, 116003.

671

672

CHAPTER 25 Miniaturization and microchips

[51] Knob R, Sahore V, Sonker M, Woolley AT. Advances in monoliths and related porous materials for microfluidics. Biomicrofluidics 2016;10. [52] He B, Tait N, Regnier F. Fabrication of nanocolumns for liquid chromatography. Anal Chem 1998;70:3790–7. [53] Callewaert M, De Beeck JO, Maeno K, Sukas S, Thienpont H, Ottevaere H, et al. Integration of uniform porous shell layers in very long pillar array columns using electrochemical anodization for liquid chromatography. Analyst 2014;139:618–25. [54] Rozing G. Micropillar array columns for advancing nanoflow HPLC. Microchem J 2021;170, 106629. [55] Stejskal K, Op de Beeck J, D€ urnberger G, Jacobs P, Mechtler K. Ultrasensitive nanoLC-MS of subnanogram protein samples using second generation micropillar array LC technology with orbitrap exploris 480 and FAIMS PRO. Anal Chem 2021;93: 8704–10. [56] Salmean C, Dimartino S. 3D-printed stationary phases with ordered morphology: state of the art and future development in liquid chromatography. Chromatographia 2019;82:443–63. [57] Gupta C, Daechsel AK, Chauhan A. Interaction of ionic surfactants with corneamimicking anionic liposomes. Langmuir 2011;27:10840–6. [58] Wang L, Pumera M. Recent advances of 3D printing in analytical chemistry: focus on microfluidic, separation, and extraction devices. TrAC - Trends Anal Chem 2021;135, 116151. [59] Broeckhoven K, Desmet G. Advances and innovations in liquid chromatography stationary phase supports. Anal Chem 2021;93:257–72. [60] Vasconcelos Soares Maciel E, de Toffoli AL, Sobieski E, Domingues Naza´rio CE, Lanc¸as FM. Miniaturized liquid chromatography focusing on analytical columns and mass spectrometry: a review. Anal Chim Acta 2020;1103:11–31. [61] Yuan X, Oleschuk RD. Advances in microchip liquid chromatography. Anal Chem 2017;90:283–301. [62] Blue LE, Franklin EG, Godinho JM, Grinias JP, Grinias KM, Lunn DB, et al. Recent advances in capillary ultrahigh pressure liquid chromatography. J Chromatogr A 2017;1523:17–39. [63] Cong Y, Katipamula S, Geng T, Prost SA, Tang K, Kelly RT. Electrokinetic sample preconcentration and hydrodynamic sample injection for microchip electrophoresis using a pneumatic microvalve. Electrophoresis 2016;37:455–62. [64] Gilar M, Olivova P, Daly AE, Gebler JC. Orthogonality of separation in twodimensional liquid chromatography. Anal Chem 2005;77:6426–34. [65] Schure MR, Davis JM. Orthogonality measurements for multidimensional chromatography in three and higher dimensional separations. J Chromatogr A 2017;1523:148–61. [66] De Vos J, Stoll D, Buckenmaier S, Eeltink S, Grinias JP. Advances in ultrahigh-pressure and multi-dimensional liquid chromatography instrumentation and workflows. Anal Sci Adv 2021;2:171–92. [67] Themelis T, De Vos J, Dores-Sousa JL, van Assche T, Eeltink S. Engineering solutions for flow control in microfluidic devices for spatial multi-dimensional liquid chromatography. Sens Actuators B 2020;320, 128388. [68] Themelis T, Amini A, De Vos J, Eeltink S. Towards spatial comprehensive threedimensional liquid chromatography: a tutorial review. Anal Chim Acta 2021;1148, 238157.

References

[69] Themelis T, Amini A, De Vos J, Eeltink S. Towards spatial comprehensive threedimensional liquid chromatography: a tutorial review. Anal Chim Acta 2021;1148. [70] Beaver LA, Guiochon GA. System and apparatus for multi-dimensional real-time chromatography [United States Patent no. 4469601]; 1984. [71] Wouters B, Davydova E, Wouters S, Vivo-Truyols G, Schoenmakers PJ, Eeltink S. Towards ultra-high peak capacities and peak-production rates using spatial threedimensional liquid chromatography. Lab Chip 2015;15:4415–22. [72] Rigano F, Tranchida PQ, Dugo P, Mondello L. High-performance liquid chromatography combined with electron ionization mass spectrometry: a review. TrAC - Trends Anal Chem 2019;118:112–22. [73] Tycova A, Prikryl J, Kotzianova A, Datinska V, Velebny V, Foret F. Electrospray: more than just an ionization source. Electrophoresis 2021;42:103–21. [74] Xiong B, Wang L, Wang Y, Bao Y, Jiang S, Ye M. Three-dimensional hydrodynamic focusing microfluidic emitter: a strategy to inhibit sample ion expansion in nanoelectrospray ionization. Analyst 2016;141:177–82. [75] Juraschek R, D€u T, Karas M. Nanoelectrospray-more than just a minimized-flow electrospray ionization source. vol. 37. UTC; 1998. [76] Tycova A, Prikryl J, Foret F. Reproducible preparation of nanospray tips for capillary electrophoresis coupled to mass spectrometry using 3D printed grinding device. Electrophoresis 2016;37:924–30. [77] Lotter C, Heiland JJ, Thurmann S, Mauritz L, Belder D. HPLC-MS with glass chips featuring monolithically integrated electrospray emitters of different geometries. Anal Chem 2016;88:2856–63. [78] Freire SLS, Yang H, Wheeler AR. A practical interface for microfluidics and nanoelectrospray mass spectrometry. Electrophoresis 2008;29:1836–43. [79] Mery E, Ricoul F, Sarrut N, Constantin O, Delapierre G, Garin J, et al. A silicon microfluidic chip integrating an ordered micropillar array separation column and a nanoelectrospray emitter for LC/MS analysis of peptides. Sens Actuators B 2008;134:438–46. [80] Chambers AG, Ramsey JM. Microfluidic dual emitter electrospray ionization source for accurate mass measurements. Anal Chem 2012;84:1446–51. [81] Dietze C, Scholl T, Ohla S, Appun J, Schneider C, Belder D. Rapid prototyping of microfluidic chips for dead-volume-free MS coupling. Anal Bioanal Chem 2015;407:8735–43. [82] Lazar IM, Trisiripisal P, Sarvaiya HA. Microfluidic liquid chromatography system for proteomic applications and biomarker screening. Anal Chem 2006;78:5513–24. [83] Lazar IM, Li L, Yang Y, Karger BL. Microfluidic device for capillary electrochromatography-mass spectrometry. Electrophoresis 2003;24:3655–62. [84] Piendl SK, Geissler D, Weigelt L, Belder D. Multiple heart-cutting two-dimensional chip-HPLC combined with deep-UV fluorescence and mass spectrometric detection. Anal Chem 2020;92:3795–803. [85] Mellors JS, Gorbounov V, Ramsey RS, Ramsey JM. Fully integrated glass microfluidic device for performing high-efficiency capillary electrophoresis and electrospray ionization mass spectrometry. Anal Chem 2008;80:6881–7. [86] Peretzki AJ, Schmidt S, Flachowsky E, Das A, Gerhardt RF, Belder D. How electrospray potentials can disrupt droplet microfluidics and how to prevent this. Lab Chip 2020;20:4456–65.

673

674

CHAPTER 25 Miniaturization and microchips

[87] Liu J, Ro KW, Nayak R, Knapp DR. Monolithic column plastic microfluidic device for peptide analysis using electrospray from a channel opening on the edge of the device. Int J Mass Spectrom 2007;259:65–72. [88] Lazar IM, Kabulski JL. Microfluidic LC device with orthogonal sample extraction for on-chip MALDI-MS detection. Lab Chip 2013;13:2055–65. [89] Foret F, Preisler J. Liquid phase interfacing and miniaturization in matrix-assisted laser desorption/ionization mass spectrometry. Proteomics 2002;2:360–72. [90] Ro KW, Liu J, Knapp DR. Plastic microchip liquid chromatography-matrix-assisted laser desorption/ionization mass spectrometry using monolithic columns. J Chromatogr A 2006;1111:40–7. [91] Yuan X, Oleschuk RD. Advances in microchip liquid chromatography. Anal Chem 2018;90:283–301. [92] Haghighi F, Talebpour Z, Nezhad AS. Towards fully integrated liquid chromatography on a chip: evolution and evaluation. TrAC Trends Anal Chem 2018;105:302–37. [93] Lynch KB, Chen A, Liu S. Miniaturized high-performance liquid chromatography instrumentation. Talanta 2018;177:94–103. [94] Harrison DJ, Fluri K, Seiler K, Fan Z, Effenhauser CS, Manz A. Micromachining a miniaturized capillary electrophoresis-based chemical analysis system on a chip. Science (80-) 1993;261. [95] Al Lawati HAJ, Kadavilpparampu AM, Suliman FO. Combination of capillary micellar liquid chromatography with on-chip microfluidic chemiluminescence detection for direct analysis of buspirone in human plasma. Talanta 2014;127. [96] Nagy A, Baranyai E, Gaspar A. Interfacing microfluidic chip-based chromatography with flame atomic absorption spectrometry for the determination of chromium(VI). Microchem J 2014;114. [97] Anderson MS. Microfluidics and chromatography with an atomic force microscope. Anal Chem 2005;77. [98] Shimizu H, Mawatari K, Kitamori T. Femtoliter-scale separation and sensitive detection of nonfluorescent samples in an extended-nano fluidic device. Analyst 2014;139. [99] Shimizu H, Mawatari K, Kitamori T. Detection of nonfluorescent molecules using differential interference contrast thermal lens microscope for extended nanochannel chromatography. J Sep Sci 2011;34. [100] Geissler D, Heiland JJ, Lotter C, Belder D. Microchip HPLC separations monitored simultaneously by coherent anti-stokes Raman scattering and fluorescence detection. Microchim Acta 2017;184. [101] Tsuyama Y, Morikawa K, Mawatari K. Diffraction-based label-free photothermal detector for separation analyses in a nanocapillary. J Chromatogr A 2021;1648. [102] Li X, Chang H. Chip-based ion chromatography (chip-IC) with a sensitive fiveelectrode conductivity detector for the simultaneous detection of multiple ions in drinking water. Microsyst Nanoeng 2020;6. [103] Bai X, Lee HJ, Rossier JS, Reymond F, Schafer H, Wossner M, et al. Pressure pinched injection of nanolitre volumes in planar micro-analytical devices. Lab Chip 2002;2. [104] Wang ZW, Wang WJ, Chen GN, Wang W, Fu FF. Liquid chromatography on a monolithic column microfluidic chip coupled with “three-T” sample injection mode and amperometric detection. J Sep Sci 2010;33. [105] Schlund M, Gilbert SE, Schnydrig S, Renaud P. Continuous sampling and analysis by on-chip liquid/solid chromatography. Sens Actuators B Chem 2007;123.

References

[106] Ding Y, Qi Y, Suo X. Rapid determination of β2-agonists in urine samples by microchip micellar electrokinetic chromatography with pulsed electrochemical detection. Anal. Methods 2013;5. [107] Abia JA, Putnam J, Mriziq K, Guiochon GA. Design and implementation of an array of micro-electrochemical detectors for two-dimensional liquid chromatography-proof of principle. J Chromatogr A 2010;1217. [108] Yang H, Zhang Q, Zhao J, Tu Y. Separation and determination of three phenolic xenoestrogens in industrial wastewater by micellar electrokinetic chromatography on polydimethylsiloxane microchip. Water Air Soil Pollut 2012;223. [109] Bowen AL, Martin RS. Integration of on-chip peristaltic pumps and injection valves with microchip electrophoresis and electrochemical detection. Electrophoresis 2010;31. [110] Vandaveer 4th WR, Pasas-Farmer SA, Fischer DJ, Frankenfeld CN, Lunte SM. Recent developments in electrochemical detection for microchip capillary electrophoresis. Electrophoresis 2004;25. [111] Liu J, Chen CF, Tsao CW, Chang CC, Chu CC, DeVoe DL. Polymer microchips integrating solid-phase extraction and high-performance liquid chromatography using reversed-phase polymethacrylate monoliths. Anal Chem 2009;81. [112] Xia L, Choi C, Kothekar SC, Dutta D. On-chip pressure generation for driving liquid phase separations in nanochannels. Anal Chem 2016;88. [113] Smirnova A, Shimizu H, Pihosh Y, Mawatari K, Kitamori T. On-chip step-mixing in a T-nanomixer for liquid chromatography in extended-nanochannels. Anal Chem 2016;88:10059–64. [114] Kim JY, Cho SW, Kang DK, Edel JB, Chang SI, De MAJ, et al. Lab-chip HPLC with integrated droplet-based microfluidics for separation and high frequency compartmentalisation. Chem Commun 2012;48. [115] Song Y, Takatsuki K, Sekiguchi T, Funatsu T, Shoji S, Tsunoda M. Rapid quantitative method for the detection of phenylalanine and tyrosine in human plasma using pillar array columns and gradient elution. Amino Acids 2016;48. [116] Liu C, Qu Y, Luo Y, Fang N. Recent advances in single-molecule detection on microand nano-fluidic devices. Electrophoresis 2011;32. [117] Liu R, Ishimatsu R, Yahiro M, Adachi C, Nakano K, Imato T. Photometric flow injection determination of phosphate on a PDMS microchip using an optical detection system assembled with an organic light emitting diode and an organic photodiode. Talanta 2015;132. [118] Fang XX, Li HY, Fang P, Pan JZ, Fang Q. A handheld laser-induced fluorescence detector for multiple applications. Talanta 2016;150. [119] Weaver MT, Lynch KB, Zhu Z, Chen H, Lu JJ, Pu Q, et al. Confocal laser-induced fluorescence detector for narrow capillary system with yoctomole limit of detection. Talanta 2017;165. [120] Yang FB, Pan JZ, Zhang T, Fang Q. A low-cost light-emitting diode induced fluorescence detector for capillary electrophoresis based on an orthogonal optical arrangement. Talanta 2009;78. [121] Das AJ, Wahi A, Kothari I, Raskar R. Ultra-portable, wireless smartphone spectrometer for rapid, non-destructive testing of fruit ripeness. Sci Rep 2016;6. [122] Gałuszka A, Migaszewski ZM, Namiesnik J. Moving your laboratories to the field advantages and limitations of the use of field portable instruments in environmental sample analysis. Environ Res 2015;140.

675

676

CHAPTER 25 Miniaturization and microchips

[123] Xiong Y, Tan J, Fang S, Wang C, Wang Q, Wu J, et al. A LED-based fiber-optic sensor integrated with lab-on-valve manifold for colorimetric determination of free chlorine in water. Talanta 2017;167. [124] Schulze P, Belder D. Label-free fluorescence detection in capillary and microchip electrophoresis. Anal Bioanal Chem 2009;393. [125] Schulze P, Ludwig M, Kohler F, Belder D. Deep UV laser-induced fluorescence detection of unlabeled drugs and proteins in microchip electrophoresis. Anal Chem 2005;77. [126] Lapainis T, Scanlan C, Rubakhin SS, Sweedler JV. A multichannel native fluorescence detection system for capillary electrophoretic analysis of neurotransmitters in single neurons. Anal Bioanal Chem 2007;387. [127] Hackl C, Beyreiss R, Geissler D, Jezierski S, Belder D. Rapid prototyping of electrochromatography chips for improved two-photon excited fluorescence detection. Anal Chem 2014;86. [128] Geissler D, Belder D. Two-photon excitation in chip electrophoresis enabling label-free fluorescence detection in non-UV transparent full-body polymer chips. Electrophoresis 2015;36. [129] Hunt HC, Wilkinson JS. Optofluidic integration for microanalysis. Microfluid Nanofluidics 2008;4:53–79. [130] Nguyen NT. Micro-optofluidic lenses: a review. Biomicrofluidics 2010;4. [131] Gustafsson O, Mogensen KB, Ohlsson PD, Liu Y, Jacobson SC, Kutter JP. An electrochromatography chip with integrated waveguides for UV absorbance detection. J Micromech Microeng 2008;18. [132] McNamee SE, Elliott CT, Greer B, Lochhead M, Campbell K. Development of a planar waveguide microarray for the monitoring and early detection of five harmful algal toxins in water and cultures. Environ Sci Technol 2014;48. [133] Lu Q, Copper CL, Collins GE. Ultraviolet absorbance detection of colchicine and related alkaloids on a capillary electrophoresis microchip. Anal Chim Acta 2006;572. [134] Salimi-Moosavi H, Jiang Y, Lester L, McKinnon G, Harrison DJ. A multireflection cell for enhanced absorbance detection in microchip- based capillary electrophoresis devices. Electrophoresis 2000;21. [135] Mogensen KB, Klank H, Kutter JP. Recent developments in detection for microfluidic systems. Electrophoresis 2004;25. [136] G€otz S, Karst U. Recent developments in optical detection methods for microchip separations. Anal Bioanal Chem 2007;387. [137] Nge PN, Rogers CI, Woolley AT. Advances in microfluidic materials, functions, integration, and applications. Chem Rev 2013;113. [138] Ou J, Glawdel T, Ren CL, Pawliszyn J. Fabrication of a hybrid PDMS/SU-8/quartz microfluidic chip for enhancing UV absorption whole-channel imaging detection sensitivity and application for isoelectric focusing of proteins. Lab Chip 2009;9. [139] Shimizu H, Smirnova A, Mawatari K, Kitamori T. Extended-nano chromatography. J Chromatogr A 2017;1490. [140] Smirnova A, Shimizu H, Mawatari K, Kitamori T. Reversed-phase chromatography in extended-nano space for the separation of amino acids. J Chromatogr A 2015;1418. [141] Ellinas K, Tsougeni K, Boulousis G, Speliotis T, Tserepi A, Gogolides E. Phosphopeptide enrichment and separation in an affinity microcolumn on a silicon microchip: Comparison of sputtered and wet-deposited TiO2 stationary-phase. Sen Actuators B Chem 2013;188.

References

[142] Kabiri S, Kurkuri MD, Kumeria T, Losic D. Frit-free PDMS microfluidic device for chromatographic separation and on-chip detection. RSC Adv 2014;4:15276–80. [143] Wade JH, Bailey RC. Refractive index-based detection of gradient elution liquid chromatography using chip-integrated microring resonator arrays. Anal Chem 2014;86. [144] St-Gelais R, Masson J, Peter YA. All-silicon integrated Fabry-rot cavity for volume refractive index measurement in microfluidic systems. Appl Phys Lett 2009;94. [145] Yin H, Killeen K, Brennen R, Sobek D, Werlich M, van de Goor T. Microfluidic chip for peptide analysis with an integrated HPLC column, sample enrichment column, and nanoelectrospray tip. Anal Chem 2005;77:527–33. [146] SCIEX. cHiPLC System, https://sciex.com/products/hplc-products/nanolc-products/ chiplc-system. [Accessed 31 January 2022]. [147] Waters. ionKey/MS, https://www.waters.com/waters/en_US/ionKey-MS- - -micro flow-UPLC-Seperation-with-iKey/nav.htm?cid¼134782630&locale¼en_US. [Accessed 31 January 2022]. [148] Pettit JJ, Davisson CW. Development of a portable liquid chromatograph for the analysis of pesticides. Abstr Pap Am Chem Soc 1980;180, 4-ASCS. [149] Otagawa T, Stetter JR, Zaromb S. Portable liquid chromatograph for analysis of primary aromatic-amines in coal-derived materials. J Chromatogr 1986;360:252–9. [150] Chatzimichail S, Rahimi F, Saifuddin A, Surman AJ, Taylor-Robinson SD, SalehiReyhani A. Hand-portable HPLC with broadband spectral detection enables analysis of complex polycyclic aromatic hydrocarbon mixtures. Commun Chem 2021;4. [151] Li D, Chen H, Ren S, Zhang Y, Yang Y, Chang H. Portable liquid chromatography for point-of-care testing of glycated haemoglobin. Sens Actuators B 2020;305, 127484. [152] Lam SC, Coates LJ, Hemida M, Gupta V, Haddad PR, Macka M, et al. Miniature and fully portable gradient capillary liquid chromatograph. Anal Chim Acta 2020;1101: 199–210. [153] Grinias JP. The potential for portable capillary liquid chromatography. Spec Issues 2020;6:15–24. [154] Rahimi F, Chatzimichail S, Saifuddin A, Surman AJ, Taylor-Robinson SD, SalehiReyhani A. A review of portable high-performance liquid chromatography: the future of the field? Chromatographia 2020;83:1165–95. [155] Sharma S, Plistil A, Simpson RS, Liu K, Farnsworth PB, Stearns SD, et al. Instrumentation for hand-portable liquid chromatography. J Chromatogr A 2014;1327:80–9. [156] Sharma S, Tolley LT, Tolley HD, Plistil A, Stearns SD, Lee ML. Hand-portable liquid chromatographic instrumentation. J Chromatogr A 2015;1421:38–47. [157] Alpert AJ. Experience and applications of a new portable HPLC machine. Chromatogr Today 2020;13:46–8.

677

CHAPTER

Mass spectrometric detection, instrumentation, and ionization methods

26

Deyber Arley Vargas Medina, Edvaldo Vasconcelos Soares Maciel, and Fernando Mauro Lanc¸ as University of Sa˜o Paulo, Sa˜o Carlos, Institute of Chemistry of Sa˜o Carlos, Sa˜o Carlos, SP, Brazil

26.1 Introduction Since the initial developments in liquid chromatography (LC) in the 1990s, it has advanced to become one of the most essential and used separation/purification techniques in analytical chemistry [1]. Its success is due to several factors such as efficiency, reliability, robustness, and principally versatility—there are many modes of separation available, contributing to a tailored selectivity for specific target compounds [1]. LC is a suitable and widespread technique for analyzing GC-noncompatible analytes, which includes most molecules of current interest, including volatile and non-volatile compounds of low- and high-molecular-weight, they only need to be soluble in an LC-compatible solvent [2]. Also, thermal stability is not crucial in most cases, as LC separations typically occur at temperatures below 60°C [2]. For example, different classes of analytes, including pharmaceutical and veterinary drugs, cosmetics, pesticides, micropollutants, human endogenous biomolecules, can be analyzed by LC. Commercially available stationary phases and instrument configurations make it easy to analyze almost any soluble compound in different complex matrices [3]. Another remarkable feature of LC is the compatibility with several detection techniques, such as ultra-violet/visible, electrochemical, fluorescence detectors, and principally mass spectrometry (MS) [4]. However, despite this flexibility, MS has become the most important because of its versatility and remarkable sensitivity. Furthermore, considering the current challenging problems analytical chemists face where samples containing hundreds of thousands of known and unknown compounds must be investigated, it is the preferred high-throughput technique [5]. In this context, liquid chromatography coupled to mass spectrometry (LC–MS) represents an essential technique as this provides adequate selectivity, remarkable sensitivity, extensive range of compatible molecules – they only need to be ionizable. Currently, Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00016-3 Copyright # 2023 Elsevier Inc. All rights reserved.

