Method Development in Analytical HPLC [1 ed.] 0443298493, 0443298505, 9780443298509, 9780443298493

Method Development in Analytical HPLC presents the essential information for understanding the process of developing an

214 44 15MB

English Pages 605 Year 2025

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Method Development in Analytical HPLC [1 ed.]
 0443298493, 0443298505, 9780443298509, 9780443298493

Table of contents :
1. Introductory concepts related to a chemical analysis2. Retention and elution processes in HPLC3. Instrumentation in HPLC4. Performance and utilization of HPLC separation and detection5. Samples for HPLC analysis6. Sample preparation for an HPLC analysis7. Chromatographic columns in HPLC8. Mobile phase and the role of solvents in HPLC9. Development of an HPLC analytical method10. Selection of the best options in developing an HPLC analytical method11. Method and data validation in HPLC analysis

Citation preview

Method Development in Analytical HPLC A Foundational Guide to Developing HPLC Analytical Methods

Serban C. Moldoveanu SM Consulting, LLC, Winston-Salem NC, United States

Victor David University of Bucharest, Faculty of Chemistry, Bucharest, Romania

Elsevier Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands 125 London Wall, London EC2Y 5AS, United Kingdom 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies. Publisher’s note: Elsevier takes a neutral position with respect to territorial disputes or jurisdictional claims in its published content, including in maps and institutional affiliations. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. ISBN: 978-0-443-29849-3 For information on all Elsevier publications visit our website at https://www.elsevier.com/books-and-journals Publisher: Candice Janco Acquisitions Editor: Gabriela Capille Editorial Project Manager: Andrea Dulberger Production Project Manager: Sruthi Satheesh Cover Designer: Vicky Pearson Esser Typeset by TNQ Technologies

Preface

The most common technique utilized in chemical analysis is very likely highperformance liquid chromatography (HPLC). This technique is capable of generating analytical information for a wide variety of samples, such as environmental, biological, industrial, and many others. To satisfy such a large utilization, the development of new HPLC analytical methods or adaptation of such methods from the available literature is frequently necessary. However, the method development based on HPLC technique is not a simple list of actions to be followed. Successful development or adaptation of an HPLC method depends on the understanding of the theoretical principles of this technique and on the consideration of the correlations between a number of variables that influence the result in order to fulfill specific requirements such as good sensitivity, selectivity, accuracy and precision, short run time, low price, and low environmental impact. The goal of the present book is to offer a useful guide for the development of analytical methods using HPLC, with the clear understanding of the reasons for specific selections. Although plenty of information has been published regarding method development in HPLC, this information is mainly scattered in peer-reviewed articles, and very few presentations of the subject in a unified and logical manner are available. This book offers at the beginning some background regarding HPLC and on the parameters used for the characterization of HPLC performance. Subsequent chapters examine the properties of samples subject to HPLC analysis, present a summary of sample preparation techniques used to make samples amenable for the HPLC analysis, and describe some details about the chromatographic columns and mobile phases used in HPLC. The core of the book details method development in different types of HPLC. The following chapter describes various aspects of the optimization process in method development, and the last chapter presents the validation of the developed method and of the data generated by those methods. For the final implementation of this goal, the help of the Elsevier editors Ms. Gabriela Capille, Maddie Wilson, Andrea Dulberger, and Sruthi Satheesh was extremely valuable. Authors

Introductory concepts related to a chemical analysis 1.1

1

Workflow in a chemical analysis

Chemical analysis is a process that includes various techniques dedicated to the separation, identification, and quantitation of specific components in a sample. These specific components are indicated as analytes, while the rest of the sample is indicated as matrix. A large number of methods are dedicated for analyzing samples, and new methods are continuously developed or adopted from the literature. The first question in developing a new analytical method or adopting it from the literature should be “What is the purpose of the analysis?” Once this question is answered, the next question is “What are the properties of the samples that must be analyzed?” Based on the answers to these questions, it can be proceeded to the process of developing a new analytical method or of selecting one from the literature. A typical flow for completing a new chemical analysis is schematically indicated in Fig. 1.1.1. This flow is dictated by various aspects of the analysis such as the purpose of analysis, the target analytes and their concentration, the sample matrix, the need of sample preparation, the time required for the analysis, available instrumentation, and other aspects that will be further discussed in this book. The first step for an analysis is the sample collection indicated as sampling. Correct sampling is very important since the results of analysis depend on sampling, and the errors generated during sampling are impossible to correct, unless additional sampling is performed and the whole analysis is repeated. The subject of sampling is beyond the purpose of this book, and only a summary discussion on this topic is given in Section 1.2. To develop or adopt successfully an analytical method, it is very important to obtain as much information as possible about various aspects related to the purpose of analysis, sample properties and its conservation, the utilization of the results of the analysis, etc. In addition, the collection of information from the literature related to the