679

680

CHAPTER 26 MS detection, instrumentation, and ionization methods

several LC–MS platforms are available, from simple and cost-effective systems to complex and costly ones, which contributes to the widespread adoption of this technique. Mass spectrometry was born in the 1890s with the experiments of Joseph Tompson and later continued by Francis W. Aston, who was awarded a Nobel Prize in chemistry for their work on this topic. Since then, many technological and theoretical advances in MS have resulted in it becoming an almost indispensable technique for many scientists [4]. Nowadays, MS is an accessible technique mainly because of compact, benchtop instruments coupled with chromatographs. Mass spectrometry analyses are based on measuring the mass-to-charge ratio (m/z) of an ionic compound—m/z is a value obtained by dividing an ion mass by its charge [6]. Therefore, it is crucial to remember that MS can only deal with ions in either negative or positive states. The results are interpreted as a function of the ion abundance versus its m/z, namely, the mass spectrum. The characteristic property of MS is its capability to discriminate compounds by m/z [4]. It is worth mentioning that the detection capability of MS is jeopardized when samples containing many interfering substances are simultaneously introduced into the mass spectrometer. So, it is not difficult to understand why LC–MS has become a mainstream analytical technique [6]. Essentially, LC separates target compounds from interfering substances in the sample before introducing it into the MS, where target compounds are ionized and further detected according to their m/z ratios. LC–MS instruments have some critical requirements to operate adequately: (i) because LC operates under high-pressure conditions while MS demands highvacuum, the volume of mobile phase introduced into the MS region must be controlled; (ii) the composition of the mobile phase—preferable it must not contain non-volatile additives like phosphate buffers, because this can cause signal suppression or damage the MS ionization source; and (iii) the nature of the matrix and the target compounds to be analyzed [7]. LC–MS can be considered a hybrid technique developed to bring together the best features of each of them in one single instrument [8]. To meet the requirements mentioned earlier on LC–MS platforms, some critical components must be overviewed to give the reader a comprehensive description of LC–MS coupling conditions and operation. This book chapter addresses these topics in the following sections covering the most relevant ionization techniques and mass analyzers.

26.2 Ionization techniques Mass spectrometers are in short ion optic systems composed of electrostatic lenses at high vacuum conditions, capable of focusing, transmitting, filtrating, and measuring ion beams. The first requirement for an LC–MS analysis is the conversion of liquid or solid neutral molecules into gas-phase ionized species, easily conveyed by an electric field. This conversion occurs in the mass spectrometer ionization source and can proceed through diverse ionization mechanisms that depend on the technology

26.2 Ionization techniques

employed. There are currently many ionization sources for LC–MS, and the more common types will be described in the following sections, introducing their advantages, limitations, and primary applications.

26.2.1 Ionization under vacuum conditions 26.2.1.1 Electron ionization (EI) Electron ionization (EI) was the first mass spectrometry ionization source developed and the first to be hyphenated with chromatography. This technique was introduced in 1918 [9], and in 1959, its coupling with capillary gas chromatography was one of the main breakthroughs in the development of the modern analytical sciences [10]. EI relies on the bombardment of gas phase molecules with a beam of high-energy electrons. EI source is contained in a vacuum chamber and consists of a tungsten filament, a compartment for ion generation, and a series of lenses for ion manipulation. Samples are introduced either in a gas phase or as a solid deposited on a heated probe [11]. A filament generates a beam of electrons that intercepts the sample following a spiral path guided by a cathode and a series of small permanent magnets. Collisions with the electron beam cause gas-phase molecules to absorb energy and ionize with loss of an electron and generate an odd-electron cation-radical (M+•). M+• are unstable species and rapidly undergo consecutive fragmentation through a series of intramolecular reactions to generate more stable species. Fragmentation can take place through diverse pathways, including (i) elimination of a radical to give an evenelectron cation; (ii) loss of neutral fragment to give another cation-radical; or (iii) molecular rearrangement, often involving migrations of hydrogens or methyl moieties, followed by fragmentation: ½M+ ! A+ + B ½M+ ! A+ + B

The cations and cations-radicals thus generated are ejected from the ionization source by applying positive potential on a repeller plate (5–10 kV) placed at the rear of the ion chamber. The ions are separated according to their m/z ratio in the mass analyzer and subsequently recorded as EI mass spectra. EI occurred under vacuum conditions, and around 1 in 104 molecules in the ionization source are ionized. Fragmentation occurs by intramolecular reactions without affecting other sample constituents. Each EI-MS spectrum is a fingerprint of each molecule, and the significant number of fragments ions recorded is a rich source of structural information for identification purposes. EI spectra obtained for given electron energy are reproducible and comparable. EI spectra are recorded at 70 eV beam electron energy in most cases. This energy provides an optimal number of fragments, with adequate intensity and robustness to create reference mass spectral libraries. As a result, there are many EI spectra libraries; in some cases, those are included in the instrument software package to

681

682

CHAPTER 26 MS detection, instrumentation, and ionization methods

annotate unknown analytes by automated spectra comparison. The potential for analyte identification is still the main advantage of EI over other ionization sources. Although EI-MS performs favorably when coupled to gas chromatography, EI-MS coupling to liquid chromatography is more intricate, and at the moment, it is still quite limited [12]. The mobile phase flow rate for convectional LC is incompatible with the vacuum conditions of the EI source. With the mobile phase entering the ionization source, most of the electrons collide with solvent molecules; analyte ionization is inefficient, can undergo chemical ionization, and generates noisy spectra with limited fragmentation. Besides, the tungsten filament quickly deteriorates under typical operating conditions. The first attempts to hyphenate LC and EI-MS dates back to the end of the 1970s but only became feasible with the introduction of the first commercial nano-LC instruments at the turn of the century. At nanoscale flowrates (nL/min), the influence of the mobile phase on the EI process is negligible, so that solvent evaporation and analyte ionization can take place in the ionization source [13]. Direct-EI was the first interface proposed for nano-LC-EI-MS interfacing. It is based on the direct insertion of separation column into the ionization source through a segment of capillary tube (Figure 26.1A) [14]. Over the last 20 years, this coupling has demonstrated suitable competence for reproducible EI-MS spectra of many LCamenable compounds [15]. Furthermore, Direct-EI is advantageous due to their minimal instrumental requirements, being the simplest and least expensive strategy for LC-EI-MS. However, the technique is insufficiently robust for routine analysis. Furthermore, the capillary tube interface is prone to clogging since the high temperature of the EI source can cause premature evaporation of the mobile phase and precipitation of analytes. To overcome these drawbacks, Cappiello and co-workers recently introduced the direct liquid electron interface (DLI) (Figure 26.1B) [16]. In this case, solvent and analyte evaporation occur in a capillary nebulization chamber, placed between the column exit and the ionization source. Evaporation is promoted by heating up to 400°C, and a helium make-up flow propels the gas-phase molecules to the EI source. Although still limited to molecular masses below 400 Da, DLI is the most efficient nano-LC-EI-MS system available.

26.2.1.2 Matrix-assisted laser desorption ionization (MALDI) MALDI is a soft ionization technique discovered by Hillenkamp and Karas in the 1980s [17], and earned Koichi Tanaka the Nobel Prize in Chemistry in 2002 for its application to the development of methods for the structural analysis of proteins [18]. Currently, MALDI is widely used for the analysis of mixtures of organic compounds, large synthetic polymers, and biomolecules, such as peptides, lipids, nucleotides, saccharides, and intact proteins. MALDI uses UV radiation to promote ionization of analytes by the simultaneous desorption/ionization of analyte molecules induced by a pulsed laser beam within a mixed crystal of the matrix and analyte. Sample solutions are dispersed in a matrix and deposited onto a sample plate (target). The solvent evaporates, and a solid sample spot remains on the target. Then, the target is inserted into the spectrometer under

26.2 Ionization techniques

FIGURE 26.1 Hyphenation of nano-LC and EI-MS. (A) Direct-EI interface for nano-LC-EI-MS coupling [63]. (B) LEI interface for nano-LC-EI-MS coupling [15].

vacuum conditions and struck by a pulsed UV laser. The matrix and sample molecules vaporize and ionize into the vacuum. Then, a high voltage propels the charged species toward the mass analyzer [a time of flight (TOF), in most cases]. Sample plates are made of polished stainless steel or conductive glass, with multiple wells where samples can be applied. The matrix is a low-mass organic molecule used to facilitate the ionization process. Commonly used matrix compounds have a conjugated pi system, like a benzene ring attached to oxygen- or nitrogen-containing groups. There are specific matrix compounds for different applications. For example, sinapic acid is frequently used for whole proteins; 3HPA (3-hydroypicolinic acid) for oligonucleotides and DNA; 2,5-dihidrxybenzoic acid (DHB) for oligosaccharides; α-cyano-4-hidroxycinnamic acid (CHCA) for peptides, small proteins, fats, and many other small molecules; and ditranol for synthetic polymers. In some cases, a combination of different matrix compounds can lead to improved ionization.

683

684

CHAPTER 26 MS detection, instrumentation, and ionization methods

There are two main classes of laser used in MALDI: (i) gas and (ii) solid-state lasers. The gas lasers are the more common. Nitrogen lasers produce UV radiation at 337 nm. They are relatively inexpensive but have a short lifetime and a low repetition rate (1–60 laser shots per second), and their intensity varies significantly between pulses. Solid-state lasers produce UV radiation around 350 nm, are more expensive, but have a more extensive lifetime and better power stability, being more suitable for quantitative determinations and mass spectrometry imaging. The energy per pulse of solid-state lasers is relatively high, with a pulse length of 1–2 ns or pulse energies 1–10 μJ. When the laser pulse hits the sample spot, matrix molecules absorb the laser energy, and the analytes ablate from the surface, being transferred into the gas phase. During the ablation process, analyte molecules are ionized, usually via protons transference reactions (in the positive ion mode) with nearby matrix molecules, generating a protonated quasimolecular ion. Adduct ions can form if there is a source of sodium or potassium in the sample. In the negative ion mode, ionization takes place via deprotonation of analytes. MALDI is a gentle ionization process and usually does not result in significant analyte fragmentation or decomposition. The main hurdle for coupling MALDI with LC is the lack of robust automated interfaces. Commonly, the LC eluent is split (for nano-LC split is not required) and collected onto a MALDI sample plate at timed intervals. This approach allows spotting and real-time mapping of each LC peak. Each peak is divided over several spots allowing a detailed analysis. Besides, the MALDI plate can be re-analyzed many times, for the exploration of different experimental parameters and to obtain additional information.

26.2.2 Atmospheric pressure ionization (API) 26.2.2.1 Electrospray ionization (ESI) Electrospray ionization (ESI) is the most popular technique for the transference of ions from the liquid phase to MS vacuum conditions. The development of ESI made LC–MS feasible with most MS configurations and expanded the boundaries of the technique to cover analytes of small mass through to large molecules (>100,000 Da). The development of this interface earned John Fenn the Nobel Prize in Chemistry in 2002 [19,20], recognizing his contribution to the MS of large polar molecules. ESI is a soft ionization technique that relies on the formation and evaporation of ions from nanometric charged droplets. The LC eluent is infused at a 1–1000 μL/min flow rate into a metallic capillary needle of 100 μm i.d. Initially, a rounded drop of solvent forms at the tip, and by applying a high voltage, ions in the liquid phase migrate to the drop’s surface, accumulating in the region of the larger electric field. When the potential is sufficient (e.g., 3–6 kV), the electric forces equal the surface tension, pulling the meniscus into a conical shape (Taylor cone). The strong electric field at the tip of the cone ejects a microscope jet of liquid formed of fine-charged droplets that rapidly evaporate, releasing gas-phase ionized analytes. The electric field between the ESI tip (emitter) and the sampling cone of the mass spectrometer

26.2 Ionization techniques

Liquid inlet

ESI emitter (1.5 kV)

Solvent evaporation

Nebulization gas (N2)

Taylor cone

Aerosol plume

Sheath gas (N2)

Coulomb fission

a.

Gas phase ion

FIGURE 26.2 ESI source. (A) Schematic representation of the main parts of the ESI and the ion formation process. (B) Picture of an ESI source.

directs the ions to the MS analyzer. Besides, a coaxial flow of drying gas (N2) to assist the spray process, improving nebulization, and a sheath gas (N2) aids the electric field in focusing the ions toward the sampling cone (Figure 26.2). Unlike EI, ESI proceeds practically without fragmentation. Analyte ions form mainly from the solution by acid/basic reactions by tuning of the mobile phase composition either before or after the separation. In addition, ESI acts as an electrochemical cell with analytes ionized by oxidation–reduction reactions. Therefore, ESI spectra are composed mainly of quasimolecular ions, either [MH]+, [MH], or as adducts formed with solvent impurities or additives. In the positive ion mode, although other ions such as NH+4 , Na+, and K+ may be present in the solution, protonation reactions are the main contributor to the charge of the sprayed droplets. Ions H+ may come from acidification of the mobile phase and charge-balancing reactions (e.g., 2H2O ! 4H+ +4e + O2) at the metal/solution interface. Similar considerations apply to the formation of negatively charged species. Depending on the nature of the analyte and analysis conditions, the applied potential can be either negative or positive, enabling the detection of anion or cations, respectively. In addition, some instruments allow rapid switching the field’s polarity, facilitating the simultaneous recording of m/z data in positive and negative modes. The transfer of ions to the gas phase proceeds by successive solvent evaporation/ Coulomb-fission steps of the droplets emitted from the Taylor cone. The surface tension of the solvent retains the ions within the droplet while Coulombic repulsions between like charges disrupt the droplet surface. This process continues until the Rayleigh limit is reached. The Rayleigh limit estimates the number of ions able

685

686

CHAPTER 26 MS detection, instrumentation, and ionization methods

to co-exist in a charged droplet before fission occurs and is given by the expression (with R < 10 nm): ZR ¼

8π √ε0 γR3 e

where ZR is the number of elementary charges in a stable drop, R is the droplet radius, ε0 is the vacuum permittivity, and γ is the surface tension of the solvent. At the Rayleigh limit, the surface tension can no longer sustain the Coulomb force of repulsion, and a “Coulomb fission” occurs, generating new smaller and highly charged droplets. Repeated evaporation/fission events of nanometer droplets ultimately yield the gas phase ionized analytes. Ions releasing from the nanometer drops can be understood through two main proposed mechanisms: (i) the ion evaporation model (IEM) for low molecular mass analytes, and (ii) the charged residue model (CRM) for large species [21]. In the ion evaporation mechanism (Figure 26.3A), ions are expelled from the surface by the electric field originating from a Rayleigh-charged nanodroplet. The ejected ion initially remains connected to the droplet by a “bridge” of solvent molecules, rupturing as a solvated cluster composed of the ion and a few solvent molecules. Remanent solvent rapidly evaporates by collision with the surrounded make-up gas, and gas-phase ions release. In the CRM, the analyte ions are not ejected from Rayleigh-charged nanodroplets. Instead, free ions result from the solvent’s complete evaporation, with the final charge transfer from the solvent to the analyte (Figure 26.3B). ESI performance is susceptible to matrix and mobile phase composition effects. In addition, non-volatile compounds can compete with analytes for the available

FIGURE 26.3 Schematic representation of the ion formation mechanism in ESI. (A) Ion evaporation model (IEM); (B) charged residue model (CRM) [21].

26.2 Ionization techniques

charges or space on the droplet surface, affecting the droplet formation/evaporation/ fission process and leading to enhancement or suppression of analyte ionization. In general, the negative ion mode is more selective and less susceptible to ionization enhancement/suppression effects than the positive ion mode. Finally, the efficiency of the ESI process depends on instrumental parameters, such as the magnitude and polarity of the applied potential, gases flow rate, and source temperature, among others, and adequate sample preparation, the proper choice of the mobile phase additives, and the quality of the chromatographic separation. ESI performs well at a conventional flow scale. However, sensitivity improves for nanoflow rates, and the matrix effects decrease. Nano-LC-ESI-MS works at nanoliter/min flow rates and uses emitters with an inner diameter of around 20 μm so that the spray is composed of smaller droplets than those observed at conventional flow rates. Smaller droplets evaporate faster, requiring a minor number of evaporation/fission cycles leading to the more efficient transfer of ions into the gas phase [11].

26.2.2.2 Atmospheric pressure chemical ionization APCI APCI is a soft ionization technique complementary to ESI, less susceptible to matrix effects, and suitable for analyzing nonpolar compounds. In some instruments, APCI and ESI co-exist in the same interface, and it is sufficient to change the spray probe to use one or another ionization mode. There are also multimode sources that can perform simultaneous ESI and APCI, allowing the analysis of compounds of a broad polarity range in a single run. APCI ions are formed directly in the gas phase, in contrast to ESI. The APCI probe consists of a spray needle with a pneumatic nebulizer, where the LC effluent vaporizes rapidly with the aid of a high-speed gas stream (N2) at 400–550°C. Solvent and analyte molecules in the aerosol undergo APCI by applying a corona discharge (1–5 μA) through an electrode connected to the end of the nebulization tube. The discharge creates plasma rich in electrons and reactant ions derived from the ionization of the mobile phase. First, the vaporized solvent molecules ionize to produce the reactant ions [X+H]+ in the positive ion and [XH] in the negative ion modes. Subsequently, the gaseous analyte is ionized by reaction with these reactant ions. For example, in the positive ionization mode, N2 molecules are excited and ionized by energetic electrons: N2 + e ! N2 + + 2 e

If only water molecules are present, the N+• 2 commences a sequence of reactions that lead to the formation of protonated water clusters: N2 + + H2 O ! H2 O+ + 2N2 H2 O+ + H2 O ! H3O+ + HO H3O+ + nH2 O ! H3O+ :ðH2 OÞn

687

688

CHAPTER 26 MS detection, instrumentation, and ionization methods

The protonated water clusters transfer their proton to methanol, acetonitrile, or other mobile phase constituent with a higher proton affinity. This new sequence of ionmolecule reactions results in mixed solvent clusters. Finally, protons are transferred from the protonated solvent clusters to the analyte molecules. Compounds with a high proton affinity will ionize more efficiently. A given analyte only will be protonated if it has a higher proton affinity than any other component in the mobile phase, i.e., when ammonium acetate is added to the mobile phase, only compounds with a proton affinity exceeding 854 kJ/mol (proton affinity of ammonium acetate) will be protonated. In the negative-ion mode, the superoxide O 2 promotes deprotonation of the mobile phase triggering the sequence of ionization reactions. Their relative gas-phase acidity determines the deprotonation of the molecules. The dependence on the proton affinity or relative gas-phase acidity makes APCI a more selective and robust ionization technique less susceptible to matrix effects than ESI. APCI mass spectra are characterized by quasi molecular ions. However, due to the high temperatures employed, APCI can occur with thermal breakdown, and some ions derived from thermal degradation products may be observed. For example, APCI-MS analysis of nitro aromatic compounds contain a fragment ion due to the loss of 30 Da, and H/D exchange studies demonstrated that this ion is a product of the thermally induced reduction of the NO2 group to an NH2 group and not a fragment ion [22].

26.2.2.3 Atmospheric pressure photoionization (APPI) APPI is a complementary technique for ionizing nonpolar compounds that barely ionize by ESI or APCI, such as polycyclic aromatic hydrocarbons, steroids, lipids, vitamins, some flavonoids, and some drugs and their metabolites. Being a more selective ionization technique, APPI is less susceptible to matrix and mobile phase composition effects than ESI or APCI. APPI is a modified version of APCI, in which a photon-emitting lamp induces ionization. Like in APCI, the LC eluent is initially vaporized by a heated pneumatic nebulizer and subsequently submitted to the radiation of a discharge lamp. A series of gas-phase reactions are triggered, ending with the analyte’s ionization. The selectivity of APPI depends on the ionization energy (IE) of the analytes, which should be lower than the photon energy emitted by the lamp. Of the different discharge lamps available, the krypton lamp is the most frequently used providing of 10.0 and 10.6 eV [21]. Direct APPI occurs when the IE of the analytes is lower than the lamp emitting photons. In this case, one of the electrons in the analyte molecule is removed by absorption of the ionizing photon (hν), forming the radical cation (M+•), that subsequently can react with a protic solvent (S), removing a proton and forming the quasimolecular ion [M+H]+: M + hν ! M+ + e M+ + S ! ½M + H+ + S½H

26.3 Mass analyzers

When the IE of the analytes is larger than the photon energy of the lamp, indirect ionization strategies are necessary. In this case, a solvent with lower IE than the photon source, such as acetone or toluene, is added as a dopant (D). Analytes are then ionized via transference of charge or proton by reaction with a cation-radical formed as a product of the photoionization of the dopant (D+•). D + hv ! D+ + e D+ + M ! D + M+ D+ + S ! ½D  H + ½S + H+ ½S + H+ + M ! S + ½M + H+

26.2.3 Ambient ionization mass spectrometry Although not coupled to LC, mention should be made to a family of techniques for the direct analysis of native samples, currently known as ambient desorption/ionization mass spectrometry (AIMS). In AIMS, desorption and ionization of analytes occur at an open-air surface under ambient conditions. No previous chromatographic separation is required, and the stages of sample preparation, analyte ionization, and transfer of free ions to the gas-phase occur almost simultaneously. Ions are produced directly from the sample via an ESI, APCI, or MALDI-like ionization mechanism and in the inlet of the mass spectrometer. AIMS techniques arose in 2004 with the introduction of solvent-based desorption electrospray ionization (DESI) [23] and the plasma-based direct analysis in real-time (DART) [24]. First, DESI is performed, generating an electrospray directly over the sample to be analyzed. Then, the same spray acts as the extractant phase, and its impact on the sample produces analytes ions in the gas phase (Figure 26.4A). DART works similarly, but in this case, a jet of electronically or vibrationally excited gas (He, Ar, or N2) is used to desorb/ionize the analytes from the sample (Figure 26.4B). In addition, new ionizations strategies, such as the use of low-temperature plasma ionization sources (LTP) [25], alternative sample/spray probes, such as the paper spray (Figure 26.4C) [26], and the coated blade spray (Figure 26.4D) [27] are beginning to be explored [28].

26.3 Mass analyzers The mass analyzer is the system component responsible for separating the ions formed in the ionization source according to their m/z. Although mass analyzer is a general term for the ion separation device, there are different types available, with varied compatibility with the ionization sources [29]. Each mass analyzer has a different m/z working range, mass accuracy, and especially mass resolution. The MS

689

690

CHAPTER 26 MS detection, instrumentation, and ionization methods

FIGURE 26.4 Some examples of ambient desorption/ionization techniques. (A) Desorption electrospray ionization (DESI) [64]; (B) plasma-based direct analysis in real time (DART) [65]; (C) paper spray [66]; (D) coated blade spray [27].

resolution is the main parameter used to define whether a mass analyzer is considered just satisfactory or high-resolution (HRMS) [29].

26.3.1 Low-resolution mass analyzers 26.3.1.1 Quadrupole So far, the quadrupole mass analyzer remains one of the most popular types for several reasons: (i) it is robust and reliable in terms of reproducibility and accuracy; (ii) it is compact, becoming easier to “box” in benchtop instruments; (iii) is a cost-effective component (low-cost of production and good durability) [30]. These features give quadrupole mass analyzers broad applicability for target compounds up to a mass of 2000–4000 m/z [30]. Indeed, LC–MS systems using quadrupoles are currently considered vital instruments for the pharmaceutical and related industries. Furthermore, when high sensitivity is required, quadrupoles are frequently combined with other mass analyzers like other quadrupoles or time-offlight analyzers [31]. Tandem and hybrid mass analyzers open up a new branch of MS operating modes (e.g., multiple reaction monitoring) and will be discussed in Section 26.3.3. By definition, a quadrupole is considered a dynamic filter composed of a cluster of four hyperbolic or cylindrical charged rods [31]. The oppositely charged rods are arranged in a parallel array and connected electrically, creating a central gap where the ions pass through. First, a direct current (DC) is applied for charging the rods, and then a radio-frequency alternating voltage (RF) is superimposed on the DC voltage

26.3 Mass analyzers

[32]. This process creates a dynamic quadrupole field in the gap, where an electrical potential pushes the ions. The DC vs. RF potential combination is the primary factor responsible for filtering the ions, which will pass to the detector or be filtered out [32]. Therefore, according to the DC and RF potential applied simultaneously, ions of different m/z can be isolated and detected [32]. In addition to their use as a mass analyzer, quadrupoles can guide ions within the mass spectrometer, transmitting them from the ionization source to the mass analyzer. In this case, the quadrupole is set to work in the “RF-only” mode, where DC is reduced by nearly zero while the RF is increased, allowing all ions to pass steadily through the quadrupole [33]. Noteworthy, MS collision cells are quadrupoles operating in the “RF-only” mode while an inert collision gas (e.g., Argon) is introduced. This action causes collision of the ions with the inert gas, causing collision-induced dissociation (CID) [33]. Figure 26.5 shows a schematic illustration of a simple quadrupole for a better understanding. Quadrupoles can be considered a suitable choice for many routine applications because of their affordable cost, compact dimensions, robustness, and reliability (low needed for maintenance) [34]. Furthermore, quadrupoles present excellent performance when coupled with liquid chromatography and ESI sources, particularly for small molecules analysis. On the other hand, they possess a limited working mass range and poor resolution, representing a downside for complex samples, biomolecules analysis, or multi-residue methods containing a broad range of compounds with comparable mass [34].

FIGURE 26.5 Schematic illustration of a quadrupole mass analyzer: A specific combination of RF and DC allows selected ions to pass through the quadrupole toward the detector. Reprinted from Somogyi [67] with the permission of Elsevier.

691

692

CHAPTER 26 MS detection, instrumentation, and ionization methods

26.3.1.2 Ion trap (IT) Ion trap (IT) is a variation of the quadrupole mass analyzer designed to provide better analytical performance [35]. The first developments on ion trap date back from the 1950s with pioneering works of Paul and Steinwedel, and a little later than the emerging of quadrupole mass analyzer [36]. However, ion trap coupled with liquid chromatography only became possible after the development of the ESI interfaces in the 1980s, which led to new ion trap MS instruments for the analysis of non-volatile, thermally unstable, and polar compounds. Ion trap mass analyzers do not behave like a mass filter; instead, they work by capturing and confining the target ions for subsequent mass analysis [37]. This capacity to confine the ions and accumulate them into a specific region sometimes improves sensitivity [35]. Unlike a quadrupole, with its rods disposed in a horizontal and parallel arrangement, ion traps have two different forms: a loop-folded structure (3D) and, more recently, the linear ion trap (LIT or 2D) [35]. Concerning 3D ion traps, a central ring-like electrode is responsible for applying the RF potential while two other hyperbolic cap electrodes located a fixed distance from it. The target ions are confined in the space between the cap electrodes, and the combination of the applied potentials, also called trapping field (DC and RF), determines which ions will be trapped in a characteristic circular orbit [38]. Inside the ion trap, target ions are sequentially ejected one after another as a function of the trapping field according to their m/z ratio [38]. Variation of the trapping field causes the circular flight of each ion to become unstable, prompting their ejection toward the detector. On the other hand, linear ion traps (LITs) employ a bi-dimensionally RF potential at its extremities, assisted by a quadrupole. In this case, ions are confined radially by characteristic trapping fields applied to the quadrupoles, and the target ions ejected radially or axially depending on the LIT design [35]. Figure 26.6 provides a general illustration of the steps occurring in an ion trap mass analyzer. Since LITs use a quadrupole as the main component, they can also be mass filters. This ability is often exploited in Orbitrap instruments. Compared to 3D ion traps, LITs possess higher ion confinement capacity and sensitivity, a fast scan rate, and a more straightforward design [36]. Because of their compact size/shape, all ion traps represent an exciting class of mass analyzers for coupling with liquid chromatography benchtop instruments. In most cases, ion traps are used for qualitative purposes, especially for “omics” studies using low-resolution with high acquisition rates in the full scan or data-dependent mode. Although ion traps can be employed to analyze small molecules, quadrupoles are the gold standard analyzer for such applications. It is worth mentioning that when more resolving power is needed, for example, in non-target compound analysis, a high-resolution mass analyzer (HRMS) is preferable (e.g., TOF, FT-MS, or Orbitrap).

26.3.2 High-resolution mass analyzers In mass spectrometry, resolution refers to the ability of the analyzer to resolve ion beams with similar m/z ratios. Higher resolution provides better separation between two isobaric ions. For low-resolution analyzers, such as single quadrupole

26.3 Mass analyzers

FIGURE 26.6 Common steps in which target ions pass through when an ion trap mass analyzer is employed. (A) Selected ions are retained in the trap; (B) The specified ions are fragmented; (C) Product ions are ejected from the trap; (D) One of the product ions may be trapped, excited, and fragmented. Reprinted from Hocart [41] with the permission of Elsevier.

or ion trap, the resolution is defined as the entire width of the peak at half its maximum height (FWHM). This value is set to around 0.7 a.m.u. and is independent of the measured m/z. For HRMS instruments, IUPAC defined two main approaches to measure resolution: (i) in the presence of two overlapping peaks of approximately equal height, the resolution is the ratio between the higher mass and the difference in masses between the peaks (ΔM) (Figure 26.7A); (ii) if only a single peak is present, the resolution is expressed as the ratio between the nominal mass of the peak (M) and the 5 or 50% of the peak height (FWHM) (Figure 26.7B). Another term commonly used to express the capability of an instrument or method to resolve ions differing in m/z by a small increment is the Resolving power, which is defined as the peak width (Δm) in mass units for a given m/z ratio.

693

694

CHAPTER 26 MS detection, instrumentation, and ionization methods

FIGURE 26.7 Definition of resolution in mass spectrometry. (A) Resolution calculated from two overlapping adjacent m/z peaks; (B) resolution calculated from a single m/z peak using full-width at half-maximum (FWHM) [68].

The higher the analyzer’s resolution, the higher the absolute mass accuracy. Typically expressed in Δppm, mass accuracy is an estimation of the difference between the experimentally measured m/z ratio (accurate mass) and the calculated m/z of that ion from the isotopic masses of their atom constituents (monoisotopic molecular mass or accurate mass): Δppm ¼

mmesuared  mmonoisotopic 6 ∗10 mmonoisotopic

Monoisotopic molecular masses are calculated from the molecular formula of the measured ion considering the mass of each atom present in the structure. For example, for the quasimolecular ion of the molecule HhCcNnOo in the positive ion mode, the monoisotopic mass is calculated as presented in Table 26.1.

Table 26.1 Example of the calculation of monoisotopic molecular mass for a compound with molecular formula HhCcNnOo, ionized in the positive mode. Atom

Mass of the atom

# Atoms

Sum

Hydrogen Carbon Nitrogen Oxygen Plus H Minus e Total

1.00783 12.00000 14.00307 15.99492 1.00783 0.00055

h c n o

1.00783*h 12.00000*c 14.00307*n 15.99492*o 1.00783 0.00055 P ( )

26.3 Mass analyzers

HRMS allows the identification of compounds with mass errors normally lower than 3 mDa. Typically, values of Δppm 0) for the 2-PIC and 1-AA columns, and Van der Waals interactions (s > 0) for the BEH and C18 columns. In another study, thin layer chromatography was used to separate 16 β-blockers [38]. In addition, the influence of various solvents on the retention of β-blockers was investigated. As a result of the obtained data, a QSRR model was developed and was used to explain the retention mechanism of betablockers on the silica gel stationary phase.

30.5 3S—Similarity, selectivity, and specificity An important aspect in the QSRR methodology is the concept of similarity, selectivity, and specificity (3S). The term similarity refers to the search for such chromatographic stationary phases that will be structurally similar to the analyzed substances. By using such chromatographic materials, it is possible to obtain a higher selectivity of the chromatographic system, which can be further improved by manipulation of the composition of the mobile phase. This, in turn, contributes to the observation of higher separation specificity, thanks to the increased selectivity moderated by chromatographic conditions [42]. An example of 3S is lipidomic analysis with the use of stationary phases imitating biological membranes, the so-called stationary phases of phospholipid. Both the stationary phase and the analyzed compounds (phospholipids) have phosphate groups and alkyl chains in their structure. Another example of 3S is amino acid analysis on peptide stationary phases. As a result of this analysis, it was possible to study the influence of the structural similarities between the analyzed molecules and chemically bonded ligands located on the surface of the stationary phase. It should be mentioned that the retention mechanism of amino acid on peptide stationary phases is a complex phenomenon and includes a range of interactions between analytes and chemically bonded ligands of amino acids and peptides (Figure 30.4). Chemical similarity plays an important role in predicting the properties of chemical compounds, especially in designing new chemicals and drugs. A measure that is used frequently to compare the similarity of chemical structures is the Tanimoto

30.6 Characterization of stationary phases

FIGURE 30.4 Possible mechanism retention of amino acids.

coefficient [43,44]. Tanimoto coefficient (similarity index) is very important in QSRR and QSAR studies and can be calculated using the following equation: SA,B ¼

c a+bc

(30.6)

where a, b—bit sets in fingerprint for A and B molecules, c—bit set in common between two fingerprints. The similarity index is mainly used as a filter to select compounds that are structurally similar to the target compounds in order to create a training set for use in constructing QSRR models.

30.6 Characterization of stationary phases There are many reports in the literature on the retention properties of stationary phases [7,45,46]. The information contained in individual scientific articles explains the molecular retention mechanism of the chromatographic system under study. In studies using molecular modeling, a series of analytes has been developed for different types of QSRR approaches, using models involving log P, reduced LSER, and molecular descriptors. For retention data, the most frequently used parameter is the logarithm of the retention factor which is correlated with the theoretically calculated log of the n-octanol/water partition coefficient (log P). This approach uses regression of log k against the calculated descriptors. Table 30.2 shows the QSRR equations for amino acids

809

Table 30.2 QSRR equation for the tested stationary phases. Equation

k1

k2

k3

k4

R2

s

F

0.2994 (0.1049) 0.5456 (0.1536)

0.1979 (0.0387) 0.0055 (0.0012)

0.9812 (0.2585) 0.1655 (0.0320)



0.6916

0.0135

14

0.0638 (0.0157)

0.7726

0.0109

12

0.3699 (0.1087) 0.6998 (0.1371)

0.1841 (0.0343) 0.0044 (0.0011)

0.8673 (0.2424) 0.1336 (0.0225)



0.7115

0.0136

15

0.0457 (0.0149)

0.7756

0.0115

13

0.7861 (0.2235) 0.9903 (0.2327)

0.1826 (0.0468) 0.0056 (0.0016)

0.0120 (0.0045) 0.2003 (0.0451)



0.5591

0.0253

8

0.0020 (0.0009)

0.6624

0.0211

7

– 0.7397 (0.2165

– 0.0056 (0.0017)

– 0.1050 (0.0363)

– 0.0746 (0.0235)

– 0.6286

– 0.0240

– 6

0.7653 (0.1456)

0.0028 (0.0010)

0.0837 (0.0265



0.5135

0.0126

6

Amino-Gly1 log k ¼ k1 + k2 log D + k3 δMin log k ¼ k1 + k2 VWS + k3 HBA + k4 Pch Amino-Gly3 log k ¼ k1 + k2 log D + k3 δMin log k ¼ k1 + k2 VWS + k3 HBD + k4 Pch Amino-Asp1 log k ¼ k1 + k2 HBD + k3 TE log k ¼ k1 + k2 VWS + k3 HBD + k4 HF Amino-Ala2 – log k ¼ k1 + k2 VWS + k3 HBD + k4 Pch Amino-[Fmoc-Tyr(tBu)] log k ¼ k1 + k2 VWS + k3 HBD

log k ¼ k1 + k2 VWS + k3 TE + k4 Pch

0.7770 (0.1361)

0.0100 (0.0020)

0.0187 (0.0045)

0.0428 (0.0151)

0.6869

0.0089

8

0.5946 (0.1698) 0.7613 (0.3304)

0.0037 (0.0012) 0.0033 (0.0013)

0.1140 (0.0309) 0.0299 (0.0112)



0.5887

0.0174

9

0.0761 (0.0296)

0.5356

0.0214

4

0.8052 (0.2170) 0.8754 (0.3936)

0.1820 (0.0454) 0.0037 (0.0015)

0.0144 (0.0044) 0.0341 (0.0133)



0.5805

0.0238

8

0.0842 (0.0353)

0.5100

0.0304

4

– 0.7438 (0.2209)

– 0.1341 (0.0454)

– 0.0141 (0.0047)

– 0.0525 (0.0202)

– 0.5637

– 0.0236

– 5

0.4187 (0.1167) 0.4135 (0.1225)

0.1893 (0.0430) 0.0993 (0.0314)

1.0744 (0.2875) 0.2901 (0.0946)



0.6376

0.0167

11

0.24442 (0.0819)

0.6066

0.0198

6

0.3563 (0.1529)

0.2168 (0.0604)

1.2061 (0.3647)



0.5798

0.0244

7

Amino-Met1 log k ¼ k1 + k2 VWS + k3 HBD log k ¼ k1 + k2 SAS + k3 HE + k4 Pch Amino-Met2 log k ¼ k1 + k2 HBD + k3 TE log k ¼ k1 + k2 SAS + k3 HE + k4 Pch Amino-Leu1 – log k ¼ k1 + k2 HBD + k3 TE + k4 Pch Amino-Leu2 log k ¼ k1 + k2 log D + k3 δMin log k ¼ k1 + k2 HBD + k3 μ + k4 Pch Amino-Leu3 log k ¼ k1 + k2 log D + k3 δMin

Continued

Table 30.2 QSRR equation for the tested stationary phases—cont’d Equation

k1

k2

k3

k4

R2

s

F

log k ¼ k1 + k2 VWS + k3 HBD + k4 P

1.0661 (0.2369)

0.0048 (0.0014)

0.2453 (0.0648)

0.0083 (0.0034)

0.6340

0.0215

6

0.3024 (0.1386) 0.7242 (0.2423)

0.2298 (0.0548) 0.1249 (0.0498)

1.4525 (0.3308) 0.0137 (0.0052)



0.6496

0.0183

11

0.0636 (0.0222)

0.5462

0.0283

4

– 0.9359 (0.2774)

– 0.0038 (0.0016)

– 0.2744 (0.0759)

– 0.0119 (0.0040)

– 0.5482

– 0.0294

– 4

0.6205 (0.1647) 1.0014 (0.2517)

0.0102 (0.0021) 0.0063 (0.0019)

0.0239 (0.0053) 0.0069 (0.0028)



0.6727

0.0142

12

0.1026 (0.0438)

0.5927

0.0192

5

0.6693 (0.1675) 0.4191 (0.1764)

0.0046 (0.0011) 0.0028 (0.0012)

140.1350 (0.0305) 0.0817 (0.0342)



0.6814

0.0169

13

0.0016 (0.0007)

0.7903

0.0121

14

Amino-Phe1 log k ¼ k1 + k2 log D + k3 δMin log k ¼ k1 + k2 HBD + k3 TE + k4 Pch Amino-Phe2 – log k ¼ k1 + k2 VWS + k3 HBD + k4 P TSKgel–NH2 log k ¼ k1 + k2 VWS + k3 TE log k ¼ k1 + k2 SAS + k3 VWS + k4 log D Silica-amino log k ¼ k1 + k2 VWS + k3 HBD log k ¼ k1 + k2 VWS + k3 HBD + k4 HF

Regression coefficients (k1, k2, k3, k4) with standard deviations, the standard error of the estimate (s), the value of the Fisher test (F), the square of the multiple correlation coefficient (R2), and the level of significance of equations P < .05 for the investigated stationary phases.

30.6 Characterization of stationary phases

FIGURE 30.5 Structure of stationary phases. A—amino-(Leu)1, B—amino-(Leu)2, C—amino-(Leu)3, D—amino-(Gly)1, E—amino-(Gly)3, F—amino-(Met)1, G—amino-(Met)2, H—amino-(Phe)1, I—amino-(Phe)2, J—amino-(Asp)1, K—amino-(Ala)2, L—silica-amino, M—TSKgel-NH2, N—amino-[Fmoc-Tyr(tBu)] [34].

separated on peptide reversed-phase stationary phases [34] (Figure 30.5). Acetonitrile and water were used as the mobile phase. When analyzing Table 30.2, it can be noticed that for phases amino-(Gly)3, amino-(Leu)2, amino-(Leu)3, and amino-(Phe)1, the greatest influence on the retention mechanism is exerted by an excess of electrons on a single atom in the amino acid molecule (δMin). This parameter is related to ion–ion electrostatic interactions between amino acids and stationary phases. Moreover, the presence of the log D descriptor together with δMin represents a specific relationship, because log D is a measure of the polarity of the analytes (lower value of log D indicates greater hydrophilic character). This may mean that for the amino-(Gly)1, amino(Gly)3, amino-(Leu)2, amino-(Leu)3, and amino-(Phe)1 stationary phases, there are both hydrophilic and ion-ionic interactions. Another descriptor that influences the retention mechanism of amino acids is the number of hydrogen bond donors (HBDs) and hydrogen bond acceptors (HBAs) for the amino-(Gly)1 column. Peptide stationary phases contain acceptors and donors of hydrogen bonding centers in their structure. Therefore, it can be assumed that hydrogen bonds play an important role in the retention mechanism of amino acids on the studied surfaces of the stationary phases.

813

814

CHAPTER 30 Prediction of retention in liquid chromatography

Another predictor of the QSRR equations is VWS. This parameter relates to the ability of solute molecules to interact with chromatographic components through hydrophobic forces. For the amino-(Met)1, amino-(Met)2, amino-(Leu)1, amino(Leu)2, and amino-(Phe)1 columns, VWS contributes negative input to retention. This means that non-specific intermolecular interactions are stronger between amino acids and mobile phase molecules than those between amino acids and a chemically bonded ligand on the stationary phase. Another descriptor that plays an important role in the retention mechanism is total energy (TE). TE is proportional to retention, which means that the higher the sum of the electron energies and the intranuclear repulsion energies, the higher the retention.

30.7 Quantitative retention biological activity relationships Quantitative structure-activity relationship (QSAR) modeling is one of the most important computational tools in medical chemistry. Classic QSAR studies cover issues related to ligands and their binding sites, inhibition constants, rate constants, and other biological endpoints, as well as lipophilicity, polarizability, electronic and spatial properties and specific structural features. QSAR models are used widely to assess the potential impact of biologically active chemicals on human health and ecological systems. The basis of QSAR is the assumption of dependency between differences in biological activity of a group of compounds and selected parameters describing changes in their structure. Based on statistically significant correlations, a model is built that can be used to predict the properties (activity) of any derivative compounds with a defined chemical structure. QSAR is based on the principle that molecules with a similar structure behave similarly under the same environmental conditions, and differences in the activity of compounds result from differences in the structure of the molecule [47]. With the QSAR study, the biological effect of biologically untested compounds can be predicted based on their physicochemical properties or other structural features. Additionally, it is possible to indicate which of the chemical properties regulate a particular type of pharmacological effect. QSARs can also provide guidance on how to modify structural properties to obtain the desired pharmacological response [48]. QSAR and QSRR analyses can contribute to the prediction of the properties of individual derivatives and the design of new compounds with better activity and targeted pharmacological action. Figure 30.6 shows the relationship between QSAR and QSRR. These are currently the most promising techniques for the study of new compounds [49].