Data processing

Collection of information Planning for a on purpose of analysis method of analysis and sample properties

Sampling and sample submission

Results interpretation

Method development Sample conservation

Core analysis for pure analytes

Sample preparation

Core analysis for samples

Figure 1.1.1 Typical flow for completing a new chemical analysis. Method Development in Analytical HPLC. https://doi.org/10.1016/B978-0-443-29849-3.00003-8 Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

2

Method Development in Analytical HPLC

analysis of the same type or similar type of samples is very important. A more detailed description of the types of information useful for the method development of a chemical analysis is given in Section 1.3. If the preliminary information is missing, preliminary experimental work must be performed to identify the analytes, and to generate approximate ranges of concentrations of the analytes in the expected samples. The adoption of a method from the literature is usually associated with some modifications of the indicated procedure since rather frequently the repeating of the literature recommendations encounters difficulties and may require some changes. In addition to that, even when a new method is developed, guidance from other similar procedures is almost always utilized. Between adopting a method and developing a new one, there is in fact a continuous of possibilities. For this reason, further discussions are equally applicable to the development of an entirely new method, and the adoption of one from the literature. Some analytical procedures consist of a direct measurement of the analytes, and others involve a separation of sample constituents followed by their measurements. The methods including a separation on line with measurement are indicated as hyphenated methods. Such methods have the advantage of being able to use for detection techniques that otherwise cannot be used for the analysis of the compounds in a mixture. Usually, such methods can analyze in one sample more than one analyte as they are separated one by one. A more detailed discussion on core analytical procedures in general is given in Section 1.4. One of the most versatile techniques used as core analysis is high performance liquid chromatography (HPLC), which is a hyphenated method and for which the development of methods of analysis is the subject of present book. After sufficient information is collected the planning for the analysis should start. The selection of a specific analytical procedure is determined by a variety of factors including the information obtained regarding the sample and the purpose of analysis, the capabilities of the laboratory, timing required for delivering the results, precision and accuracy required for the results, number of samples that are expected to be analyzed, etc. An overview regarding the decision to use an HPLC analytical procedure that fits the requirements of an analysis is given in Section 1.6, but such information is also presented in many other parts of this book. Once a decision to develop an analytical method is made based on the purpose of analysis and the properties of the sample, five stages can be differentiated for a method development: (1) develop a method for a synthetic mixture of pure analytes that should be measured, (2) process the raw sample (sample preparation step) to make the raw sample amenable for the method of analysis previously developed for the mixture of pure analytes, (3) apply the core analysis developed for the pure compounds to the real samples, after they were subject to sample preparation, (4) evaluate the results and decide about the analysis adequacy, and (5) validate the method following a specified protocol. In order to develop a reliable method of analysis, it is very important that every step along the method development is fully verified. In a multistep process as a method development is, the errors from each step are accumulating and the final result can be inadequate if any of the intermediate steps generated errors:

Introductory concepts related to a chemical analysis

3

1) The initial core method is usually developed only for a mixture of pure analytes the mixture resembling the composition of the analytes from the real samples. That assures that at least in an ideal situation, the core analytical procedure is functional. At this stage, also one or more internal standards must be selected and a calibration for the quantitation of the analytes is obtained. In addition, the evaluation of some potential interferences can be tested and verified if they affect the measurements and need to be eliminated by sample preparation. 2) After an analytical method has been developed for the pure compounds and some potential interfering parts of the matrix are evaluated, the next step is the sample preparation. Although sample preparation is performed in connection with the core analytical method, it usually consists of a set of different operations. The sample preparation can be as simple as the dissolution of the sample in an adequate solvent since most core analytical procedure requires the sample to be present in a solution. However, elaborate sample preparations may also be needed and these can include sample clean-up (simplification of the sample matrix), analyte concentration, changing of the chemical nature of the analytes or matrix (by derivatization), etc. The method development for HPLC which is presented in this book is concerned mainly with the core analytical method and a discussion on sample preparation for HPLC is limited to Chapter 6. For detailed sample preparation procedures, the dedicated literature must be considered (e.g., Ref. [1]). 3) Following sample preparation, the core analysis is performed on the processed real samples. Additional problems may occur when real samples are analyzed. The real samples, although were subject to sample preparation, may still contain a residual matrix. Also, the concentration of the analytes in some of the real samples may be outside the expected range of calibration for the analytes. This type of problem must be resolved, either by adjusting the initial method developed for pure analytes, or by adjusting the sample preparation. 4) After the analysis is performed, the generated data are processed, and the results are interpreted. The quality of the results and their usefulness are dependent on the attributes of the analytical process (sample preparation, core analysis). These attributes are important for evaluating the quality of the analytical method. The list of such qualities includes selectivity, accuracy, sensitivity, reproducibility, adequate concentration range, adequate time for the analysis to run, robustness, simplicity of sample preparation, limited cost, etc. If the analytical method is satisfactory, it is not usually recommended to continue with its improvements, unless improvements are truly necessary. All methods can be continued to be improved, but this may require time and effort which in many cases is not an efficient process. 5) For routine utilization of an analytical method, it is necessary that the method is also subject to a rigorous validation process. The validation is not limited to the core separation analysis being extended to the whole method, including the sample preparation step. The validation is commonly performed using a specific protocol. Various such protocols are recommended in the literature and the subject of method validation is further presented in Chapter 11.