30.8 Conclusions and future perspectives Undoubtedly, the use of QSRR allows for a reduction in the time required to develop new methods of chromatographic analysis; however, this technique is not able to completely replace the experimental approach to method development. Moreover,

30.8 Conclusions and future perspectives

FIGURE 30.6 Theoretical approach for determination of bioactivity. (A) An example of 3D-QSAR molecular contour map for flavonoids. Steric field contours are in green, yellow; electrostatic fields are in blue and red; (B) representation of the graph for determination of drug binding percentage; (C) specific example of the connection between QSAR and QSRR techniques [49].

the accuracy of the retention data prediction can be improved significantly if a small number of experiments are performed in conjunction with the use of statistical methods. This hybrid approach has great potential for method development and typically involves QSRR to find the approximate chromatographic conditions to achieve the desired separation, followed by a structured experimental optimization to determine the precise chromatographic conditions. In addition, the use of new stationary phases (mimicking biological membranes) also extends the use of QSRR, especially in the determination of drugs. This combination (theory and practice) ensures a high level of accuracy. Currently, QSRR is a useful technique for predicting chemical retention, but there is a pressing need for more extensive retention databases on a wider range of analytes and chromatographic conditions and techniques to strengthen the retention prediction models. The use of advanced instrumental techniques with a package of statistical and chemometric tools gives great opportunities in predicting the retention of biologically active compounds in the area of QSRR research. The improvement of cheminformatics methods allows for better development of new molecular modeling strategies. The next milestone in QSRR research will be the introduction of new strategies to improve compound retention prediction methods, to assess molecular similarity, and to synthesize new stationary phases mimicking biological membranes.

815

816

CHAPTER 30 Prediction of retention in liquid chromatography

Acknowledgment Bogusław Buszewski and Justyna Walczak-Skierska are members of Toru n Center of Excellence “Towards Personalized Medicine” operating under Excellence Initiative-Research University. Paul Haddad is a member of the Australian Centre for Research on Separation Science at the University of Tasmania.

References [1] Kaliszan R. Quantitative structure-retention relationships. Anal Chem 1992;64 (11):619A–31A. https://doi.org/10.1021/ac00035a001. [2] Kaliszan R, Foks H. The relationship between the Rm values and the connectivity indices for pyrazine carbothioamide derivatives. Chromatographia 1977;10(10):346–9. https:// doi.org/10.1007/BF02274482. [3] https://www.scopus.com/results/results. Data available 07.10.2021. [4] Kaliszan R. QSRR: quantitative structure-(chromatographic) retention relationships. Chem Rev 2007;107:3212–46. https://doi.org/10.1021/cr068412z. [5] Kaliszan R. Quantitative structure-chromatographic retention relationships. New York: John Wiley & Sons; 1987. [6] Taraji M, Haddad PR. Method optimisation in hydrophilic-interaction liquid chromatography by design of experiments combined with quantitative structure–retention relationships. Aust J Chem 2021. https://doi.org/10.1071/CH21102. [7] Buszewski B, Gadzała-Kopciuch RM, Markuszewski M, Kaliszan R. Chemically bonded silica stationary phases: synthesis, physicochemical characterization, and molecular mechanism of reversed-phase HPLC retention. Anal Chem 1997;69:3277–84. https:// doi.org/10.1021/ac9612032. [8] Al-Haj MA, Kaliszan R, Buszewski B. Quantitative structure-retention relationships with model analytes as a means of an objective evaluation of chromatographic columns. J Chromatogr Sci 2001;39:29–38. https://doi.org/10.1093/chromsci/39.1.29. [9] Kaliszan R, Ba˛czek T, Bucinski A, Buszewski B, Sztupecka M. Prediction of gradient retention from the linear solvent strength (LSS) model, quantitative structure-retention relationships (QSRR), and artificial neural networks (ANN). J Sep Sci 2003;26: 271–82. https://doi.org/10.1002/jssc.200390033. [10] Kaliszan R. Chromatography and capillary electrophoresis in modelling the basic processes of drug action. Trends Analyt Chem 1999;18:400–10. https://doi.org/10.1016/ S0165-9936(99)00120-X. [11] Sadek PC, Carr PW, Doherty RM, Kamlet MJ, Taft RW, Abraham MH. Study of retention processes in reversed-phase high-performance liquid chromatography by the use of the solvatochromic comparison method. Anal Chem 1985;57:2971–8. https://doi.org/ 10.1021/ac00291a049. [12] Baczek T, Kaliszan R. Combination of linear solvent strength model and quantitative structure-retention relationships as a comprehensive procedure of approximate prediction of retention in gradient liquid chromatography. J Chromatogr A 2002;962: 41–55. https://doi.org/10.1016/S0021-9673(02)00557-5. [13] Studzinska S, Molikova M, Kosobucki P, Jandera P, Buszewski B. Study of the interactions of ionic liquids in IC by QSRR. Chromatographia 2011;73:S35–44. https://doi.org/ 10.1007/s10337-011-1960-3.

References

[14] Remya GS, Suresh CH. Quantification and classification of substituent effects in organic chemistry: a theoretical molecular electrostatic potential study. Phys Chem Chem Phys 2016;18:20615–26. https://doi.org/10.1039/c6cp02936a. [15] Purcell WP, Bass GE, Clayton JM. Strategy of drug design: a guide to biological activity. New York: John Wiley and Sons; 1973. [16] Cherkasov A, Muratov EN, Fourches D, Varnek A, Baskin II, Cronin M, Dearden J, Gramatica P, Martin YC, Todeschini R, Consonni V, Kuz’min VE, Cramer R, Benigni R, Yang C, Rathman J, Terfloth L, Gasteiger J, Richard A, Tropsha A. QSAR modeling: where have you been? Where are you going to? J Med Chem 2014;57:4977– 5010. https://doi.org/10.1021/jm4004285. [17] Kaliszan R. Structure and retention in chromatography. A chemometric approach. Amsterdam: Harwood Academic Publishers; 1997. [18] Moriwaki H, Tian Y-S, Kawashita N, Takagi T. Mordred: a molecular descriptor calculator. J Cheminformatics 2018;10:4. https://doi.org/10.1186/s13321018-0258-y. [19] Haddad PR, Taraji M, Sz€ucs R. Prediction of analyte retention time in liquid chromatography. Anal Chem 2021;93:228–56. https://doi.org/10.1021/acs.analchem.0c04190. [20] Bogusław B, Noga S. Hydrophilic interaction liquid chromatography (HILIC)—a powerful separation technique. Anal Bioanal Chem 2012;402:231–47. https://doi.org/ 10.1007/s00216-011-5308-5. [21] Jandera P. Stationary and mobile phases in hydrophilic interaction chromatography: a review. Anal Chim Acta 2011;692:1–25. https://doi.org/10.1016/j.aca.2011.02.047. [22] Poole CF, Lenca N. Applications of the solvation parameter model in reversed-phase liquid chromatography. J Chromatogr A 2017;1486:2–19. https://doi.org/10.1016/j. chroma.2016.05.099. [23] Poole CF. Influence of solvent effects on retention of small molecules in reversed-phase liquid chromatography. Chromatographia 2019;82:49–64. https://doi.org/10.1007/ s10337-018-3531-3. [24] Chirita R-I, West C, Zubrzycki S, Finaru A-L, Elfakir C. Investigations on the chromatographic behaviour of zwitterionic stationary phases used in hydrophilic interaction chromatography. J Chromatogr A 2011;1218:5939–63. https://doi.org/10.1016/j. chroma.2011.04.002. [25] Buszewski B, Zˇuvela P, Sagandykova G, Walczak-Skierska J, Pomastowski P, David J, Wah Wong M. Mechanistic chromatographic column characterization for the analysis of flavonoids using quantitative structure-retention relationships based on density functional theory. Int J Mol Sci 2020;21:2053. https://doi.org/10.3390/ijms21062053. [26] Nord LI, Fransson D, Jacobsson SP. Prediction of liquid chromatographic retention times of steroids by three-dimensional structure descriptors and partial least squares modeling. Chemom Intell Lab Syst 1998;44:257–69. https://doi.org/10.1016/S0169-7439(98) 00070-7. [27] Wen Y, Amos RIJ, Dolan JW, Talebi M, Haddad PR, Szucs R, Pohl CA. Retention prediction in reversed phase high performance liquid chromatography using quantitative structure-retention relationships applied to the hydrophobic subtraction model. J Chromatogr A 2018;1541:1–11. https://doi.org/10.1016/j.chroma.2018.01.053. [28] Grisoni F, Ballabio D, Todeschini R, Consonni V. Molecular descriptors for structure–activity applications: a hands-on approach. Comput Toxicol 1800;2018: 3–53. https://doi.org/10.1007/978-1-4939-7899-1_1.

817

818

CHAPTER 30 Prediction of retention in liquid chromatography

[29] Belhassan A, Chtita S, Lakhlifi T, Bouachrine M. 2D- and 3D-QSRR studies of linear retention indices for volatile alkylated phenols. In: Dysfunction of olfactory system. Intech Open; 2019. https://doi.org/10.5772/intechopen.8957. [30] Kaliszan R, Ba˛czek T, Cimochowska A, Juszczak P, Wisniewska K, Grzonka Z. Prediction of high performance liquid chromatography retention of peptides with the use of quantitative structure-retention relationships. Proteomics 2005;5: 409–15. https://doi.org/10.1002/pmic.200400973. [31] Studzinska S, Buszewski B. Different approaches of quantitative structure-retention relationships in the prediction of oligonucleotides retention. J Sep Sci 2015;38:2076– 84. https://doi.org/10.1002/jssc.201401395. [32] Szultka-Mlynska M, Buszewski B. Chromatographic behavior of selected antibiotic drugs supported by quantitative structure-retention relationships. J Chromatogr A 2016;1478:50–9. https://doi.org/10.1016/j.chroma.2016.11.057. [33] Walczak-Skierska J, Szultka-Mły nska M, Pauter K, Buszewski B. Study of chromatographic behavior of antibiotic drugs and their metabolites based on quantitative structure-retention relationships with the use of HPLC-DAD. J Pharm Biomed Anal 2020;184, 113187. https://doi.org/10.1016/j.jpba.2020.113187. [34] Skoczylas M, Bocian S, Buszewski B. Quantitative structure retention relationships of amino acids on the amino acid- and peptide-silica stationary phases for liquid chromatography. J Chromatogr A 2020;1609, 460514. https://doi.org/10.1016/j. chroma.2019.460514. [35] Buszewski B, Walczak J, Skoczylas M, Haddad PR. High performance liquid chromatography as a molecular probe in quantitative structure-retention relationships studies of selected lipid classes on polar-embedded stationary phases. J Chromatogr A 2019;1585:105–12. https://doi.org/10.1016/j.chroma.2018.11.053. [36] Ciura K, Belka M, Kawczak P, Ba˛czek T, Nowakowska J. The comparative study of micellar TLC and RP-TLC as potential tools for lipophilicity assessment based on QSRR approach. J Pharm Biomed Anal 2018;149:70–9. https://doi.org/10.1016/j. jpba.2017.10.034. [37] Buszewski B, Walczak-Skierska J, Wrona O, Wojtczak I. Linear solvation energy relationships in the determination of phospholipids by supercritical fluid chromatography. J Supercrit Fluids 2021;173, 105206. https://doi.org/10.1016/j.supflu.2021.105206. [38] Ciura K, Kawczak P, Greber KE, Kapica H, Nowakowska J, Ba˛czek T. Application of reversed-phase thin layer chromatography and QSRR modelling for prediction of protein binding of selected β-blockers. J Pharm Biomed Anal 2019;176, 112767. https://doi.org/ 10.1016/j.jpba.2019.07.015. [39] Gao H, Huang H, Zheng A, Yu N, Li N. Determination of quantitative retention-activity relationships between pharmacokinetic parameters and biological effectiveness fingerprints of Salvia miltiorrhiza constituents using biopartitioning and microemulsion high-performance liquid chromatography. J Chromatogr B Analyt Technol Biomed Life Sci 2017;1067:10–7. https://doi.org/10.1016/j.jchromb.2017.09.018. [40] Santoro AL, Carrilho E, Lanc¸as FM, Montanari CA. Quantitative structure–retention relationships of flavonoids unraveled by immobilized artificial membrane chromatography. Eur J Pharm Sci 2016;88:147–57. https://doi.org/10.1016/j.ejps.2015.12.009. [41] Feenstra P, Gruber-W€olfler H, Brunsteiner M, Khinast J. Retention-time prediction for polycyclic aromatic compounds in reversed-phase capillary electro-chromatography. J Mol Model 2015;21:124. https://doi.org/10.1007/s00894-015-2668-3.

References

[42] Sagandykova G, Buszewski B. Perspectives and recent advances in quantitative structure-retention relationships for high performance liquid chromatography. How far are we? Trends Analyt Chem 2021;141:116294. https://doi.org/10.1016/j. trac.2021.116294. [43] Bajusz D, Racz A, Heberger K. Why is Tanimoto index an appropriate choice for fingerprint-based similarity calculations? J Cheminformatics 2015;7:20. https://doi. org/10.1186/s13321-015-0069-3. [44] Willett P. Similarity methods in chemoinformatics. Annu Rev Inf Sci Technol 2009;43:3–71. https://doi.org/10.1002/aris.2009.1440430108. [45] Zˇuvela P, Skoczylas M, Jay Liu J, Baczek T, Kaliszan R, Wah Wong M, Buszewski B. Column characterization and selection systems in reversed-phase high-performance liquid chromatography. Chem Rev 2019;119:3674–729. https://doi.org/10.1021/acs. chemrev.8b00246. [46] Daghir-Wojtkowiak E, Studzinska S, Buszewski B, Kaliszan R, Markuszewski MJ. Quantitative structure–retention relationships of ionic liquid cations in characterization of stationary phases for HPLC. Anal Methods 2014;6:1189. https://doi.org/10.1039/ C3AY41805G. [47] Chen B, Zhang T, Bond T, Gan Y. Development of quantitative structure activity relationship (QSAR) model for disinfection byproduct (DBP) research: a review of methods and resources. J Hazard Mater 2015;299:260–79. https://doi.org/10.1016/j. jhazmat.2015.06.054. [48] Eriksson L, Johansson E. Multivariate design and modeling in QSAR. Chemom Intell Lab Syst 1996;34:1–19. https://doi.org/10.1016/0169-7439(96)00023-8. [49] Sagandykova GN, Pomastowski PP, Kaliszan R, Buszewski B. Modern analytical methods for consideration of natural biological activity. Trends Anal Chem 2018;109:198–213. https://doi.org/10.1016/j.trac.2018.10.012.

819

CHAPTER

Validation of liquid chromatographic methods

31

Kimber L. Barnetta, Brent Harringtonb, and Timothy W. Graula,c a

Analytical Research and Development, Pfizer Inc., Groton, CT, United States, bNonclincial Statistics, Pfizer Inc., Groton, CT, United States, cGlobal CMC, Pfizer Inc., Groton, CT, United States

31.1 Discussion 31.1.1 Traditional method validation The purpose of method validation is to demonstrate that an analytical method is suitable for its intended purpose and, for a quantitative method, provides a reliable estimate of the actual value of the sample tested. Method validation involves assessing method performance against predefined criteria that are established based on product or sample specifications and the type of measurement to be performed, for example, assay, identification, or limit test. An assessment of method performance versus this predefined criterion is performed to provide a reasonable level of assurance that the method is suitable for use. Method characteristics that are commonly evaluated during a validation study are described by several guidances [1,2], some of which are shown in Tables 31.1 and 31.2. In general, an analytical method is validated prior to use. In addition, it is a good practice to periodically re-evaluate the method itself as well as the associated validation criteria and data to ensure that the method remains appropriate. High-profile cases of failure of analytical methods illustrate the importance of periodically reevaluating methods. The first example involves heparin, an injectable anticoagulant drug. The analytical methods used to quantitate heparin were unable to differentiate between heparin and over-sulfated chondroitin sulfate, an adulterant that caused severe allergic reactions and death in some patients [10]. A second example is the adulteration of milk products and pet food with melamine that killed several infants [11] and as many as one thousand dogs and cats [12]. Melamine was added to the products to increase nitrogen levels of the products and was not discriminated by the analytical method used to measure protein levels of the samples. The heparin and melamine examples illustrate the importance of demonstrating and then periodically confirming that an analytical method has the ability to discriminate the intended analyte. Liquid Chromatography. https://doi.org/10.1016/B978-0-323-99968-7.00035-7 Copyright # 2023 Elsevier Inc. All rights reserved.

821

Table 31.1 Examples of method validation guidelines. Organization

Guideline

Description

International Conference on Harmonization (ICH) [1] International Union of Pure and Applied Chemistry (IUPAC) [2] Eurachem

Validation of Analytical Procedures [1]

Guidance for analytical methods presented in pharmaceutical applications Guidance for methods of all types

Association of Official Analytical Chemists (AOAC)

Harmonized Guidelines for Single-Laboratory Validation of Methods and Analysis [2] The Fitness for Purpose of Analytical Methods: a Laboratory Guide to Method Validation and Related Topics [3] Guidelines for single-Laboratory Validation of Analytical Methods for Trace-Level Concentrations of Organic Chemicals [4]

Food and Agriculture Organization (FAO) International Atomic Energy Agency (IAEA) European Medicines Agency (EMA)

Guideline on Bioanalytical Method Validation [5]

World Health Organization (WHO)

Analytical Method Validation [6]

National Association of Testing Authorities, Australia (NATA) Irish National Accreditation Board

Guidelines for the validation and verification of chemical test methods [7] Guide to Method Validation for Quantitative Analysis in Chemical Testing Laboratories [8] Defines terminology associated with metrology, including method validation [9]

International Vocabulary of Metrology—Basic and General Concepts and Associated Terms

Guidance for methods of all types

Harmonized guidance providing recommendations for validation of methods used to quantitate trace level organic analytes such as pesticide residues in food and water

Guidance for methods used in pharmacokinetic studies and filed in pharmaceutical applications in the EU Guidance for inspectors of pharmaceutical manufacturing facilities and manufacturers of pharmaceutical products Guidance for NATA accredited labs for validation of methods of all types Guidelines for validation for accreditation in accordance with ISO 17000 series of standards Applies to all method types

Table 31.2 Summary of validation characteristics recommended by ICH and IUPAC guidelines. Validation characteristics for liquid chromatography Specificity (ICH) Selectivity (IUPAC) Accuracy (ICH) Trueness (IUPAC) Precision

Precision: Repeatability Precision: Intermediate precision (ICH) Precision under run-torun conditions (IUPAC) Precision: Reproducibility (ICH)

ICH definition [1]

IUPAC definition [2]

“…ability to assess unequivocally the analyte in the presence of components which may be expected to be present. Typically these might include impurities, degradants, matrix, etc.” “…expresses the closeness of agreement between the value which is accepted either as a conventional true value or an accepted reference value and the value found” “…the closeness of agreement (degree of scatter) between a series of measurements obtained from multiple sampling of the same homogeneous sample under the prescribed conditions” “…expresses the precision under the same operating conditions over a short interval of time..” Also called intra-assay precision “…expresses within-laboratory variations: different days, different analysts, different equipment, etc.”

“…the degree to which a method can quantify the analytes accurately in the presence of inteferents”

“…expresses the precision between laboratories…”

“… the closeness of agreement between a test result and the accepted reference value of the property being measured” “…the closeness of agreement between independent test results obtained under stipulated conditions”

“...variations observed during a single run”

Variations observed from run-to-run in a single laboratory

“..it may be useful for the assessment of….separate operator and run effects, between or within-day effects or the precision attainable using one or several instruments” Continued

Table 31.2 Summary of validation characteristics recommended by ICH and IUPAC guidelines—cont’d Validation characteristics for liquid chromatography Detection Limit (DL) (ICH and IUPAC) Also called Limit of Detection (IUPAC) Quantitation Limit, QL (ICH) Limit of Determination or Limit of Quantificaiton (IUPAC) Linearity Range

Robustness (ICH) Ruggedness (IUPAC)

ICH definition [1]

IUPAC definition [2]

“...the lowest amount of analyte in a sample which can be detected but not necessarily quantitated as an exact value”

“…the smallest amount or concentration of analyte in the test sample that can be reliably distinguished from zero”

“…the lowest amount of analyte in a sample which can be quantitatively determined with suitable precision and accuracy”

“...a concentration below which the analytical method cannot operate with an acceptable precision.” Not recommended by IUPAC as part of method validation. Instead, it is recommended to perform an assessment of measurement uncertainty as a function of concentration and compare with a fit for purpose criterion “Linearity can be tested informally by examination of a plot of residuals produced by linear regression for the responses…” “...the interval of analyte concentration within which the method can be regarded as validated”

“…ability (within a given range) to obtain test results which are directly proportional to the concentration (amount) of analyte in the sample” “…the interval between the upper and lower concentration (amounts) of analyte in the sample (including these concentrations) for which it has been demonstrated that the analytical procedure has a suitable level of precision, accuracy and linearity” “...a measure of its capacity to remain unaffected by small, but deliberate variations in method parameters and provides indication of its reliability during normal usage”

“...the resistance to change in the results produced by an analytical method when minor deviations are made from the experimental conditions described in the procedure”

31.1 Discussion

The focus of this chapter is validation of liquid chromatographic methods. There are numerous articles and guides that describe details and provide examples of validation of LC methods [13–15]; therefore, a step-by-step guide is not provided here. This chapter will provide an overview as well as a discussion about some of the difficulties associated with traditional validation and some alternative practices that are currently being discussed in the literature. The common LC method validation characteristics are listed in Table 31.2. Also listed in Table 31.2 are definitions for the different validation characteristics according to the International Conference on Harmonization (ICH) [1] and the International Union of Pure and Applied Chemistry (IUPAC) [2]. ICH guidelines apply to methods used for testing of pharmaceutical materials intended for human use, while the IUPAC guidance provides recommendations for analytical methods, including LC methods, used across multiple industries. In addition to the guidelines shown in Tables 31.1 and 31.2, industries, such as environmental, pharmaceutical, and health, have regulatory requirements and standards which must be followed in order to receive approval from a regulatory body. While validation requirements of most regulatory bodies are similar, there are differences [16] and it is prudent to be aware of regulatory requirements for a particular industry in a given country to ensure appropriate validation characteristics are evaluated. The terms and definitions provided below are consistent with the ICH guidance, Q2(R1) [1]: Identification (ID) test: The objective of an ID test is to confirm that the chemical species being analyzed is the correct species. An ID test must be able to discriminate the intended analyte from other related species, impurities, or matrix components that may be present; this is termed as the method specificity. It is typically demonstrated for an LC method by measuring resolution between the analyte of interest and other species of that may be present and confirming the homogeneity of the peak of interest using an orthogonal approach, such as 2D-HPLC or detector response (e.g., mass spectrometry or UV spectra). Limit test: According to the ICH guidance, the following characteristics are required to validate an LC limits test: method specificity and the detection limit (DL). Method specificity is discussed above for ID tests. The detection limit can be assessed using different approaches: (1) visual evaluation of the peak, (2) peak signal-to-noise (S/N), or (3) calculated from the response and the standard deviation of the slope of the calibration curve. The IUPAC [2] guidance Harmonized Guidelines for Single-Laboratory Validation of Methods of Analysis discusses the difficulties associated with accurately estimating the DL. For example, DL (a) estimates are often biased low, (b) they can have significant variability due to the fact that the different approaches (outlined above) often lead to different DL estimates, and (c) estimates can be highly variable unless a large data set is collected [2]. Assay for content: The following characteristics are recommended for LC assay validation: accuracy, precision, specificity, linearity, and range. The quantitation limit (QL) is assessed for analytes having responses near the QL. Specificity is evaluated as discussed above for ID tests.