The development of an analytical method (either newly developed or adopted from the literature) can be a trial-and-error process, and sometimes an initially developed method may prove to be inadequate for the desired analysis. In such case, a second (or even third) choice must be made. This book is dedicated to the process of developing an analytical method that utilizes HPLC as a core analysis. The information provided is intended to help in shortening as much as possible the trial-and-error process, and to guide the reader in obtaining efficiently a useful HPLC analytical method.

4

Method Development in Analytical HPLC

Key points • •



The development of a method of analysis is based on the purpose of analysis and the properties of the samples that must be analyzed. Method development includes five parts: (1) development of a core method for the analytes in a pure solution, (2) sample preparation of the raw samples, (3) application of core analysis developed to real samples, (4) evaluation of results, and (5) validation following a recommended protocol. The use of HPLC as a core analytical technique is very common.

1.2

Basic information about sampling and preservation of samples

Sampling in analytical chemistry is the operation of sample collection with the purpose of analyzing it. Sampling of materials for being analyzed by HPLC is no different from general sampling. The procedure frequently consists in obtaining only a portion from the bulk material that is the object of analysis, but the collected portion must be representative for the investigated material. The use of the whole material to be analyzed as a sample, or sampling as an integrated part of the analysis, is also possible. The incorrect results of an analysis are frequently caused by incorrect sampling, for example, because the composition of the collected sample has differences from that of the material to be analyzed. After a (representative) sample is obtained from a specific material, it is common that subsamples are generated which are further used for the analysis. Sampling procedure is determined by a number of factors. Among these are the purpose of analysis, the physical and chemical nature of the sample, whether the sampling is performed for a material characterization or a process control (material from a process), the proportion of the analytes in the sample matrix (if it can be estimated), the required precision and accuracy of the analysis (see Chapter 11), the available quantity of the material to be sampled, the sample price, and several other factors. A summary of bulk material properties to be considered during sampling is indicated in Table 1.2.1.

Table 1.2.1 Bulk material characteristics to consider during sampling. Physical property

Chemical nature

Stability/value

Gas/vapor/aerosol Liquid Solid Mixed phases Homogeneous Nonhomogeneous

Organic Inorganic Mix Biological Pharmaceutical Environmental

Stable Unstable Unique/expensive Perishable Hazardous (toxic or explosive)

Introductory concepts related to a chemical analysis

5

Besides being representative for the material to be analyzed, sampling must fulfill other requirements such as to provide sufficient material for the analysis and to avoid any contamination. If possible, and the price of sample is not prohibitive, at least 3e5 times more than the minimum amount of sample in which the analytes can be measured should be available. The sample should be associated with the appropriate information regarding stability, potential hazard, the number of replicates required to be analyzed, and if after analysis the remaining of the sample must be returned to the owner. For the same material, it is also common to obtain several samples which will be separately analyzed. This is usually necessary to verify the uniformity of a specific material. The samples can be obtained, for example, from different batches. The correct sampling is also essential in such cases, a wrong sampling being capable to indicate differences which are not real. In many instances, the samples are already collected and submitted for analysis without any input from the analyst. In such cases, sample characteristics cannot be anymore changed. However, if sampling can be influenced by the analyst, information obtained before sampling can be utilized regarding the sampling procedure, deciding the amount of sample to be collected, and in some instances the time of sampling, location of sampling, etc. These aspects of sampling are based on the purpose of the planned analysis. The amount (mass) or the volume of sample or subsample that is the direct subject of analysis must be precisely known when a quantitative analysis is required.

Types of sampling Several sampling techniques can be utilized in practice [2e5]. A common procedure is the random sampling. For solid materials, this procedure assumes that the bulk material is made from a large number of discrete portions, and each portion has the same probability to be selected. A number of these portions must be collected to serve as a representation of the bulk. If performed correctly, a random sampling procedure will lead to unbiased representative samples. For a uniform sample, just taking a portion of the sample can be considered random sampling. Another procedure is systematic sampling which is achieved by taking portions of the bulk material on a regular basis in space and/or time. For instance, a material moving steadily on a conveyer belt or in a pipeline may be sampled systematically by removing equal amounts of material at fixed time intervals. Stratified sampling is executed when the bulk material is distributed in a number of subgroups, or zones that are assumed homogeneous. This distribution can be done either in reality (as when the material is put in a number of containers) or schematically (as when a material is subdivided into a number of areas). Sampling is then accomplished either randomly or systematically in each zone. Cluster sampling is a sampling technique used when “natural” but relatively homogeneous groupings are evident in material. In this technique, the material is divided into these groups (or clusters) and a simple random sample of the groups is selected. Convenience sampling is performed without any regard to representativeness for the bulk. This is, for example, the case of analysis of inclusions, and most of the time this