825

826

CHAPTER 31 Validation of liquid chromatographic methods

Accuracy: Accuracy is a measure of agreement of the measured value with the known or accepted reference value. It is also called trueness. It is generally established by measuring samples of known composition (either authentic samples or artificially prepared solutions) across the range of expected concentrations and demonstrating the values are within the pre-established criteria. Precision: Precision reflects the ability of a method to provide consistent measurements and can be assessed in multiple ways: •





Repeatability (also called intra-assay precision) is a measure of the ability of a method to provide consistent measurements over a short time period, which is generally within a given run or sample sequence. It is assessed by measuring the precision of multiple test results and confirming adherence to pre-defined criteria. Intermediate precision is a measure of the ability of a method to provide consistent measurements within a given laboratory when testing is performed by different analysts using different equipment on different days, etc. Reproducibility is a measure of the ability of a method to provide consistent performance over time in different labs, with different analysts using different equipment. This is the most difficult to assess but most closely approximates the behavior that the method will display over its lifetime.

Linearity: Linearity is established by demonstrating a mathematical relationship between the detector response and analyte concentration. While many LC methods used for routine testing rely on calibration curves exhibiting a straight line, other relationships are acceptable if they are shown to be appropriate. Range: The validated method should state the range of concentrations over which the method has been shown to meet validation criteria. Quantitation limit (QL): The method QL should be confirmed to meet preestablished criteria when the sample response is near the QL. Method QL is established in a manner similar to the DL: (1) visual evaluation of the peak, (2) peak signal-to-noise (S/N), or (3) calculated from the response and the standard deviation of the slope of the calibration curve. Robustness: Method robustness is the ability of the method to provide suitable performance when small changes are made. For LC methods, it is traditionally demonstrated by measuring the effect of small, deliberate changes in method parameters (e.g., flow rate, column temperature, etc.) on performance and is usually assessed during method development. Although not described in ICHQ2(R1), experimentally designed studies are often executed when assessing method robustness [2,17–19]. Data acquired during robustness testing, method development, and validation studies can be used to establish system suitability. System suitability parameters are method specific criteria which an LC system in routine use must meet in order to be considered suitable for use [20]. Solution stability: The storage requirements of calibration standards and samples, including maximum allowable storage time, should be appropriate for the

31.1 Discussion

intended use of the method. Solution stability should be investigated as part of method development and confirmed during method validation. The validation approaches described above provide reasonable assurance that a method will perform as needed. However, these validation studies often occur as a single event early in the life cycle of a method and may not always predict performance during routine use that can take place over years. For example, validated LC methods can exhibit loss of required resolution, peak shape changes or shifting retention times which can lead to variability in the final results or inability to meet system suitability. This can occur if the effect of method conditions on performance is not well understood and controlled. This includes the effect of small and expected dayto-day variation in method conditions, different instruments, different analysts, and column changes that can occur over time. Product variability can also influence method performance, for example, changes in extraction efficiency may occur as a product ages or small changes in the sample matrix or inactive ingredients may contribute to method variability. Interlaboratory validation studies can be performed to better understand how the method will perform under different conditions [21–23] though these studies also represent a single event in the method life cycle and may not predict future performance. Difficulties can also be encountered if validation exercises are treated as isolated exercises, disconnected from knowledge gained during method development, and the future use of the method. Establishing incorrect validation criteria where the origin of the criteria is not well understood can also contribute to method difficulties. Enhanced approaches, which can be applied over the entire method life cycle from development through routine use and subsequent changes, are being evaluated to address some of these concerns [24–27].

31.1.2 Enhanced approaches Enhanced approaches, also captured by the Procedure Life Cycle approach [26], rely on a combination of scientific knowledge and risk management to understand and control the sources of variability. They are aligned with Quality by Design (QbD) principles, which are described for the pharmaceutical industry in ICH guidances Q8-Q12 [2,10–14,16–18,20,28–30]. These concepts can also be applied to provide a framework for greater analytical method understanding, from method development, through routine use and changes that may be needed over the life cycle of the method. This philosophy is not new for chromatographers as many of these concepts are good scientific practices that have been applied to methods well before QbD terminology was introduced. Contemporary efforts focus on providing a more structured framework for better consistency. Many of these efforts have leveraged increasingly powerful in silico capabilities to provide a good understanding of the effect of method conditions and controls on method performance without significantly increasing the amount of experimental work required. Additionally, in an enhanced approach, greater emphasis is placed in the importance of connecting all activities, from development through routine use, and ensuring this knowledge remains available throughout the method life cycle.

827

828

CHAPTER 31 Validation of liquid chromatographic methods

Enhanced approaches to the method life cycle emphasize the following [31]: A. Establishing method criterion that is linked to the purpose of the method. The criterion has been termed the Analytical Target Profile (ATP) [26,32,33]. B. Applying a systematic approach to method development that relies on sound scientific and experimental design principles, which includes prior knowledge. This may also include in silico simulations. a. Taking a risk-based approach to prioritize experiments designed to understand the effect of method conditions (e.g., mobile phase composition, temperature, instrument type, dissolving solvent, etc.) and sample preparation on chromatographic behavior and ultimately the test results that are generated by the method. b. Application of fundamental knowledge, in silico, and experimental modeling to develop a good understanding of how method parameters affect chromatographic behavior. C. Establishing a meaningful analytical procedure control strategy (for example, system suitability) based on an understanding of the relationship between method conditions and method performance. D. Periodically reviewing the analytical method to ensure it remains appropriate for its intended use and updating or replacing it as appropriate. The extent of application of these concepts depends on how the test results are used, i.e., the decisions that will made with the test results. It is not necessary to fully apply these concepts in all cases to all methods. Examples of application of these principles can be found in the literature, [8,9,23–25,27,34–94] and each topic is discussed below.

31.1.2.1 The analytical target profile (ATP) The ATP describes method performance criteria that an analytical method must meet in order to be used for the described purpose [26,27,33]. The ATP directs initial method development, transfer, and any subsequent changes to the method over its life cycle. It has been proposed that the ATP should describe the requirements for the final results [26] (sometimes called reportable results or test results) that are generated by the method and not necessarily all the individual validation characteristics commonly applied. When structured this way, the ATP describes the “quality” of the test results to be generated. This provides a direct link to the decisions that will be made using the results and is therefore independent of the type of analytical technique that is applied. The ATP is a relatively recent concept and there is not yet a consensus as to what criteria should be included in an ATP, how to establish the criteria, or how a method can be shown to meet an ATP. An example ATP is discussed here; however, other approaches may also be appropriate [26,27,33,61–63,89–92,95–103]. In order to make proper inferences about results, acceptable analytical methods must produce accurate and precise results over the range of analyte for which the method is intended. To better assess this, it has been proposed that method validation

31.1 Discussion

criteria should include (1) agreement of a measurement with the true value and (2) a statement about the risk of future results differing from the nominal (or true) value [95,104,105]. All analytical measurements display an associated level of disparity in the form of systematic error (bias) and random error (standard deviation). It is good practice to determine the sources of LC method variability and to minimize them. Of particular interest should be the minimization of measurement variability to be aligned with the decisions to be made concerning sample specifications or requirements [106,107]. For example, Figure 31.1 illustrates the risk of making an incorrect decision when using a hypothetical method in the following situation: true method accuracy of 100% (0% bias), true method precision of approximately 0.6% standard deviation (std), true product sample average of 97%, and sample lower specification of 98%. In this example, it is assumed that only measurement variability accounts for any deviation from 97%. The curve in Figure 31.1 illustrates the distribution of possible test results provided by the method assuming 0% bias and 0.6% standard

50 Distribution of possible reportable values (based upon method accuracy & precision)

Frequency

40

Lower Product Specification

risk of (incorrectly) accepting a subpotent lot ≥5%

30

True Sample Value

20

10 95

96

97

98

99

100

Analyte

FIGURE 31.1 Understanding risk, precision, and accuracy coexistence on inferences made from reportable quality measurements. Illustrated is the distribution (bell shaped curve) of measurements generated by an analytical method with a true accuracy of 100% (0% bias) and true precision of approximately 0.5% standard deviation. Illustrated is an example for a sample having a true average of 97% (1% below the specification of 98%), and only measurement error accounts for any deviation of reported values from 97%. The x-axis represents an analyte assay value and the y-axis represents the frequency of a corresponding test result.

829

CHAPTER 31 Validation of liquid chromatographic methods

50 Distribution of possible reportable values (based upon method accuracy & precision)

40

Frequency

830

+0.5 Bias

Lower Product Specification 30

20

risk of (incorrectly) accepting a subpotent lot ∼16%

True Sample Value

10 95

96

97

98

99

100

Analyte

FIGURE 31.2 Effect of positive bias of analytical method on risk. Illustrated is the distribution (bell shaped curve) of measurements generated by an analytical method with a true accuracy of 100.5% (0.5% bias) and true precision of approximately 0.5% standard deviation. Illustrated is an example for a sample having a true average of 97% (1% below the specification of 98%) and only measurement error accounts for any deviation of reported values from 97%. The x-axis represents an analyte assay value and the y-axis represents the frequency of a corresponding test result.

deviation. In this case, the true mean of the batch being tested is below the minimum limit or specification. A measurement above 98% for a sample may result in a decision to incorrectly accept a sub-potent lot as being within the specification. This hypothetical method will correctly make the decision to reject the sub-potent lot with high reliability because a large proportion of test results will be below the 98% specification as illustrated by the large area under the curve for test values less than 98%. Figure 31.2 illustrates the risk of making an incorrect decision when using a hypothetical method under the same conditions as shown in Figure 31.1 but with the method having a positive bias of 0.5% (vs 0% bias shown in Figure 31.1). Figure 31.2 illustrates how a positive analytical bias will increase the risk of accepting a sub-potent lot or sample as represented by the larger area under the curve that is above 98%. It is the risk of making an incorrect inference by accepting a sub-potent lot (as illustrated by this example) that is controlled by the criteria in an ATP. This concept of risk is not currently included in traditional method validation guidelines; however, it has been incorporated in other applications [106].

31.1 Discussion

70 Distribution of possible reportable values (based upon method accuracy & precision)

60

Lower Product Specification

Frequency

50

40

30

risk of (incorrectly) accepting a subpotent lot ≥5%

True Sample Value

20

10 95

96

97

+0.5 Bias

98

99

100

Analyte

FIGURE 31.3 Effect of increased precision (reduced variability) in the presence of bias of analytical method on risk. The solid curve is generated under the conditions discussed in Figure 31.1. The dashed curve represents the distribution of measurements generated by an analytical method with a true accuracy of 100.5% (0.5% bias) and true precision of approximately 0.26% standard deviation. Illustrated is an example for a sample having a true average of 97% (1% below the specification of 98%) and only measurement error accounts for any deviation of reported values from 97%. The x-axis represents an analyte assay value and the y-axis represents the frequency of a corresponding test result.

Figure 31.3 illustrates the risk of making an incorrect decision when using a hypothetical method under the same conditions as shown in Figure 31.1 but with the method having higher precision (standard deviation approximately 0.26%). Figure 31.3 illustrates how a method with measurable bias (dotted curve) mandates an increase in method precision (or decrease in method variability) in order to maintain the same level of risk of incorrectly accepting the sub-potent lot as that of a method with 0% bias. Note that the area under both curves greater than 98% is the same at 5%. The above examples illustrate why method accuracy, precision, and risk of failing a defined criterion should be considered when establishing method performance criteria. An example ATP statement is shown below that describes both risk influencing factors (accuracy and precision) as well as a maximum risk level for measurements that fall outside this criterion.

831

CHAPTER 31 Validation of liquid chromatographic methods

1.00

0.75

Sigma

832

0.50

0.25

0.00 –1

Target

+1

Deviation from Target

FIGURE 31.4 Graphical representation showing the difference between ATP criteria as described here (parabolic-shaped curve) and traditional acceptance criteria where accuracy and precision are established independently (rectangular area.) The ATP criteria: the measured value is within 1.0% of the true value with 95% probability. Traditional criteria: no more than 1.0% bias and no more than 1.0% standard deviation. The x-axis represents method accuracy (bias, % deviation from true value) and the y-axis represents method precision (σ, std).

Example ATP The procedure must be able to accurately quantify {analyte} in {matrix or sample} over a range of 90–110% of the nominal concentration with accuracy and precision such that measurements fall within ±1.0% of the true value with at least a 95% probability. A graphical representation of the ATP is shown in Figure 31.4 below (parabolic shaped curve) and is compared to an example of traditional criteria (rectangle). Figure 31.4 illustrates a notable benefit of the ATP statement as defined above: it provides a trade-off between precision and accuracy criteria in order to maintain the same probability of measurements being within a given range of the true value as described by the ATP. This is not achieved with the traditional approach where criteria for accuracy and precision are established and assessed separately [108]. The traditional approach allows a method to operate in the upper corners of the rectangle, which are regions of lower precision and higher bias, resulting in a method that may not be fit for purpose. The formula below defines these criteria for the ATP. The formula states a performance-based criterion, as a composite of the true method mean (accuracy, μ) and the true method standard deviation (precision, σ). Thus, the ATP defines the acceptable set of {μ, σ} combinations that meet or exceed the predetermined

31.1 Discussion

performance-based criterion, which can be visualized as a two-dimensional region as shown in Figure 31.4. The expression for defining the acceptance region is the following: Z ATP ¼ fμ, σ g |

T+e

ϕðy : μ, σ Þdy  p

Te

where μ ¼ true method mean/accuracy; a parameter. σ ¼ true method sigma/precision; a parameter. e ¼ distance from true value (allowable analytical window), a fixed constant. y ¼ individual assay value; a random variable with mean μ and standard deviation σ T ¼ true analytical content, fixed target. p ¼ minimum probability for individual assay to reside within error bound e, fixed constant. Φ(.) ¼ normal density function centered at μ, with standard deviation σ. In summary, the ATP example shown here describes critical method performance requirements, that is, those characteristics that have a direct impact on the ability of a method to quantitate an analyte. LC method attributes such as peak resolution, linearity, or efficiency are not included. Although they are important features of an LC method and should be evaluated and incorporated into the control strategy as required, they do not provide a direct measurement of the ability of a method to accurately quantitate an analyte and should not appear in the ATP. Alternatively, it is possible to construct an ATP that describes the required measurement uncertainty [51,66,95,99–103]. The International Vocabulary of Basic and General Terms in Metrology (VIM) [9] defines measurement uncertainty as the “… parameter characterizing the dispersion of the quantity values being attributed to a measurand…” where the measurand is the quantity being measured. The target measurement uncertainty is the “measurement uncertainty specified as an upper limit and decided on the basis of the intended use of measurement results.” The Eurachem guide [102] provides more information about measurement uncertainty, including how to establish criteria and assess it. It also provides examples for a variety of measurement types, including LC.

31.1.2.2 Technique selection and method development Several analytical techniques may be capable of meeting the requirements of an ATP; for example, LC, GC, or NIR may all be suitable. Considerations that enter into technique selection include physicochemical properties of the analytes, previous development knowledge, and the measurement technique that best meets business requirements. Typical business requirements include the desire for at/online measurements, technology available in the testing laboratories, analysis/cycle time and cost, and the need for sample preparation. LC methodology should be designed in a systematic manner and there are numerous articles and books detailing rational method development procedures [35,37–44,52–55], including a chapter in this book.

833

834

CHAPTER 31 Validation of liquid chromatographic methods

In addition to identifying chromatographic conditions, appropriate dissolving solvents and conditions should be identified for the preparation of standards and test solutions. The solvents and sample preparation conditions should be chosen based on knowledge and in silico calculation of the properties of the analytes and compatibility with the chosen LC conditions [109]. While understanding the impact of method parameters on method performance is important, a deep understanding of the sample, particularly how it relates to the manufacturing process, provides insight on what the method must deliver.

31.1.2.3 Analytical method risk assessment Risk assessments can be conducted once method conditions have been identified. The purpose of a risk assessment is to identify method and sample preparation parameters (e.g., flow rate and mobile phase composition) that can affect system performance and measurement. This is consistent with the approach outlined in ICH Q9 for Quality Risk Management [29]. The risk assessment process relies on (1) knowledge of the physical and chemical properties of the analyte and fundamentals of the measurement system, (2) experience with similar methods, and (3) data collected during method development experiments. Quality risk management tools used for analytical risk assessments are similar to those used to identify possible hazards within a manufacturing process. A variety of tools are used [110], including the cause and effect matrix, Ishikawa diagram (fishbone diagram), and failure mode and effects analysis (FMEA). Ideally, the end-users of the LC method should participate in the risk assessment in order to better understand differences in lab practices that could lead to method non-robustness. For example, differences in equipment, reagents, and sample handling can contribute to unexpected method difficulties and should be considered.

31.1.2.4 Develop understanding and identify operating conditions Following completion of a risk assessment, an experimental plan is developed to investigate the method parameters ranked highest in terms of ability to influence performance. Design of Experiments (DoE) can be used for this purpose so that statistical relationships between method parameters (e.g., flow rate, organic composition, or gradient) and attributes (e.g., accuracy, precision, resolution) can be established [36–43,50,111]. Mechanistic modeling can also be used in combination with empirical modeling [43,111]. With these relationships, suitable operating conditions can be identified from combinations of method parameters where it is predicted that method performance meets ATP criteria or surrogate criteria such as resolution, sensitivity, peak shape, retention time, etc. Surrogate criteria are beneficial since it can be time-consuming to evaluate accuracy, precision, and measurement uncertainty at this stage given the number of design points in typical DoEs.

31.1 Discussion

31.1.2.5 Method validation After suitable method conditions are identified, the method can be validated according to the relevant standards and guidelines. There are numerous references that describe traditional method validation practices [1–7,15,56–60] and some alternative approaches have also been proposed. [44–50,104] The method can also be qualified against ATP criteria at this stage [26,27,33,67,95–103]. It is helpful to have knowledge of statistics in order to design and interpret the studies needed to qualify a method vs an ATP statement [24,27].

31.1.2.6 Method control strategy The control strategy, which includes system suitability, is established based on several considerations, including: • •



Adherence to validated/qualified method conditions. Establishment of system suitability criteria based on data collected during development, including DoE studies and validation/qualification studies. This provides a rich data set on which to base method controls compared to the traditional approach where the depth of knowledge about method performance can vary. Establishment of other controls to ensure appropriate method performance. Examples of this include identification of specific equipment, reagent grades, and data processing methods needed to ensure appropriate method performance.

The benefit of an enhanced approach is that method controls are established based on a more extensive data set which provides a better understanding of the relationship between method parameters such as resolution, peak shape, retention time, etc., and method performance [26,112]. This enables establishment of more appropriate control strategies.

31.1.2.7 Life cycle management It is helpful to monitor performance and periodically re-evaluate analytical methods to ensure that they remain appropriate. Based on these assessments, methods can be modified or replaced as appropriate [93,94]. An enhanced approach to analytical procedures is a current topic in the pharmaceutical industry. Many articles showing examples of the concepts above have been published by industry, with many related to application of DoEs for method development [37–50]. In 2015, the FDA revised their method validation guideline to incorporate some of these elements, such as recommending that method robustness be assessed by performing risk assessments and assessing method performance using multifactor DoE studies to understand sources of variation [19]. Beginning in 2014, the USP published several articles discussing considerations related to the life cycle approach, decision rules, ATPs, and control strategies with the chapter, h1220i Analytical Procedure Life Cycle, becoming official in 2022 [98]. The Step 2 versions of ICH draft guidances for industry Q14 Analytical Procedure Development Q2(R2) Validation of Analytical Procedures were released in 2022 [1,68] and include many

835

836

CHAPTER 31 Validation of liquid chromatographic methods

of these life cycle concepts. In addition, more articles are expected from across the industry as the discussion of the merits vs the challenges continues.

31.2 Conclusion Traditional method validation experiments described by ICH and the harmonized IUPAC guidelines provide a reasonable assurance that a method will perform as needed. However, a method validation activity represents a single event that may not represent future performance of a method that is routinely used, often for years. It is important to develop a good level of understanding about how method conditions affect performance so that appropriate method conditions can be identified and controls established. The life cycle approach can enhance the manner in which analytical methods are developed, validated, and maintained by ensuring a good level of understanding that underpins activities across the life cycle. The approach relies on establishing appropriate criteria and performing systematic experiments to understand and control sources of variability.

References [1] ICH. International conference on harmonization of technical requirements for registration of pharmaceuticals for human use, validation of analytical procedures: text and methodology Q2(R1)., 2005, http://www.ich.org. [2] Thompson M, Ellison SR, Wood R. Harmonized guidelines for single-laboratory validation of methods of analysis (IUPAC technical report). Pure Appl Chem 2002;74(5): 835–55. € [3] Magnusson B, Ornemark U, editors. Eurachem guide: the fitness for purpose of analytical methods—a laboratory guide to method validation and related topics. 2nd ed; 2014. ISBN 978-91-87461-59-0. Available from www.eurachem.org. [4] Fajeli A, Ambrus A, editors. Guidelines for single-laboratory validation of analytical methods for trace-level concentrations of organic chemicals in principles and practices of method validation. Royal Soceity of Chemistry; 2000. p. 179–252. [5] European Medicines Agency, EMA. Guideline on bioanalytical method validation, 2012. available for download at http://www.ema.europa.eu/ema/. [6] World Health Organization. Analytical Method Validation. In: Supplementary guidelines on good manufacturing practices: validation, Technical Report Series, Annex 4, No. 937; 2006. Appendix 4 and revised daft (2016). [7] National Association of Testing Authorities. National association of testing authorities, Australia, validation and verification of quantitative and qualitative test methods; 2018. [8] INAB. Guide to method validation for quantitative analysis in chemical testing laboratories, INAB Guide PS15., 3 March 2019, www.inab.ie. [9] BIPM. International vocabulary of metrology—basic and general concepts and associated terms (VIM)., 2017, www.bipm.org. [10] Kemsley J. Heparin undone. Chem Eng News 2008;86(18):38–40. [11] Rovner SL. Silver lining in the melamine crisis. Chem Eng News 2009;87(21):36–8. [12] Rovner SL. Anatomy of a pet food disaster. Chem Eng News 2008;86(18):41–3.

References

[13] Bliesner DM, editor. Validating chromatographic methods: a practical guide. Wiley; 2006. [14] De Bie`vre P, G€unzler H, editors. Validation in chemical measurement. Springer; 2005. [15] Eurachem. The Eurachem reading list, Validation of analytical methods, www. eurachem.org. [16] Rozet E, Ceccatob A, Hubert C, Ziemons E, Oprean R, Rudaz S, Boulanger B, Hubert P. Analysis of recent pharmaceutical regulatory documents on analytical method validation. J Chromatogr A 2007;1158:111–25. [17] Vander Heyden Y, Nijhuis A, Smeyers-Verbeke J, Vandeginsteb BGM, Massart DL. Guidance for robustness/ruggedness tests in method validation. J Pharm Biomed Anal 2001;24:723–53. [18] Borman P, Chatfield MJ, Damjanov I, Jackson P. Method ruggedness studies incorporating a risk based approach: a tutorial. Anal Chim Acta 2011;703:101–13. [19] U.S. Department of Health and Human Services, Food and Drug Administration, editor. Analytical Procedures and Methods Validation for Drugs and Biologics Guidance for Industry; 2015. [20] Wiggins DE. System suitability in an optimized HPLC system. J Liq Chromatogr 1991;14:3045–60. [21] ISO/TR 22971. Accuracy (trueness and precision) of measurement methods and results—Practical guidance for the use of ISO 5725-2:1994 in designing, implementing and statistically analysing interlaboratory repeatability and reproducibility results., 2005, www.iso.org. [22] ASTM E691–14. Standard practice for conducting an interlaboratory study to determine the precision of a test method. ASTM International (www.astm.org). [23] Heyden YV, Smeyers-Verbeke J. Set-up and evaluation of interlaboratory studies. J Chromatogr A 2007;1158:158–67. [24] Rozet E, Ziemons E, Marini RD, Boulanger B, Hubert P. Quality by design compliant analytical method validation. Anal Chem 2012;84:106–12. [25] Kochling J, Bridgewater J, Naji R. Introducing a science-based quality by design concept to analytical methods development. In: Huynh-Ba K, editor. Pharmaceutical stability testing to support global markets. Springer; 2010. p. 169–79. [26] Martin G, et al. Life cycle management of analytical procedures: method development, procedure performance qualification, and procedure performance verification. USP validation and verification expert panel, Pharmacopeial Forum 2013;39(6). [27] Colgan S, Hanna-Brown M, Pellett J, Wrisley L, Sluggett G, Morgado J, Harrington B, Szucs R, KBarnett TG, Steeno G. Pharm Technol 2014;38:9. Add title. [28] Pharmaceutical Development Q8(R2), editor. International Conference on harmonization of technical requirements for registration of pharmaceuticals for human use, ICH harmonized tripartite guideline; 2009. [29] Quality Risk Management Q9, editor. International Conference on harmonization of technical requirements for registration of pharmaceuticals for human use, ICH harmonized tripartite guideline; 2005. [30] Pharmaceutical Quality System Q10, editor. International Conference on harmonization of technical requirements for registration of pharmaceuticals for human use, ICH harmonized tripartite guideline; 2008. [31] Graul TW, Barnett KL, Bale SJ, Gill I, Hanna-Brown M. Quality by design for analytical methods. In: Ende DJa, editor. Chemical engineering in the pharmaceutical industry. Wiley; 2011. p. 545–62.

837

838

CHAPTER 31 Validation of liquid chromatographic methods

[32] Schweitzer M, Pohl M, Hanna-Brown M, Nethercote P, Borman P, Hansen G, Smith K, Larew J. Implications and opportunities of applying QbD principles to analytical measurements. Pharm Technol 2010;34:52–9. [33] Barnett K, McGregor P, Martin G, LeBlond D, Weitzel MLJ, Ermer J, Walfish S, Nethercote P, Gratzl G, Kovacs E. Analytical target profile: structure and application throughout the analytical life cycle. Pharmacopeial Forum 2016;42(5). [34] Monks KE, Rieger H-J, Molna´r I. Expanding the term “design space” in high performance liquid chromatography. J Pharm Biomed Anal 2011;56:874–9. [35] Snyder LR, Kirkland JJ, Glajch JL, editors. Practical HPLC method development. 2nd ed. Wiley; 1997. [36] Jelena T, Igor P, Ana S, Anja T, Jancic-Stojanovic B. Application of analytical quality by design concept for bilastine and its degradation impurities determination by hydrophilic interaction liquid chromatographic method. J Pharm Biomed Anal 2016;125:385–93. [37] Tol T, Kadam N, Raotole N, Desai A, Samanta G. A simultaneous determination of related substances by high performance liquid chromatography in a drug product using quality by design approach. J Chromatogr A 2016;1432:26–38. [38] Mallik R, Raman S, Liang X, Grobin AW, Choudhury D. Development and validation of a rapid ultra-high performance liquid chromatography method for the assay of benzalkonium chloride using a quality-by-design approach. J Chromatogr A 2015;1413:22–32. [39] Karmarkar S, Yang X, Garber R, Szajkovics A, Koberda M. Quality by design (QbD) based development and validation of an HPLC method for amiodarone hydrochloride and its impurities in the drug substance. J Pharm Biomed Anal 2014;100:167–74. [40] Barnett K, Doyle K, Wang K, Morgado J, Harwood J. Applying quality by design principles to analytical methods to gain enhanced method understanding. Am Pharma Rev 2015. [41] Boulanger B, Chiap P, Dewe W, Crommen J, Hubert P. An analysis of the SFSTP guide on validation of chromatographic bioanalytical methods: progresses and limitations. J Pharm Biomed Anal 2003;32(4–5):753–65. [42] Anja T, Ana S, Jancic-Stojanovic B, Mirjana M. Quality by design in the development of hydrophilic interaction liquid chromatography method with gradient elution for the analysis of olanzapine. J Pharm Biomed Anal 2017;134:18–26. [43] Jianmei K, Wei W, Yimin H, Qian G, Juan C-M. A platform analytical quality by design (AQbD) approach for multiple UHPLC-UV and UHPLC-MS methods development for protein analysis. J Pharm Biomed Anal 2016;125:130–9. [44] Hubert C, Lebrun P, Houari S, Ziemons E, Rozet E, Ph H. Improvement of a stabilityindicating method by quality-by-design versus quality-by-testing: a case of a learning process. J Pharm Biomed Anal 2014;88:401–9. [45] Saffaj T, Ihssane AB, Jhilal BF, Bouchafra CH, Laslamib BS, Sosseb SA. An overall uncertainty approach for the validation of analytical separation methods. Analyst 2013;138(16):4677–91. [46] Eric R, Cedric H, Attilio C, Walthere D, Eric Z, Francois M, Karim M, Reinhold W, Bruno S, Bruno B, et al. Using tolerance intervals in pre-study validation of analytical methods to predict in-study results. The fit-for-future-purpose concept. J Chromatogr A 2007;1158(1–2):126–37. [47] Saffaj T, Ihssane B. Uncertainty profiles for the validation of analytical methods. Talanta 2011;85(3):1535–42.

References

[48] Rozet E, Ziemons E, Marini RD, Boulanger B, Hubert P. Reply to the responses on the comments on "Uncertainty profiles for the validation of analytical methods" by Saffaj and Ihssane. Talanta 2012;100:290–2. [49] Saffaj T, Ihssane B. Remarks on "Reply to the responses to the comments on "uncertainty profiles for the validation of analytical methods" by Saffaj and Ihssane". Talanta 2013;106:155–7. [50] Ferreira SLC, Bruns RE, da Silva EGP, Santos WNLD, Quintella CM, David JM, de Andrade JB, Breitkreitz MC, Jardim ICSF, Neto BB. Statistical designs and response surface techniques for the optimization of chromatographic systems. J Chromatogr A 2007;1158:2–14. [51] Meyer VR. Measurement uncertainty. J Chromatogr A 2007;1158:15–24. [52] Molna´r I, Rieger H-J, Monks KE. Aspects of the “design space” in high pressure liquid chromatography method development. J Chromatogr A 2010;1217:3193. [53] Molna´r I, Rieger H-J, Korma´ny R. Modeling of HPLC methods using QbD principles in HPLC. In: Grushka E, Grinberg N, editors. Modeling of HPLC methods using QbD in HPLC Advances in Chromatography, vol. 53. CRC Press; 2016. [54] Xiao KP, Xiong Y, Liu FZ, Rustum AM. Efficient method development strategy for challenging separation of pharmaceutical molecules using advanced chromatographic technologies. J Chromatogr A 2007;1163:145–56. [55] Debrus B, Lebruna P, Ceccato A, Caliaroc G, Rozet E, Nistor I, Opreand R, Ruperez FJ, Barbas C, Boulanger B, Hubert P. Application of new methodologies based on design of experiments, independent component analysis and design space for robust optimization in liquid chromatography. Anal Chim Acta 2011;691(1–2):33–42. [56] Gowik P. The validation of methods for regulatory purposes in the control of residues. J Chrom A 2009;1216(46):8051–8. 13. [57] Shabir GA. Validation of high-performance liquid chromatography methods for pharmaceutical analysis Understanding the differences and similarities between validation requirements of the US Food and Drug Administration, the US Pharmacopeia and the International Conference on Harmonization. J Chromatogr A 2003;987:57–66. [58] Swartz ME, Krull IS. Handbook of analytical validation. CRC Press; 2012. [59] Chan CC, Lam H, Zhang X-M, editors. Practical approaches to method validation and essential instrument qualification. Wiley-Interscience; 2010. [60] Ermer J, Nethercote PW, editors. Method validation in pharmaceutical analysis: A guide to best practice. 2nd ed. Wiley; 2014. [61] Yadav Nand K, Sanju N, Ashish R, Gajanand S, Sarwar B, Katare Om P. QbD-based development and validation of a stability-indicating HPLC method for estimating Ketoprofen in bulk drug and Proniosomal vesicular system. J Chromatogr Sci 2016;54 (3):377–89. [62] Pantovic J, Malenovic A, Vemic A, Kostic N, Medenica M. Development of liquid chromatographic method for the analysis of dabigatran etexilate mesilate and its ten impurities supported by quality-by-design methodology. J Pharm Biomed Anal 2015;111:7–13. [63] Orlandini S, Pasquini B, Stocchero M, Pinzauti S, Furlanettoa S. An integrated quality by design and mixture-process variable approach in the development of a capillary electrophoresis method for the analysis of almotriptan and its impurities. J Chromatogr A 2014;1339:200–9.

839

840

CHAPTER 31 Validation of liquid chromatographic methods

[64] ICH Development and Manufacture of Drug Substances, editor. International Conference on harmonization of technical requirements for registration of pharmaceuticals for human use, ICH harmonized tripartite guideline; 2012. [65] ICH. International Conference on harmonization of technical requirements for registration of pharmaceuticals for human use, ICH harmonized tripartite guideline, Regulatory Considerations for Pharmaceutical Product Lifecycle Management; 2019. [66] Harrington B, Barnett K, Chesnut S, Clayton N, Cohen M, Ensing J, Graul T, HannaBrown M, Morgado J. Defining the analytical target profile. Pharm Eng 2018;52–7. [67] Harrington B, Barnett K. Experimental conditions define the ATP inference space. Pharm Eng 2018;66–71. [68] ICH. International Conference on harmonization of technical requirements for registration of pharmaceuticals for human use, Q14 Analytical Procedure Development (guideline under development), http://www.ich.org. [69] Kopp J, Zauner F, Pell A, Hausjell J, Humer D, Ebner J, Herwig C, Spadiut O, Slouka C, Pell R. Development of a generic reversed-phase liquid chromatography method for protein quantification using analytical quality-by-design principles. J Pharm Biomed Anal 2020;188, 113412. [70] Sharma G, Thakur K, Raza K, Katare OP. Stability kinetics of fusidic acid: development and validation of stability indicating analytical method by employing analytical quality by design approach in medicinal product(s). J Chromatogr B Analyt Technol Biomed Life Sci 2019;1120:113–24. [71] Mitrovic M, Protic A, Malenovic A, Otasevic B, Zecevic M. Analytical quality by design development of an ecologically acceptable enantioselective HPLC method for timolol maleate enantiomeric purity testing on ovomucoid chiral stationary phase. J Pharm Biomed Anal 2020;180, 113034. [72] Zacharis CK, Vastardi E. Application of analytical quality by design principles for the determination of alkyl p-toluenesulfonates impurities in Aprepitant by HPLC. Validation using total-error concept. J Pharm Biomed Anal 2018;150:152–61. [73] Deidda R, Avohou H, Baronti R, Davolio PL, Pasquini B, Del Bubba M, Hubert C, Hubert P, Orlandini S, Furlanetto S. Analytical quality by design: development and control strategy for a LC method to evaluate the cannabinoids content in cannabis olive oil extracts. J Pharm Biomed Anal 2019;166:326–35. [74] Radic I, Runje M, Babic S. Development of an analytical method for the determination of pimavanserin and its impurities applying analytical quality by design principles as a riskbased strategy. J Pharm Biomed Anal 2021;201, 114091. [75] Teng J, Zhu C, Lyu J, Pan L, Zhang M, Zhang F, Haoxiang W. Analytical lifecycle management (ALM) and analytical quality by design (AQbD) for analytical procedure development of related substances in tenofovir alafenamide fumarate tablets. J Pharm Biomed Anal 2022;207, 114417. [76] Bandopadhyay S, Sarwar Beg OP, Katare TS, Singh B. Integrated analytical quality by design (AQbD) approach for the development and validation of bioanalytical liquid chromatography method for estimation of valsartan. J Chromatogr Sci 2020;58(7):606–21. [77] Rmandic M, Malenovic A. Chaotropic chromatography method development for the determination of aripiprazole and its impurities following analytical quality by design principles. J Sep Sci 2020;43(16):3242–50. [78] Gurrala S, Raj S, Cvs S, Anumolu P. Quality-by-design approach for chromatographic analysis of metformin, Empagliflozin and Linagliptin. J Chromatogr Sci 2021.

References

[79] Abreu JC, Mahr AG, do Lago CL. Stability-indicating method development for quantification of bromopride, its impurities, and degradation products by ultra-high performance liquid chromatography applying analytical quality by design principles. J Pharm Biomed Anal 2021;205, 114306. [80] Renai L, Scordo CVA, El Ghadraoui A, Santana-Viera S, Rodriguez JJS, Orlandini S, Furlanetto S, Fibbi D, Lambropoulou D, Del Bubba M. Quality by design optimization of a liquid chromatographic-tandem mass spectrometric method for the simultaneous analysis of structurally heterogeneous pharmaceutical compounds and its application to the rapid screening in wastewater and surface water samples by large volume direct injection. J Chromatogr A 2021;1649, 462225. [81] Peraman R, Bandi J, Kondreddy VK, Kalva B, Kothakota SG, Paritala J, Nagappan K, Yirgamreddy PR. Analytical quality by design approach versus conventional approach: development of HPLC-DAD method for simultaneous determination of etizolam and propranolol hydrochloride. J Liq Chromatogr Relat Technol 2021;44(3–4): 197–209. [82] Gundala A, Bharathi K, Prasad K. Analytical quality by design approach in rp-hplc method development for the assay of pitavastatin in pharmaceutical dosage forms. Eur J Biomed Pharm Sci 2018;5(5):1–11. [83] Alkhateeb FL, Wilson I, Maziarz M, Rainville P. Ultra high-performance liquid chromatography method development for separation of formoterol, budesonide, and related substances using an analytical quality by design approach. J Pharm Biomed Anal 2021;193, 113729. [84] Prajapati P, Jayswal K, Shah S. Application of quality risk assessment and DoE-based enhanced analytical quality by design approach to development of chromatography method for estimation of combined pharmaceutical dosage form of five drugs. J Chromatogr Sci 2021;59(8):714–29. [85] Jiaxin D, Chang Y, Zhang X, Changqin H. Development of a method of analysis for profiling of the impurities in phenoxymethylpenicillin potassium based on the analytical quality by design concept combined with the degradation mechanism of penicillins. J Pharm Biomed Anal 2020;186:11330. [86] Prajapati P, Patel A, Shah S. DoE-based analytical quality risk management for enhanced AQbD approach to economical and eco-friendly RP-HPLC method for synchronous estimation of multiple FDC products of antihypertensive drugs. J Chromatogr Sci 2021;60:786–99. [87] Saini S, Sharma T, Patel A, Kaur R, Tripathi SK, Katare OP, Singh B. QbD-steered development and validation of an RP-HPLC method for quantification of ferulic acid: rational application of chemometric tools. J Chromatogr B Analyt Technol Biomed Life Sci 2020;1155, 122300. [88] Arce MM, Sanllorente S, Ruiz S, Sanchez MS, Sarabia LA, Ortiz MC. Method operable design region obtained with a partial least squares model inversion in the determination of ten polycyclic aromatic hydrocarbons by liquid chromatography with fluorescence detection. J Chromatogr A 2021;1657:462. [89] Jackson P, Borman P, Campa C, Chatfield M, Godfrey M, Hamilton P, Hoyer W, Norelli F, Orr R, Schofield T. Using the analytical target profile to drive the analytical method lifecycle. Anal Chem 2019;91(4):2577–85. [90] Ermer J. Analytical target profile: establishment of precision requirements for assay. J Pharm Biomed Anal 2018;160:73–9.

841

842

CHAPTER 31 Validation of liquid chromatographic methods

[91] Arce MM, Ruiz S, Sanllorente S, Ortiz MC, Sarabia LA, Sanchez MS. A new approach based on inversion of a partial least squares model searching for a preset analytical target profile. Application to the determination of five bisphenols by liquid chromatography with diode array detector. Anal Chim Acta 2021;1149:338217. [92] Borman P, Campa C, Delpierre G, Hook E, Jackson P, Kelley W, Protz M, Vandeputte O. Selection of analytical technology and development of analytical procedures using the analytical target profile. Anal Chem 2022;94(2):559–70. [93] Ermer J, Aguiar D, Boden A, Ding B, Obeng D, Rose M, Vokrot J. Lifecycle management in pharmaceutical analysis: how to establish an efficient and relevant continued performance monitoring program. J Pharm Biomed Anal 2020;181:11305. [94] Oliva A, Martinez ML. Application of capability indices and control charts in the analytical method control strategy. J Sep Sci 2017;40(15):3046–53. [95] Hubert P, Nguyen-Huu JJ, Boulanger B, Chapuzet E, Chiap P, Cohen N, Compagnon PA, Dewe W, Feinberg M, Lallier M, Laurentie M, Mercier N, Muzard G, Nivet C, Valat L. Harmonization of strategies for the validation of quantitative analytical procedures. A SFSTP proposal—Part I. J Pharm Biomed Anal 2004;36:579–86. [96] Burgess C, Curry P, LeBlond D, Kovacs E, Martin G, McGregor P, Nethercote P, Pappa H, Weitzel MLJ. Fitness for use: decision rules and target measurement uncertainty. Pharmacopeial Forum 2016;42:2. [97] Kovacs E, Ermer J, McGregor P, Nethercote P, LaBrutto R, Martin G, Pappa H. Analytical control strategy. Pharmacopeial Forum 2016;42(5). [98] The Analytical Procedure Life cycle . Pharmacopeial Forum 2020;46(5). Proposed New USP General Chapter. [99] Weitzel MLJ. The estimation and use of measurement uncertainty for a drug substance test procedure validated according to USP . Accred Qual Assur 2012;17: 139–46. [100] Hubert C, Houari S, Rozet E, Lebrun PL, Hubert PH. Towards a full integration of optimization and validation phases: an analytical quality by design approach. J Chrom A 2015;1395:88–98. [101] BIPM. Evaluation of measurement data—guide to the expression of uncertainty in measurement (GUM)., 2008, www.bipm.org. [102] Ellison SLR, Williams A, editors. Eurachem/CITAC guide CG4: Eurachem/CITAC, quantifying uncertainty in analytical measurement. 3rd ed. Eurachem; 2012. www. eurachem.org. [103] Weitzel J, Forbes R, Snee R. The use of the analytical target profile in the life cycle of an anaytical procedure. J Validation Technol 2014;XX(4). [104] Feinberg M. Validation of analytical methods based on accuracy profiles. J Chrom A 2007;1158:174–83. [105] Kadis R. Analytical procedure in terms of measurement (quality) assurance. In: De Bie`vre P, G€unzler H, editors. Validation in chemical measurements. Berlin: Springer-Verlag; 2005. p. 148–52. [106] Kringle R, Khan-Malek R, Snikeris F, Munden P, Agut C, Bauer M. A unified approach for design and analysis of transfer studies for analytical methods. Drug Inf J 2001;35:1271–88. [107] Chatfield MJ, Borman PJ. Acceptance criteria for method equivalency assessments. Anal Chem 2009;81:9841–8.