6

Method Development in Analytical HPLC

type of sampling is performed without the intention of quantitation. Sampling using a specific protocol is another type of sampling. In this case, the sampling is done in the same manner for a variety of samples, to provide the results for a specific parameter that may be used for the comparison of samples. The sampling performed in a series of steps each step consisting in taking a portion from the sample taken in the previous step is indicated as multistage sampling. Homogenization for each step can be necessary during multistage sampling. Particular procedures of sampling are sometimes used, depending on samples characteristics. For example, the sample collection from archeological artifacts must be performed with as low invasiveness as possible to the evaluated object, and in some cases not violating specific community or religious traditions [6]. Details regarding sampling can be found in numerous publications, such as books (e.g., Refs. [7e11]), papers from peer-reviewed journals (e.g., Refs. [12e21]), or on the web (e.g., Ref. [22]).

Procedures for gas sampling Gasses can be analyzed by various procedures, but HPLC is also one of the analytical methods applied for gas analysis. Because the sample in HPLC must be a liquid that is injected in the flow of the mobile phase, the collected gas sample must be further dissolved in an adequate solvent. For the initial gas sampling, the collection of a specific volume of gas is necessary. As any sample, the gas consists of analytes and matrix. For example, the matrices for atmospheric samples are the main components of air (78.084% nitrogen and 20.947% oxygen, 0.934% argon), the other components (e.g., carbon dioxide, neon, water vapors, etc.) being classified as minor and trace components. The analytes in gases can be other gases, vapors, liquid particles, solid particles, or a mixture of those. HPLC is mainly utilized for the analytes of vapors and aerosols in a gas matrix, and not for the measurement of permanent gases. The composition of the sample indicating the concentration of a specific analyte in the gas can be reported as volume of analyte to the volume of the gas, mass of analyte to the volume of gas, mass of analyte to the mass of the gas, etc. The change from volume to mass can be done if the density is known (m ¼ Vr, where m is mass, V is volume, and r is density). The gas density depends on pressure and temperature, and for this reason, the collection of a gas volume must be done in controlled pressure and temperature conditions. When flowing gases are sampled, both static pressure and pressure due to the velocity of the fluid molecules must be considered. In these specific measurements, the gas flow rate must be measured. The gas collection techniques can be grouped in static sampling, dynamic sampling, and sampling based on diffusion (another classification recognizes active and passive sampling). The homogeneity of gasses is not always certain since, for example, high density vapors may separate to a certain extent from lighter gases. The presence of aerosols (liquid and/or solid particles) also poses problems for collecting a homogeneous gas sample since they may be retained on the walls of the container. The dynamic type of gas sampling consists of the collection of a stream of gas sample. If

Introductory concepts related to a chemical analysis

7

the gas volumetric flow rate is U, the collection time is t, and the collection efficiency is Ef , the volume of the gas sample (V) collected is given by the relation: V ¼ Ef Ut

(1.2.1)

The gas flow rate U can be measured using Venturi meters, rotameters, mass flow meters, etc. When the sampling is done from a stream of flowing gases, such as for smoke stacks or dryer exhausts, the velocity of the sampled gas must be measured. The gas velocity measurement can be done using Pitot tubes or electronic velocity meters. From the gas velocity and the cross-section area of the pipe in which the gas is flowing, the flow rate of the sampled gas can be obtained. The initial gas collection by dissolution of the gas in a liquid medium, or the transfer of a collected gas sample into a liquid for injection can be done with or without any chemical reaction. This procedure can be applied using impingers containing a solvent. For the case when the solvent does not react with the analytes, the solvent provides only an appropriate medium for dissolving the analytes. The yield of recovery in this case can be approximated with the equilibrium constant of the gas liquid partition process. At equilibrium, the absorption of a gas X in a liquid A; when no chemical reaction is involved, follows Henry’s law given by expression: KX;A ¼

½XA pX

(1.2.2)

where ½XA is the molar concentration of X in the liquid medium A and pX is its partial pressure of X in gas phase. Expression 1.2.2 shows that increased pressure increases the analyte concentration in the solvent. However, the temperature and other factors such as ionic strength (through activity coefficient) of the liquid medium influence the equilibrium. Besides solutions that simply dissolve the analyte, in many cases, the liquid used for the retention reacts with the analytes. For example, SO2 can be collected in solutions containing H2O2 to transform it into SO2 4 ions [23], SO2, NO2, CO2 can be collected  in a basic solution of KMnO4 to change them into the corresponding SO2 4 , NO3 , 2 CO3 ions [24], carbonyl compounds can be collected in a solution of 2,4dinitrophenylhydrazine to transform them in the corresponding 2,4dinitrophenylhydrazones [25], etc. An interesting application of using a chemical reaction to trap an analyte is the collection of hydroxyl radicals from atmosphere. These radicals are generated from various photochemical reactions. The collection is performed using the reaction with salicylic acid, 2,3-dihydroxybenzoic acid, or 2,5-dihydroxybenzoic acid [26], or with terephthalic acid [27]. This type of collection can be considered a combination of sampling and sample preparation. Condensation of the analytes at low temperature (cryogenic collection) from the flowing gas is another common collection technique for gases. This kind of sampling is used for the collection of volatile and semivolatile compounds that can condense at a certain low temperature in a gas matrix that does not condense (e.g., air). Another

8

Method Development in Analytical HPLC

sampling procedure for gasses is based on adsorption on a solid surface that has high affinity for the gaseous analyte. This procedure can be applied for adsorbing gases and vapors to be analyzed from a diluting gas not retained by the adsorbing solid. A variety of adsorbing materials are available for gas collection such as active carbon, silica gel, zeolites, certain inorganic adsorbents, and many synthetic polymers. Various adsorbents with specific trade names are commercially available (e.g., CarbosieveTM, Anasorb®, Amberlite XAD, Chromosorb, Carboxen, Tenax GR, etc.). The adsorbing material can be placed in a tube or cartridge, a stream of gas with specific volume being passed through the tube. This volume can be either directly measured [28] or in case that a pump is used to aspirate the gas through the tube with the adsorbing material, the flow rate of the analyzed gas and the time period of collection are measured. The adsorption on a solid material can be also used in static conditions [29]. Desorbing of the analytes from a solid material can be performed using a variety of procedures. One such procedure is the extraction of the analytes using a solvent that does not dissolve the solid adsorbent and the extract can be analyzed by HPLC, or other analytical techniques [30]. Other procedures use thermal desorption in an analytical instrument, but this is mainly applicable for gas chromatographic (GC) analysis. Filtration using special filter media is frequently used for the sampling of aerosols from gases. The filters can be made from various materials such as glass fiber, porous nylon, polypropylene, polyurethane, Teflon (PTFE), etc. [31,32]. Filters coated with a specific reagent where the analyte reacts with the coating are also utilized for gas collection [33]. Diffusive gas collection is another sampling technique. A diffusive sampler is a device capable of taking samples of gases or vapors at a rate controlled by diffusion through a membrane and collection in a sorbent. This procedure does not involve any active movement of the gas through the sampler. In an ideal diffusion process, no convection takes place and the concentration remains constant at the surface of the sampler [34]. Although this type of sampling is more frequently followed by GC analysis or directly by a mass spectrometric (MS) analysis, there are situations when the technique can be coupled to HPLC [35].

Procedures for sampling of liquids The common requirements for sampling are applied to liquids and although many liquids are homogeneous the collection of a representative sample must be carefully controlled. Some liquids are not homogeneous, and may contain solids that are not uniformly distributed. Also, certain liquid samples that are collected as one phase may separate in time in a different layer, or solids may separate from the sample in time. Such problems must be individually considered and proper technique must be used for sampling. Liquid sampling can be done similar to gas sampling using static, dynamic, and diffusive techniques. In static sampling, a certain volume or mass of liquid is collected. In some instances, this volume or mass needs to be measured, but in other cases, only a subsample used for the analysis is measured. Special static collection devices can be used, such in the case of rain sampling [36]. Dynamic sampling is done mainly for moving liquids [37]. Sampling of only specific components from water can

Introductory concepts related to a chemical analysis

9

be performed using specific adsorbent materials. For quantitative information, a measured volume of liquid must be subject of adsorption [38]. Similar to the gas sampling, this procedure can be viewed as sample preparation with sampling being limited to the collection of the specific volume of liquid and the retention of analytes as a concentration step. Quantitation can also be performed based on the flow rate of a liquid and collection time, as well as collection efficiency, by using a relation of the type 1.2.1 for estimating the volume of a flowing liquid.