References

[108] Williams RL, Abernethy DR, Koch WF, Hauck WW, Cecil TL. Performance-based monographs. Pharmacopeial Forum 2009;35:765–70. [109] Nickerson B, editor. Sample preparation of pharmaceutical dosage forms: challenges and strategies for sample preparation and extraction. Springer; 2011. [110] Mire-Sluis A, Ramnarine E, Siemiatkoski J, Weese D, Swann P, O’Keeffe R, Kutza J, Edwards J, McLeod LD. Practical applications of quality risk management. BioProcess Int 2010;8(3):20–32. [111] Monks K, Molna´r I, Rieger H-J, Boga´ti B, Szabo´ E. Quality by design: multidimensional exploration of the design space in high performance liquid chromatography method development for better robustness before validation. J Chrom A 2012;1232:218–30. [112] Vander Heydena Y, Jimidarb M, Hunda E, Niemeijerb N, Peetersb R, Smeyers-Verbeke J, Massart DL, Hoogmartens J. Determination of system suitability limits with a robustness test. J Chrom A 1999;845:145–54.

843

Index Note: Page numbers followed by f indicate figures and t indicate tables.

A Abraham descriptors, 265–267, 266f, 422–423 Absolute molar mass, 521–528 Absorption detection, 665 Accuracy, defined, 826 Acetonitrile (ACN), 284–286, 402–405 Acid–base secondary equilibria, 122–124 Acid dissociation constants, 343–344 Active pharmaceutical ingredients (APIs), 299 Additives chaotropic anions, 297–300 column performance measurement using, 131–132 deep eutectic solvents (DESs) as, 292 ionic liquids (IL) as, 288–291 kosmotropic additives, 300–308 modeling, 615–616 surfactant, 308–314 Adequate sample preparation, 158 Adsorption chromatography. See Normal-phase chromatography (NPC) Adsorption isotherms, 608–609 data determination, 609–610 Frontal analysis, 610 inverse method, 610 Langmuir adsorption isotherm, 608–609 Adsorption kinetics, 231–233 Adsorption model, 606–610 adsorption isotherms (see Adsorption isotherms) band shape dependence, 606–608 Affinity chromatography, 346–347, 442 components, 540–542 ligands used in, 541t scheme for application, 539, 540f Agglomerated ion-exchangers, 482 Alcohols, 284–288 Aliphatic amines, 129 Alkyl ethoxylates, 309 Alkylphosphates, 309 Alkylsulfates, 308 Ambient ionization mass spectrometry (AIMS), 689 Amines, 284–288 Amino acid analysis, 808, 809f Amperometry, 493

Amphoteric surfactants, 308, 310 Amylose tris(3,5-dimethylphenylcarbamate (ADMPC), 416–417 Analytical affinity chromatography, 549–551, 550f Analytical target profile (ATP), 828–833 Anion-exchange columns, 479–481t, 482–483, 483f, 485–487t Anion exchange eluents, 473, 474t, 475f Anionic surfactants, 308–309 Anisotropic etching, 655f Apparent formation constants, 344–346 Aptamers, 552–553 Area height method, 50 Artificial neural networks (ANNs), 799 Atmospheric pressure chemical ionization (APCI), 186, 639, 687–688 Atmospheric pressure ionization (API), 684–689 Atmospheric pressure ionization-electrospray ionization (API-ESI), 215–218 Atmospheric pressure photoionization (APPI), 663, 663f, 688–689 Average surface hydrophobicity, 448 Axcend Focus LC, 667f, 668

B Band broadening, 516–519 Band shape, 606–608 Batch-process chromatography, numerical optimization of, 612, 612f β blockers, 159–160 Bi-Langmuir adsorption isotherm, 418 Bi-Langmuir model, 609, 618 Bioaffinity chromatography, 542–543 Biomimetic affinity chromatography, 546–547 Biomimetic liquid chromatography, 354–356 Biopartition constants, 353–354 Biopartitioning micellar liquid chromatography, 137 Bispecific antibodies (BisAb), 455 Boltzmann constant, 753–754 Bonded-phase chromatography (BPC), 442 Bovine serum albumin (BSA), 195 Box–Behnken design (BBD), 292 Brij-35, 133–135 Buffer system, 124

845

846

Index

C Cannabinoids, 218 Capillary electrochromatography (CEC), 625, 807–808 applications, 642–643 injection, 629 instrumentation, 627–640 mass spectrometry detection, 637–640 miniaturized systems, 640–642 principles, 626–627 stationary phases, 630–636 Capillary electrophoresis (CE), 760 Capillary isotachophoresis (cITP), 760, 768, 786 Carbohydrate analysis, 241 Carbonate type selectivity, 483 Carboxylic acids, 284–288 Cation exchange columns, 484, 485–487t Cation exchange eluents, 473–474 Cationic surfactants, 128–129 Cellulose tris(3,5-dichlorophenylcarbamate) (CDCPC), 392–394 Cellulose tris(3,5-dimethylphenylcarbamate) (CDMPC), 392–394 Cetyl trimethyl ammonium bromide (CTAB), 309 Chaotropic anions, 297–300 Chaotropicity, 130 Charge detector, 492–493, 492f Charged residue model (CRM), 686, 686f Chelation ion chromatography (CIC), 469–470 Chemical composition distribution (CCD), 531 Chemically induced dynamic nuclear polarization (CIDNP), 758 Chemical similarity, 808–809 Chemometric approaches, 733–734 Chip-based HPLC instruments, 666–667 Chiral selectivity, 158 Chiral selector (CS), 383–400, 403, 408, 416–417, 422 Chiral separation, 4 Chiral solvating agents (CSA), 389 Chiral stationary phase (CSP), 154–156, 158, 383–384, 389, 395, 398–400, 403, 405–406, 416–417, 603–604 computation/molecular modeling, 418–423 in early liquid chromatographic (LC) enantioseparations, 384t Choline chloride (ChCl), 292 Chromatographic dilution, 178–179 Chromatographic immunoassay, 544, 545f Chromatographic mode, 123–124, 128–129, 227 hydrophilic-interaction chromatography (HILIC), 227–228, 231

selection, 267–268 Circular dichroism (CD), 779 Clinical analysis, 203–215, 209–211t Closed-loop recycling (CLR) chromatography concept, 584–585 optimization, 595 Colloidal crystal, 167 Column, 4–8, 66 diameter, 478 efficiency, 49–50 hardware and design, 38–42 hydrodynamic parameters, 51–52 ion chromatography, 475–484 model, 605–606 packing, 42–46, 44t peak symmetry, 51 plate height, 49–50 resolution, 50–51 retention and retention volume, 48 retention factor, 48–49 screening, priority of, 329–330 selectivity/separation factor, 49 testing and evaluation, 53–56 Column mass-balance model, 612–613 Column plate number, 325 Compact solvent delivery systems, 647–650 Comparative molecular field analysis (CoMFA), 798–805 Comparative Molecular Similarity Indices Analysis (CoMSIA) methods, 798 Competitive binding immunoassays, 544–545 Compound descriptors, 338–343, 339t, 342f Computer-assisted interpretive optimization, 274–275 Computer numerical control (CNC) micromilling, 767 Computer simulation, 332 Conductimetric detection, 490–492 Conductivity detectors, 69–70 Connective tubing and fittings, 71–72 Connolly solvent-excluded volume, 112 Continuous simulated moving bed chromatography, 586–595 Control strategy, 835 Core performance tests, 333 Core-shell particles, 24 Co-solvents and mobile phase additives, in HPLC chaotropic anions, 297–300 deep eutectic solvents (DESs), 292–297 fluorinated ion-pairing agents, 284–288 ionic liquids (IL), 288–291 kosmotropic ions, 300–308

Index

reversed-phase (RP) mode, 283 surfactant additives, 308–314 Critical micelle concentration (CMC), 132, 310, 311f, 312–313 CSP. See Chiral stationary phase (CSP) Cyanopropyl silica, 236 Cyclodextrins (CD), 237, 391, 633–634 Cyclofructan-based chiral selectors, 397 Cyclofructans (CFs), 237–238

D Data dependent acquisition (DDA) mode, 701 Data independent acquisition (DIA) mode, 701 Deep eutectic solvents (DESs), 292–297 Desorption electrospray ionization (DESI), 689, 690f Detection limit (DL), 825 Detection sampling frequency, 168 Detectors, 67–70, 185–186 Differential mass-balance equation, 22 Differential refractometer (DRI), 521, 521t Differential scanning calorimetry (DSC), 406, 417–418 Dilute-and-shoot technique, 162–163 Diol silica, 236 Direct liquid electron interface (DLI), 682, 683f Dispersive mechanisms, 18–19, 18f Displacement effect, 580–581, 608–609 DNA affinity chromatography, 543 Double-gradient elution, 274 Droplet evaporation (EV), 751–753 Dry column packing, 42–43 Dwell volume, 379–381 Dye-ligand affinity chromatography, 546–547 Dynamically modified ion-exchangers, 482 Dynamic chelating chromatography, 139

E Echo peak calibration, 722–723 Efficient HPLC method development, 371 Electrochemical detection, 492–493, 664 Electrochemical detectors, 10–11, 69 Electronic circular dichroism (ECD) spectroscopy, 414–416 Electronic records, 334 Electron ionization (EI), 681–682 Electroosmotic flow (EOF), 625–626, 630–631, 633–634 Electroosmotic pumps, 649–650, 649f Electrospray ionization (ESI), 186, 494–495, 637–638, 663, 684–687, 685f

Electrostatic ion chromatography (EIC), 128–129, 470 Eluents anion exchange, 473, 474t, 475f cation exchange, 473–474 generators, 63, 484–489, 488f, 489t suppressor, 490 Eluotropic scales, 254 Elution strength, 251–253 assessment, 254–257 Embedded SCAffold RemovinG Open Technology (ESCARGOT), 756–758 Empirical optimization, 611 Enantiomer elution order (EEO), 385, 403, 408 Enantioselective liquid chromatography chiral selector (CS), 383–398, 386t inert carrier, 398–401 mobile phase, 401–405 molecular chirality, 383–384 recognition mechanisms, 384–385 separation process, 405–414 Engelhardt column test, 54–55 Enthalpy-entropy compensation plot, 100 Entropy-controlled process, 512–514 Environmental analysis, 203, 205–206t Enzyme purification, 542–543 Equilibrium-dispersive (ED) model, 579, 605–606 Equilibrium process, 514–516 Evaporative light-scattering detector (ELSD), 451–452, 531–532, 532f Evosep One system, 194 Excess Rayleigh factor, 525 Exothermodynamic relationships, 98–99 Exponentially modified Gaussian (EMG) method, 50 External cavity–quantum-cascade lasers (EC-QCLs), 736–737 External column porosity, 231–232 Extracolumn band broadening, 179–180 Extra-column effects, 519 Extract enrichment (EE) process, 592, 593f

F Fast atom bombardment (FAB), 637 Fixed solvent modulator (FSM), 568–569 Fixed-wavelength detector, 69 Flip-flop chromatography, 582 Fluorescence detection, 664–665 Fluorescence detector (FLD), 69, 451–452 Fluorinated ion-pairing agents, 284–288 Food analysis, 198–203, 199–201t Formal mechanistic retention models, 94–95

847

848

Index

Fourier-transformed ion cyclotron resonance (FTICR-MS), 696–697, 697f Fourier transform mass analyzers, 696–699 Fraction-feedback (FF) process, 592, 593f Free induction decay (FID), 748 Frontal affinity chromatography (FAC), 549, 550f Frontal analysis, 604, 609–610

G Gas lasers, 684 Gaussian function, 29–30 Gel filtration chromatography (GFC), 510–511 Gel permeation chromatography (GPC), 2, 510–511, 779 General elution problem, 369, 370f General rate model (GRM), 21–25, 579 for core-shell particles, 24 moment analysis, 24–25 for monolith columns, 23–24 Generic high-pressure injection valve, 64, 65f Generic separation, 371 Gibbs-Helmholtz relationship, 99 Giddings’ coupling theory, 517 Giddings–Eyring model, 33 Giddings plate height equation, 30–32 Global polarity estimators, 254–256 for solvent mixtures, 256 Gradient chromatography, 582–584, 583f Gradient elution, 32–33, 63–64, 328, 617–618 experimental conditions, effects of, 371–378 vs. isocratic elution, 369, 371–373 method development, 378–379, 378f, 380f problems associated with, 379–381 Gradient polymer elution chromatography (GPEC), 531 Gradient process, 590–591, 591f Graphene based nanomaterials, 634

H Hadamard NMR spectroscopy, 776, 786 Half peak height method, 50 Hansen parameters, 265 Hard polymers, 652–653 Hexadecyl trimethyl ammonium bromide (HTAB), 309 1,1,1,3,3,3-hexafluoro-2-methyl-2-propanol (HFTB), 288 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP), 288 High-performance immunoaffinity chromatography (HPIAC), 544 High-performance liquid chromatography (HPLC), 2–3, 37, 383–384, 442

columns, 5–8, 6t co-solvents and mobile phase additives (see Cosolvents and mobile phase additives, in HPLC) detectors, 10–11 enantioseparations, 390–391 equipment, 8–10 modes and techniques, 4 particles and column packing, 5–7 theory and practice, 3–5 High-pressure multiple-column packing, 46, 47f High-pressure pumps, 63 High-resolution mass analyzers Fourier transform mass analyzers, 696–699 Orbitrap, 697–699, 698f time-of-flight measurements, 695, 696f High-temperature superconducting (HTS) magnet, 780–782, 786 Hildebrand solubility parameter, 254–256 HILIC. See Hydrophilic interaction liquid chromatography (HILIC) Hofmeister series, 297, 304–306, 305f Holdup volume, 613–614 Hot embossing, 652–653 Hybrid micellar liquid chromatography, 135–137 Hydrodynamic parameters, 51–52 Hydrogen bond acceptor (HBA), 292 Hydrogen bond donor (HBD), 292, 813 Hydrophilic interaction liquid chromatography (HILIC), 97, 124, 253–254, 296–297, 799, 800–803t adsorption kinetics, 231–233 adsorption, thermodynamics of, 228–231 advantages, 228 applications, 239–242 chemically bonded phases, 236 chromatographic mode, 227–228, 231 drawback, 228 hydrophilic macromolecules bonded phases, 237–238 ion exchange, 237 mobile phases, 227, 239 overview, 227–228 polar compounds, separation of, 227 polar stationary phases, 234–235f principles, 228–233 silica gel, 235–236 stationary phases, 233–239 surface-confined ionic liquids stationary phases, 238–239 zwitterionic stationary phase, 237 Hydrophobic charge induction chromatographic (HCIC) resins, 450–451 Hydrophobic interaction, 159–160

Index

Hydrophobic interaction chromatography (HIC), 300–302, 306–307, 442–443 challenges, 456 detector, 451–452 elution gradient, 447–448 historical perspective, 443–444 limitations, 456 mobile phase, 442, 446–447 multi-dimensional separation platforms, 455–456 operating principles of, 444–452 operation of, 445–452 protein refolding, 454 solute characterization, 454–455 solute properties, 448 solute purification, 452–454 stationary phase, 442, 448–451 Hydrophobic metallochromic ligands, 139–140 Hydrophobic-subtraction model, 111–113, 113t Hydrophobic surfaces, 44 Hydrophylic interaction chromatography (HILIC), 402–405 Hydroxyl type selectivity, 483

I Identification parameter (IP), 734 Identification (ID) test, 825 Immobilized artificial membranes (IAMs), 354–355 Immobilized metal-ion affinity chromatography (IMAC), 548–549, 548f Immunoaffinity chromatography (IAC), 544–546 Immunoglobulin-binding proteins, 543 In-column ion pair (IP) formation, 284 Infrared spectroscopy (IR), 393, 414–416 mobile-phase components, 729, 730f off-line hyphenation, 730–732, 731f on-line hyphenation, 733 physical principles, 727–728 quantitative analysis, 728 Injection profiles, 614–615, 614f Injection technique, 166 In situ formed monolithic columns, 633–634 Intermediate precision, 826 Internal diameter (ID), 38 Inter-particle transport and dispersion mechanism, 21–22 Interphase model, 93–94 Interstitial reduced velocity, 231–232 Interstitial velocity, 231–232 In-tube solidphase microextraction (IT-SPME), 203 Inverse liquid chromatography, 348–349

Inverse method, 610 Inverse size-exclusion chromatography (ISEC), 358 Ion chromatography (IC), 799, 800–803t applications, 496–501 definitions, 465 eluents, 473–474 environmental applications, 501 industrial applications, 500 instrumentation, 475–496 two-dimensional, 495–496, 497f, 497–499t Ion evaporation model (IEM), 686, 686f Ion exchange, 237, 476 Ion exchange capacity, 476 Ion-exchange chromatography (IEC), 2, 442, 466–467, 468f Ion-exchanger matrix, 476, 478t Ion-exclusion chromatography (IEC), 467–469 Ionic liquids (IL), 130–131, 288–291 Ionic surfactants, 312–313 Ionic vs. hydrated radius, 468f Ion-interaction chromatography, RPLC column performance, enhancement of, 131–132 inorganic anions, separation of, 128–129 ionic liquids, 130–131 perfluorinated carboxylate anions and chaotropic ions, 130 reagents and operational modes, 126–128, 127f retention mechanism, 125–126 silanol effect, 129–130 Ionization energy (IE), of analytes, 688–689 Ion pair chromatography (IPC), 302–304 Ion-pair chromatography (IPC), 4, 125 Ion-pair reagents (IPRs), 616–617 Ion trap (IT) mass analyzers, 692, 693f Isobaric tags for relative and absolute quantitation (iTRAQ), 724 Isocratic batch elution mathematical modeling, 579–581 principle, 578, 578f Isocratic elution, 270–273, 369, 371–373 Isoeluotropic mixtures, 257–259 Isoselective gradients, 274 Isothermal titration calorimetry (ITC), 406, 417–418 Isotropic wet etching, 655, 655f

K Kamlet–Taft solvatochromic parameters, 264–265 Kinetic differentiation chromatography, 139 Kinetic measurements, 356–358

849

850

Index

Kinetic theories, of liquid chromatography. See Macroscopic kinetic theories; Microscopic kinetic theories Knox equation, 3 Kosmotropic ions, 300–308 Kozeny-Carman equation, 39

L Laboratory-assembled nano-liquid chromatography instrumentation, 185, 186f Laboratory notebook, 334 Langmuir adsorption isotherm, 608–609 Langmuir model, 582–584 Laplace-domain solution, 23 Larmor frequency, 753–754 Laser machining, 652–653 Lectins, 543 Legal and forensic analysis, 215–219, 216–217t Life cycle management, 835–836 Light Lab 3, 667f, 668 Lignosulfonates, 455–456 Limit test, 825 Linear free energy relationships (LFERs), 797 Linear free energy relationships, RPLC hydrophobic-subtraction model, 111–113 solvation parameter model, 102–111, 102t Linear gradient program, 371, 372f Linear ion traps (LITs), 692 Linearity, 826 Linear photodiode array (LDA), 451–452 Linear solvation energy relationships (LSER), 422–423 Linear solvation energy relationships (LSERs), 797 Linear solvent strength (LSS) gradient relationship model, 306–308 Linear solvent strength (LSS) model, 4, 268 Liquid chromatography-mass spectrometry (LC-MS), 11, 661, 679–680, 720 Liquid chromatography-nuclear magnetic resonance (LC-NMR) improvement strategies droplet evaporation (EV), 751–753 loop-storage mode, 750 solid-phase extraction (SPE), 750–751 stop-flow mode, 749–750 Liquid chromatography under critical conditions (LCCC), 779 Liquid-core waveguide (LCW), 739 Liquid-film linear driving force model, 19 Liquid–liquid chromatography, 338 Liquid-liquid extraction (LLE), 158 Liquid partition chromatography, 2 Liquid–solid chromatography (LSC), 89

Longitudinal diffusion, 232 Loop-storage mode, 750 Low-angle static light scattering (LALS) detector, 526–527 Low-resolution mass analyzers ion trap (IT) mass analyzers, 692, 693f quadrupole mass analyzers, 690–691, 691f Lumped kinetic model, 19–21, 21f Lumped pore diffusion model, 25

M Machine vision (MV) devices, 751–753 Macrocyclic glycopeptide-based chiral stationary phases, 149 Macroporous polymer particles, 94 Macroscopic kinetic theories, 18–27 equivalence of, 25–26 general rate model, 21–25 lumped kinetic model, 19–21, 21f lumped pore diffusion model, 25 microscopic kinetic model vs., 33–34 of non-linear chromatography, 26–27 van Deemter plate height equation, 20–21 Mass analyzers high-resolution, 692–699 low-resolution, 690–692 Mass flux, 22 Mass spectrometry (MS), 70, 156–159, 186–187, 494–495, 637–640, 743–744 Mass transfer in hydrophilic interaction chromatography (HILIC), 231, 233 in packed column, 18–19, 18f Mass-transfer, 17–19, 231, 233 Mass transfer coefficient, 25–26, 357–358 Mass transfer kinetics, 356–357 Matrix-assisted laser desorption ionization (MALDI), 639, 682–684 Matrix match calibration, 722 McGowan algorithm, 797 Membrane evaporation device, 569, 570f Membrane permeation, 354–356 Metabolic isotopic labelling, 723 Metabolites, 708–709, 709f Metal complexation, RPLC metal ions, determination of, 138–140 organic compounds, determination of, 140 Method development documentation, 334–335 goals, 324 gradient elution, 378–379, 378f, 380f in practice, 329–332