Procedures of sampling for solids Solid materials are frequently nonhomogeneous, and sampling poses the problem of collecting a representative sample. In such cases, the collection of a larger amount of material should be initially done, followed by homogenization and obtaining a subsample. Some theoretical aspects about the sampling of a solid material can be obtained by considering a bulk material made from individual particles but only a part of them containing the analyte. If the analyte is present only in a fraction p (0 < p < 1) of particles, and n particles are sampled, the average number of particles containing the analyte is m given by the following formula: m¼n p

(1.2.3)

If multiple sampling is performed, the average resulting number m of samples containing the analyte will be affected by errors and the standard deviation for m will be that of a binomial distribution [39]: s¼

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi n p ð1  pÞ

(1.2.4)

In order to ensure that the collected material is representative, the relative standard deviation s=m % for a number of collections should be as small as possible. The dependence of s=m % on n for a given p is exemplified in Fig. 1.2.1, and the dependence of s=m % on p for different n values is exemplified in Fig. 1.2.2. The graphs in Fig. 1.2.1 and 1.2.2 show, as expected, that a lower s= m % is obtained when the number of sampled particles n is higher, and when the fraction p is higher. Translation in practice of those observations can be taken as directional recommendations. In theory, the representativeness of an analytical sample may be quantified by the reliability of the obtained mean value m from a number n samples as compared to the true mean m of the bulk material. The confidence interval ðm  dÞ < m< ðm þdÞ; which characterizes representativeness, depends on the value d that can be derived for a small number of measurements n following Student’s “t” distribution, with the formula: tn1;P s d ¼ pffiffiffi n

(1.2.5)

10

Method Development in Analytical HPLC

100 90 80 70

s/m %

60

p = 0.01

50

40

p = 0.05

30 20

p = 0.1

10 0

0

200

400

n

600

800

1000

Figure 1.2.1 Variation of s=m % as a function of the number n of sampled particles for a bulk material for three fractions p of analyte.

100 90 80

70

s/m %

60

n = 10

50 40

n = 100

30 20

n = 1000

10 0 0

0.1

0.2

p

0.3

0.4

0.5

Figure 1.2.2 Variation of s=m % as a function of the fraction p of analyte for three values of the number n of sampled particles.

Introductory concepts related to a chemical analysis

11

where tn1;P is the tabulated variable for Student’s “t” distribution and depends on the number of measurements n and the chosen statistical certainty P (tn1;P can be replaced with zP , for Gaussian distribution when n is large). For a smaller desired d, expression 1.2.5 shows why the number of sampled particles must be larger. A common procedure in sampling solid materials is to obtain initially a larger sample that is further resampled after a homogenization step obtained, for example, by grinding (when a lower amount of sample may have a large number of particles). The previous theory assumes a p value identical for the bulk material. In practice, this assumption is not always true, and the analyte may be at a higher p in some parts of the bulk material than in other. Some homogenization of solid bulk materials can be done, for example, by quartering. This consists of flattening the whole or a relatively large part of bulk material (initial large sample) into a circular or square layer. This layer is quartered and the alternate quarters are discarded. The remaining material is mixed and again distributed into a circular or square layer the process being repeated until a sample of suitable size is obtained. Intermediates samples generated by quartering may or may not be subject to grinding for obtaining smaller particles. Besides being representative, the amount of sample must be sufficiently large for the analysis to be successful. This amount depends on sensitivity of the analytical procedure, the method of sample preparation, the number of required replicates, etc. Sampling with representative result can be difficult for some solid materials such as solid waste, or some minerals such as gold ores that are very nonhomogeneous. Solids that contain pores filled with liquids, gases, or biological materials may pose problems depending on the purpose of analysis. The answer to the lack of representativeness for this type of sample is the presentation of data indicating precisely the conditions for the collection, and evaluation of the results as a scattered data set, possible for a large number of sampling points and over a longer period of time.

Sampling of mixed-phases materials Mixed-phase is a common type of material that needs to be analyzed. Examples of mixed-phase are various foods, soils, suspended particles in liquids, biological materials, etc. Soil, for example, contains a solid portion, and also liquids and gases. Specific sampling protocols must be utilized for such samples [40,41]. Separate sampling for the different phases and the evaluation of the ratio of each phase may be chosen for obtaining the correct information on the sample. Sampling with high reproducibility of materials like mud, oil shales, sewage, and sludge can be difficult to make [42]. Also, depending on the purpose of analysis, the whole material or only a component (e.g., the solid part) must be analyzed. For such materials, special sampling protocols may be designed, or for assuring that the sample is representative, the collection should be done in a larger number of points (in space and/or time). In the category of mixed phases can be included emulsions, suspensions, pastes, foams, etc. The general requirements of sampling with special attention to the homogeneity should be applied for such materials.

12

Method Development in Analytical HPLC

Sampling of composite materials Some materials that must be analyzed are present in the form of composite material, such as coated tablets of pharmaceutical drugs, where the coating and the content of the tablet are different materials. Depending on the purpose of analysis, sampling for such materials should follow a special protocol, or a specific sample preparation technique should be used as part of sampling that allows the separation of the sample components [43,44]. There are no general rules directing the sampling of composite materials that can be very diverse. A common example of special sampling is that for historic and artistic objects when sampling must be nondestructive.

Sampling of biological materials Biological samples can be present in various forms such as gases, liquids, solids, or mixed phases, and they may include gas and liquid body fluids, cellular components, tissues, biogenic substances, etc. Many specific techniques and protocols are described for sampling of biological materials, depending on the purpose of sampling, the type of material to be collected, the type of analysis to be performed on the sample, etc. One such specific protocol is, for example, the dried blood spot technique which became a routine method for the collection, transport, and storage of blood as a dried sample on a paper-type substrate [45]. For biological samples, special care must be given to avoid contamination (of both the sample as well as of the environment), and to avoid the change of sample composition, or affect its stability. Sample transport and preservation of biological samples may also pose specific requirements. Limiting the sample size is also important in sampling biological materials, in particular when sampling involves invasive techniques such as the collection of blood, amniotic fluid, cerebrospinal fluid, or biopsy samples. Metabolomic studies providing information of the metabolic phenotype can be affected by various genetic or environmental factors, and consequently the HPLC-MS based investigations require specific protocols in the preanalytical stage in order to avoid unwanted results caused by improper operations for the collection and preparation of biofluids and extracts (blood, plasma, urine, tissues, and cells) [46]. Sampling of biological materials is covered in detail in various dedicated publications (see e.g., Refs. [47,48]).

On-line sampling coupled with HPLC analysis Usually, the step of sampling is carried out independently on the HPLC analysis, and the samples collected are subjected to their preparation and further analysis. However, certain applications are based on direct coupling the sampling to the HPLC system, eliminating the intermediate step of sample preparation. An example of this type is the use of dialysis for sampling procedure applied for the isolation of some inorganic 2 3 anions (F, Cl, Br, NO 2 , SO4 , PO4 , and others) that are transferred across a membrane into a liquid receiver and carried to the injection valve of an ionchromatographic system [49]. The enrichment factor for this procedure can be adjusted by controlling the flow rate ratio of sample and receiving solutions in the dynamic

Introductory concepts related to a chemical analysis

13

dialysis system [50]. The on-line dialysis sampling method coupled with HPLC can also be applied to organic compounds from an aqueous matrix [51], or for simultaneous investigation of the interactions between multiple bioactive compounds contained in different herbal medicines [52].

Handling and storage of samples Sample handling (including transport), and storage must maintain sample integrity. These include avoiding contamination of both sample and environment, as well as keeping sample composition unchanged. Losses of analyte may occur because of adsorption on the container walls, loss of volatiles, and for reactive samples photodecomposition, thermal decomposition, condensation reactions of small molecules, oxidation or reduction processes, etc. Changes in physical aspect of the sample may also occur, such as separation of phases (separation of nonmiscible solvents, precipitation of solids, etc.). Changes in moisture content are, for example, a common effect, which affect the results of a chemical analysis. Absorption of water is also a potential change. Storing samples at low temperature and in the absence of light is a common procedure to avoid composition changes. Addition of preservatives, of antioxidants, or adjustment of pH may be needed in special cases for keeping the sample unchanged. Special attention must be paid to the containers used for storage such as the plastic ones that may cause contamination with volatile compounds or plasticizers, or may produce losses due to permeability. Biological samples, which can easily change their composition, are usually stored in frozen state under liquid N2 [53]. An important issue related to the biological samples subjected to a freezing process is the stability of the samples following a number of freeze-thaw cycles [54] (see Section 11.2). Sample stability is an important issue in chemical analysis and some details regarding potential sources of sample lack of stability are further presented in Section 11.2.

Key points • • • •

Sampling must provide a representative sample for the material to be analyzed. Errors in sampling affect significantly the final result of an analysis. For sampling solid samples, a large representative sample can be initially collected, followed by its homogenization and resampling to generate a smaller representative sample. Sample stability is an important subject in any chemical analysis, including HPLC.

1.3

Collection of information and planning for developing a method of analysis

For a specific required analysis, it is either necessary to develop a new method of analysis or to adopt a method already described in the literature (printed or on the web). Before starting this process, the collection of information is a very important activity. This information is essential for generating an adequate method and this is not limited

14

Method Development in Analytical HPLC

to HPLC methods of analysis but to analytical methods in general. Planning for the implementation of the new or adopted analytical method is also necessary in order to understand which resources are necessary and which is their availability. The information collection and planning are of informational nature and do not necessarily involve any real operation (unless preliminary analyses must be performed).