Index

prevalidation, 332–333 report, 334 structured approach, 325–328 triangle, 271–272, 271f validation, 333–334 Method development, NPC aim, 82 example, 84–85 mobile phase selection, 82–84 thin-layer chromatography, 82 Method document, 333–334 Method specificity, 825 Method validation characteristics, 823–824t, 825 control strategy, 835 guidelines, 822–824t life cycle management, 835–836 operating conditions, understanding and identify development of, 834 purpose, 821 risk assessments, 834 technique selection and method development, 833–834 Methylene selectivity, 157 Meyerhof equation, 526 Micellar liquid chromatography (MLC), 255, 312–314, 355–356, 807–808 hybrid liquid chromotography, 135–137 microemulsion liquid chromotography, 137–138 secondary equilibrium in mobile phase, 132–135 three-phase systems in, 315f Microchip-based CEC system, 640 Microchips absorption detection, 665 analyte injection, 658–659 electrochemical detection, 664 fluorescence detection, 664–665 HPLC instruments, 666–667 materials, 651–656 microfluidic LC coupled to mass spectrometry, 661–663 portable HPLC systems, 667–668 refraction detection, 665 separation channels and beds, 656–658 stationary phases, 656, 657t two-dimensional LC, 659–661 Microchromatographic techniques, 72 Microdetectors, LC-NMR improvements faster NMR experiments, 775–778 general features, 753–755 microsaddle detectors, 755–756 parallel NMR detection, 771–775 planar spiral microcoil detectors, 761–764

solenoid microcoil detectors, 756–761 stripline detectors, 765–770 Microemulsion liquid chromatography (MELC), 137–138, 356, 807–808 Microextraction by packed sorbent (MEPS), 214 Microfabrication technologies, 651–656 Microfluidic analytical techniques, 178–180 Microfluidic LC coupled to mass spectrometry, 661–663, 662–663f Microfluidic pump systems, 182–184 Microfluidic water-assisted trap (M-WAFT) focusing system, 203, 204f Micromachining, 652 Microsaddle detectors, 755–756 Microscopic kinetic theories, 27 Giddings plate height equation, 30–32 vs. macroscopic kinetic theories, 33–34 non-linear chromatography, Monte Carlo simulations of, 32–33 stochastic–dispersive model, 30 stochastic model, 27–30 Microsolenoid, general scheme of, 769f Micro total chemical analysis systems (μTAS), 642 Miniaturized chromatographic techniques, 214 Miniaturized HPLC, sample injection in, 650–651, 650t, 651f Mobile phase composition, optimization of chromatographic mode selection, 267–268 combined mobile phases, 276–277 computer-assisted interpretive optimization, 274–275 modifier content, 268–269 systematic trial-and-error optimization, 30–33 Mobile-phase reservoir, 62–63 Mobile phase selection, 82–84 “Modicon” concept, 591–592, 592f Molar mass distribution (MMD), 509–510, 510f Molecular diffusion, 357–358 Molecular dynamics (MD) simulations, 91, 419, 422 Molecularly imprinted polymers (MIPs), 552, 553f, 643, 644f Molecular mass distributions (MMDs), 779 Molecular mechanics (MM), 419 Molecular modeling, 797 Moment analysis, 24–25 Monoisotopic molecular mass, 694, 694t Monolith columns, 7, 23–24, 37, 157, 159–163 Monolithic packed CEC columns, 630f, 633–634 Monte Carlo simulations, of non-linear chromatography, 32–33 MSDial software program, 708

851

852

Index

Multi-angle static light scattering (MALS) detection, 521, 521t, 527–528, 528–530f Multi-column parallel approach, 159 Multi-column solvent gradient process (MCSGP), 594 Multidetector separations, 529–532 Multidimensional chromatography, 197 Multidimensional liquid chromatography (MD-LC), 563 fundamentals, 565–566 instrumental set-up and data analysis, 566–570 Multifunctional surfactants, 308 Multiple heart-cutting analysis, 164 Multiple injections in a single experimental run (MISER), 167–168 Multiple linear regression (MLR), 340, 799 Multiple reaction monitoring mode (MRM), 700–701 Multiplexing, 157 Multiwalled carbon nanotubes (MWCNTs), 202

N Nano-liquid chromatography applications, 187–219 capillary column preparation, 181–182 environmental analysis, 203, 205–206t food analysis, 198–203, 199–201t instrumentation, 182–187 legal and forensic analysis, 215–219, 216–217t pharmaceutical and clinical analysis, 203–215, 208–211t proteins and peptide analysis, 187–198, 189–193t stationary phases (SPs), 180–181 Nano-volume injection, 184–185 Nano-well-mediated fractionation system, 197–198 Nelder–Mead simplex algorithm, 613 Nitrogen lasers, 684 Nitroimidazoles, 202–203 Nonafluoropentanoic acid (NFPA), 286 Non-competitive immunoassays, 545–546 Noncovalent stationary-phase coatings, 634–635 Non-ionic surfactants, 309 Non-linear chromatography, 91 kinetic theories, 26–27 macroscopic kinetic theory, 26–27 Monte Carlo simulations, 32–33 Non-suppressed conductivity, 490 Non-target analysis, 707–708 challenge, 712 metabolites, 708–709, 709f Non-uniform sampling nuclear magnetic resonance (NUS NMR), 778

Normal bore chromatography, 72 Normal-phase chromatography (NPC) displacement process, 76–77 method development, 82–85 problems, 85–86 retention, 76 separations, 76 Normal-phase liquid chromatography (NPLC), 253–254 Nuclear magnetic resonance (NMR) spectrometry hyphenated systems, general features of, 743–744 on-flow LC-NMR limitations, 749–778 on-flow LC-NMR operation mode, 745–749, 746f Nuclear magnetic resonance (NMR) spectroscopy, 389, 414–416 Nuclear Overhauser effect (NOE), 416 Nucleic acids, 543 Numerical injection-volume optimization, 613 Numerical optimization, 611–613

O Octadecylsilane (ODS), 631–632 Octanol–water partition constant, 350–352 Off-line comprehensive LC (off-line LC x LC), 564 Off-line heart-cutting (LC/LC) technique, 564 Off-line hyphenation infrared spectroscopy, 730–732, 731f Raman spectroscopy, 732 Oil-in-water (O/W) microemulsion, 137 Omeprazole, enantiomeric separation of, 619, 619f Omics, QSRR applications in, 805–807 One-dimensional liquid chromatography (1D-LC), 563 One standard per substance class, 723 On-line comprehensive LC (on-line LC x LC), 564 On-line heart-cutting (LC-LC) technique, 564 On-line hyphenation infrared spectroscopy, 733 Raman spectroscopy, 733–734, 739–740 On-line viscometry, 522–525 Open-tubular CEC system, 630f, 634–636 Open tubular (OT) columns, 195–196 Open-tubular liquid chromatography (OTLC), 178 Orbitrap mass analyzers, 697–699, 698f Organic-based monolithic columns, 160 Organic monoliths, 399

P Packed columns, 188 mass transfer in, 18–19, 18f particle-packed columns, 630–632

Index

in situ formed monolithic columns, 633–634 Parahydrogen-induced polarization (PHIP), 778 Parallel reaction monitoring (PRM), 701 Partial least squares (PLS) regression, 734, 799 Particle-packed CEC columns, 630–632, 630f, 632f Particle size, 477f, 478 Partition constants, 338 Partition-displacement retention model, 92–93 Partitioning mechanism, HILIC, 228 Peak parking method, 357 Peak symmetry, 51 Pentadecafluorooctanoic acid (PDFOA), 284–286 Peptide analysis, 187–198, 189–193t Per aqueous liquid chromatography (PALC), 231 Perfluorinated carboxylate anions, 130 Perfluorinated carboxylic acids, 284, 286–287 Pharmaceutical analysis, 203–215, 208t pH measurements, 124 Photodiode array, 69 Photometric detection, 493–494 Photoresists, 654 Picoflow liquid chromatography-mass spectrometry (picoLC-MS) system, 196 Piston pumps, 647–648 Planar spiral microcoil detectors, 761–764 Planck constant, 753–754 Plasma-based direct analysis in real-time (DART), 689, 690f Plate height, 49–50 Plate height equation, 21–22 Plate number (N), 49 Poisson distribution, 28 Polarity scales, 254 Polar stationary phases, HILIC, 234–235f Polydimethylsiloxane (PDMS), 653–654 Polymer coated ion-exchangers, 482 Polymeric monolithic columns, 160 Polymer-shielded dye-affinity chromatography, 547 Poly-methyl methacrylate (PMMA), 779 Polynucleotides, 543 Polysaccharide phenylcarbamates, 392–393 Polystyrene (PS), 779 Pore size distribution, 358 Porous graphitic carbon (PCG), 94, 286 Porous shell micro-pillar array column, 657–658, 658f Portable high-performance LC systems, 667–668, 668t Post-column apparatus, 66–67 Post-column immunodetection, 546 Post-column internal standard infusion, 723 Post-column reaction detection, 494

Post-detection eluent processing, 71 “Powerfeed” operation, 591–592, 592f Precision, 826 Pre-column apparatus, 65 Preparative liquid chromatography, 72 adsorption model, 606–610 column model, 605–606 equilibrium-dispersive (ED) model, 605–606 process optimization, 610–618 Principal-component analysis (PCA), 734–736 Probability density function, 28 Procedure Life Cycle approach, 826–827 Process optimization, 610–618 Protein binding constants, 346–347 Protein precipitation (PP), 158 Proteins analysis, 187–198, 189–193t “Pseudo-enantiomeric” behavior, 395–396 Pulsed amperometric detection (PAD), 493 Purnell equation, 3

Q Quadrupole mass analyzers, 690–691, 691f Quality by design (QbD) principles, 323–324, 827 Quantification procedure advantages and challenges, 717, 717t analytical approach, 718, 718f calibration, 719 data analysis and reporting, 720–722, 721f LC–MS analysis, 720 matrix effects, 721–722 sample collection, 718 sample preparation, 719 validation, 719 Quantitation limit (QL), 825–826 Quantitative affinity chromatography, 549 Quantitative structure-activity relationship (QSAR) modeling, 814, 815f Quantitative structure–retention relationships (QSRR), 796f, 814–815 data sets, 795–796 definition, 795 mathematical models, 796–797 omics, applications in, 805–807 other chromatographic techniques, applications in, 807–808 quantitative structure-activity relationship (QSAR) modeling, 814, 815f retention prediction, in RPLC, HILIC, and IC, 798–799, 800–803t similarity, selectivity, and specificity (3S), 808–809, 809f stationary phases, characterization of, 809–814, 810–812t, 813f

853

854

Index

Quantitative structure–retention relationships (QSRR) (Continued) structural parameters, 797–798 three-dimensional quantitative structurebiological activity relationship (3DQSAR) method, 799–805, 805f Quantum-cascade lasers (QCLs), 728–729, 736–737 Quantum mechanical (QM) method, 419

R Raman spectroscopy off-line hyphenation, 732 on-line hyphenation, 733–734, 739–740 physical principles, 727–728 quantitative analysis, 728 Range, 826 Rayleigh-Gans-Debye approximation, 525 Reaction–dispersive model, 19–20 Reaction monitoring, 780–783 Reactive ion etching (RIE), 655–656, 655f Reciprocating pumping, 647–648, 648f Recognition mechanisms, enantioselective computation/molecular modeling, 418–423 experimental techniques, 414–418 Recycling chromatography, 584, 585f Redox reactions, 140–141 Reference spectra matrix (RSM), 733–734, 735f Refraction detection, 665 Refractive index (RI) detectors, 10–11, 70 Relative response factors (RRF), 780 Repeatability, 333, 826 Reproducibility, 826 Resolution, 50–51 Resolution-modeling software, 329 Response surface methodology (RSM), 292 Retention affecting parameters, RPLC exothermodynamic relationships, 98–99 formal mechanistic retention models, 94–95 interphase model, 93–94 pressure, 100–101 semi-empirical retention models, 95–99 solvent strength, 95–98 surface excess adsorption, 92–93 system properties, 90–92 temperature, 99–100 Retention correlation models, 349–356 Retention factor, 48–49, 326 Retention mechanism, HILIC, 229–231 Retention time, 48 Retention volume, 48 Reversed-phase (RP), 401–402, 404–405 Reversed-phase chromatography (RPC), 4

Reversed-phase liquid chromatography (RPLC), 188, 253–254, 284, 300–302, 312–313, 799, 800–803t acid–base secondary equilibria, 122–124 acid dissociation constants, 343–344 apparent formation constants, 344–346 biopartition constants, 353–354 compound descriptors, 338–343 ion-interaction chromatography, 125–132 linear free energy relationships, 101–113 metal complexation, 138–140 micellar liquid chromatography, 132–138 octanol–water partition constant, 350–352 redox reactions, 140–141 retention affecting parameters, 90–101 retention vs. pH trends, 122–124, 123f soil–water distribution constants, 352–353 Reversed-phase packing materials, 284 Reverse-phase high performance liquid chromatography (RP-HPLC), 448–450, 450t Ring-closing metathesis (RCM) reaction, 780–782 Risk assessments, 834 Robustness, 332–333, 826

S Salting-out chromatography, 300–302, 443–444 Sample introduction device, 64–65 Sample matrix (SM), 733–734 Sample preparation, 65, 188, 212–214, 371, 719 Sandwich immunoassays, 546 Schoenmakers’ rule, 256–257 Science-based calibration (SBC), 734–736 Score card approach, 711–712, 711t Segmented gradient program, 371, 372f Selective LC x LC system (sLC x LC), 568 Selectivity, 49, 79–82, 326–328 Semi-empirical retention models, 95–99 Separation factor. See Selectivity Separation process, enantioselective liquid chromatography, 405–414 kinetics of, 410–414 thermodynamics of, 405–409 Serially connected columns, 158 Setschenow equation, 302 Signal-to-noise (S/N) ratio, 748 Silanol effect, 129–130 Silica-based monolithic columns, 160–161 Silica gel, 235–236 SilicaRod, 161 Silicon-based materials, 654–656 Silver ion chromatography, 140 Similarity, selectivity, and specificity (3S), 808–809, 809f

Index

Simple-to-use self-modeling algorithm (SIMPLISMA), 734–736 Simulated moving bed (SMB) chromatography, 586–595 design, 589, 589f double-layer configuration, 594, 594f extract enrichment (EE) process, 592, 593f fraction-feedback (FF) process, 592, 593f gradient process, 590–591, 591f “Modicon” process, 591–592, 592f optimisation, 595 “Powerfeed” operation, 591–592, 592f principle, 587, 587f simple three-zone open-loop setup, 592, 593f ternary separation, 593, 593f two-column capture processes, 594, 595f “Varicol” process, 591–592, 592f Single-chain variable fragments (scFv), 455 Size-exclusion chromatography (SEC), 442, 511 absolute molar mass determination, 521–528 band broadening, 516–519 multidetector separations, 529–532 physicochemical characterization, 529–532 retention in, 512–516 two-dimensional techniques, 529–532 Size-exclusion columns (SECs), 56 Size-exclusion process, 512, 513f Slurry column packing, 42–43 Slurry packing method, 630–631 Slurry solvents, 43–44, 44t Smart LifeLC portable HPLC, 667f, 668 Snyder’s solvent-selectivity triangle (SST), 259–267, 261f, 264f Soaps, 308 Soczewinski–Snyder equation, 797 Sodium dodecyl sulfate (SDS), 133–135 Soft polymers, 653–654 Soil column liquid chromatography, 347 Soil–water distribution constants reversed-phase liquid chromatography, 352–353 soil column liquid chromatography, 347 Solenoid microcoil detectors, 756–761 Solid-film driving force model, 20 Solid-film linear driving force model, 19 Solid-phase extraction (SPE), 158, 750–751 Solid-state lasers, 684 Solid-state NMR spectroscopy, 416–417 Solute and solvent localization, 79 Solution stability, 826–827 Solvation parameter model, 102–111, 102t, 265–266 Solvatochromic scales, 254 Solvatochromic triangle, 264–265

Solvent delivery system, 63–64 Solvent demixing, 86 Solvent nomograph, 78, 78f Solvent-selectivity triangles, 257–259 Solvent strength, 95–98 Solver method, 340–341, 342f Solvophobic theory, 94–95 Spacer arms, 448–450 Spectroscopic detection, 493–494 Spider diagrams, 265–266, 266f Stainless-steel frit, 41–42, 41f Static light scattering (SLS), 525–528 Stationary phases (SPs), 7–8 capillary electrochromatography (CEC), 630–636 hydrophilic interaction liquid chromatography (HILIC), 233–239 hydrophobic interaction chromatography (HIC), 442, 448–451 microchips, 656, 657t nano-liquid chromatography, 180–181 particle-packed CEC columns, 631 quantitative structure-retention relationships (QSRR), 809–814, 810–812t, 813f Steady-state recycling (SSR) chromatography, 585–586 Stochastic–dispersive model, 30 Stochastic model, 27–30 Stop-flow experiments, 517, 518f Stop-flow mode, 749–750 Stripline detectors, 765–770 Stripper column, 490 Strong anion-exchange (SAX) phases, 632 Strong cation-exchange (SCX) phases, 632 Structures for Lossless Ion Manipulation (SLIM), 197–198 Sub-2-μm fully porous packing materials, 146–150, 164 Sulfoalkylbetaine zwitterions, 237 Sulfonated compounds, 308 Supercritical fluid chromatography (SFC), 768, 786, 807–808 Superficially porous particles (SPP), 150–156 Suppressed conductivity, 490–492, 491f Surface-confined ionic liquids stationary phases, HILIC, 238–239 Surface-enhanced Raman scattering (SERS), 728–729, 732, 739–740 Surface enhanced resonance Raman spectroscopy (SERRS), 732 Surface excess adsorption, 92–93, 92f Surfactant additives, 308–314 alkyl ethoxylates, 309

855

856

Index

Surfactant additives (Continued) alkylphosphates, 309 alkylsulfates, 308 amphoteric surfactants, 308, 310 anionic surfactants, 308–309 cationic surfactants, 309 cetyl trimethyl ammonium bromide (CTAB), 309 copolymers, 309 hexadecyl trimethyl ammonium bromide (HTAB), 309 ionic surfactants, 312–313 multifunctional surfactants, 308 non-ionic surfactants, 309 soaps, 308 sulfonated compounds, 308 Surfactant coatings, 128 Suspected target analysis, 707 Syringe pumps, 648 Systematic evolution of ligands by exponential enrichment (SELEX) process, 552–553 System suitability tests, 334

T Tag-along effect, 580–581 Tanaka column test, 55 Tandem mass spectrometry (MS/MS), 699–701 Tangent line method, 49 Tanimoto coefficient, 808–809 Target analysis, 707 Tartaramide-based chiral selectors, 397 Taylor–Aris method, 357 Temperature-controlled autoinjectors, 65 Ternary separation, 593, 593f Therapeutic drug monitoring (TDM), 212 Thermal field flow fractionation (ThFFF), 531–532, 532f Thermal polymerization, 640–641 Thermodynamics of adsorption, HILIC, 228–231 Thin-layer chromatography (TLC), 82, 807–808 Thomas model, 26–27, 33, 34f 3D printing, 653 Three-dimensional quantitative structurebiological activity relationship (3DQSAR) method, 799–805, 805f Three-dimensional quantitative structure-retention relationship (3D-QSRR) model, 799–804 Three-point interaction principle, 389–390 Time-decoupled online comprehensive 2D-UHPLC-ion mobility MS method, 566–567, 567f Time-of-flight (TOF) mass spectrometry, 695, 696f

Time-resolved NUS (TR-NUS), 778 Total energy (TE), 814 Trans-particle transport mechanisms, 21–22 Transport–dispersive model, 19 “Triad” methodology, 779 Tridecafluoroheptanoic acid (TDFHA), 284–286 Trifluoroacetic acid (TFA), 617 Triple-quadrupole (QqQ) mass analyzers, 699–700, 700f Trisacetamidohexyl isocyanurate (TAAHI), 214–215 Trisaminohexyl isocyanurate (TAHI), 214–215 True moving bed (TMB) chromatography, 586, 587f Two-column capture processes, 594, 595f Two-dimensional chromatographic separations, 164–165 Two-dimensional insertable separation tool (TWIST) concept, 569, 571f Two-dimensional ion chromatography, 495–496, 497f, 497–499t Two-dimensional liquid chromatography (2D-LC), 455–456, 659–661 Two-dimensional size-exclusion chromatography, 529–532 Type I anion-exchangers, 472 Type II anion-exchangers, 472

U Ultrafast high-performance liquid chromatography (UHPLC), 61, 64, 145–146 applications, 163–165 challenges, 165–168 high-efficiency high-speed separations, 146–156 high-throughput HPLC analysis, 156–163 monolithic columns, 159–163 with MS detection, 156–159 overview, 145–146 pharmaceutical compounds drugs of abuse, analysis of, 163–164 sample preparation, 165 sub-2-μm fully porous packing materials, 146–150 superficially porous particles, 150–156 two-dimensional chromatographic separations, 164–165 Ultrafast nuclear magnetic resonance (UF NMR), 776–778, 786 Ultraviolet-initiated polymerization, 640–641 Ultraviolet (UV) photometers, 10–11 Ultraviolet/vis absorbance detectors, 67–69 Ultraviolet–vis detection, 636–637

Index

Universal calibration, 522–525, 523f US Pharmacopeia tailing factor, 51

V Validation protocol, 333 Validation report, 334 Valve-based sample introduction system, 65 van Deemter equation, 51–52, 52f, 400, 410, 413–414, 516–517, 627 van Deemter plate height equation, 20–21 van Deemter plot, 180, 181f van’t Hoff’s equation, 406–408, 418 Variable-wavelength detector, 69 “Varicol” process, 591–592, 592f Vibrational circular dichroism (VCD), 414–417 Vibrational spectroscopy, 729 Void volume, 48

W Water-organic solvent mixtures, 91–92 Waters ionKey docking station, 666–667, 666f Weak affinity chromatography (WAC), 539 Wheatstone bridge-type differential viscometer, 522–524, 524f

X X-ray diffraction (XRD), 417

Z Zonal elution, 549, 550f Zone retention factor, 232 Zwitterionic ion chromatography (ZIC), 470–472, 471–472f, 475f Zwitterionic stationary phase, HILIC, 237

857