Collection of information The collection of information should help clarifying the following aspects: (1) the purpose of analysis, (2) the nature of the materials to be analyzed and their properties, (3) the analytes to be measured and possibly an estimation of their abundance in the sample (major constituents, trace, etc.), and (4) the available literature related to the analysis of similar of identical samples. In addition, information regarding available instrumentation, necessary chemicals, funding, time for delivering results, etc. must be collected. The information collected about a requested analysis cannot be always comprehensive. However, whatever information is available, it should be collected and thoroughly reviewed. Each of these aspects is further commented: 1) To start the development of a method of analysis, it is obviously required to know for what kind of samples the method must be developed. This should be associated with the information explaining why the analysis is necessary and how the results will be utilized. Samples are frequently analyzed for health-related purposes (e.g., medical analyses, analysis of pharmaceuticals, analysis of metabolites), for evaluating environmental issues, for forensic purposes, for exploratory reasons, or for fundamental research. An analysis can be performed for responding to official or legal requirements, for the assessment of the quality of raw materials, for controlling a process, for assuring the quality of finished products, for reverse engineering, or for product development purposes. Based on the purpose of analysis, it should be clear if qualitative, quantitative, or both types of results are expected from the analysis, or if a special type of analysis such as that of enantiomers or of the structure of the compounds is needed. Before starting to develop an analytical method, it is important to know what number of samples is expected to be analyzed, how fast the results must be delivered, and what is the limit for the cost of the analysis. The methods developed for a large number of samples must be more carefully structured. Usually, they need to be shorter and more stable to small changes. Also, the information regarding the required precision of the results is important. In addition, it must be known whether a specific protocol must be followed during the analysis, if the validation of the method must be done following a specific protocol, or if no regulations are imposed. Some analyses are required to be nondestructive and the sample must be returned to the owner, and in certain cases the analysis is done in conjunction with preparative purposes and this should be known. 2) The information about the material to be analyzed should start with knowledge about the available amount of sample (large quantity, small quantity, readily available, unique, etc.), sample homogeneity, the age of the sample, potential contamination, the value of the sample, the origin of the sample, the number of samples to be analyzed and if the same type of samples will continue to be submitted for an extended period of time. Also, it must be known if the whole sample should be analyzed or only a specific part (surface, soluble component, selected points, etc.). Following this information, it is useful to obtain data about the physical

Introductory concepts related to a chemical analysis

15

state of the sample (homogeneous, nonhomogeneous, solid, liquid, gas). If possible, details should be obtained about the nature of the sample matrix (organic or inorganic material, biological, environmental, composite, etc.). Also, information about the sample thermal stability and perishability, safety concerns about the sample, etc. should be collected at this point. 3) The information about the analytes to be measured is another important component in planning an analysis. This includes the nature of the analytes or at least the class of the analytes (inorganic, organic, functional groups in organic compounds, ionic character, etc.). If this information is missing, it is important to know at least if the analytes are small molecules or polymeric ones. In the case that very little information is available about the sample composition, preliminary analyses are sometimes performed. These preliminary analyses can be qualitative or semiquantitative. For example, a gas chromatographic-mass spectrometric (GC-MS) analysis (if possible) may provide valuable qualitative information for volatiles and semivolatiles components of an unknown sample. In case of nonpolymeric analyte molecules, data regarding volatility, solubility, and reactivity are very useful. For macromolecules, a general characterization is always useful such as if the polymer is natural or synthetic, or if it is only a component of a composite material. Other data regarding the analytes are helpful, such as information on the estimated level of analytes in the sample (trace, medium levels, major constituent). The aspect regarding the level of analytes is very important in deciding the required sensitivity of the analytical method. 4) The collection of information on methods of analysis similar or identical with a projected one is a very useful step. The scientific literature contains an enormous number of analytical methods. These methods may include information about the analysis of the same type of samples and analytes as those for which an analytical method is necessary. In such cases, a decision can be made to simply implement the published method unmodified, or to modify it to a certain extent in order to adjust it to a specific new demand. However, it is rather common that a direct implementation of a published method without any change is not possible. In this case, one or more reported analytical methods should be compared and a new method can be developed. A third possibility exists, when no such analysis as the one of interest is known, and an entirely new method should be developed. In this case, the information about methods having only some similarities is still very useful. The selection or development of an analytical method must also take into account other aspects besides the nature of the sample and its analytes. This includes the availability in the laboratory where the method will be developed of instruments, of expertise, and of funding.

Planning and collection of necessary resources The planning for developing a method of analysis is based on the collected information as previously described. One important step for the planning is the decision if a method from the literature should be adopted for the required analysis, if that method must be modified, or if a new method should be developed because no adequate method for the specific task was previously reported (or found). Also, because every analytical method consists of the development of a core analytical method on pure compounds, followed by the sample preparation of real samples, and then the application of the core analytical method on real (processed) sample, separate plans are made for each of these steps. The initial step must be based on the information collected before starting the

16

Method Development in Analytical HPLC

analysis, or on some preliminary experiments that indicate which analytes must be analyzed and what would be the ranges of content in the samples that will be analyzed (traces of compounds, some specific content, major constituents, etc.). For selecting the core analytical technique, besides the verification that it works on pure compounds, other considerations must include knowledge about the resources available in the laboratory such as instrumentation, expertise, available chemicals, and funding. HPLC is a common technique and frequently it is selected as core analytical procedure. The following sample preparation step depends on several factors. These factors include the sample characteristics (such as sample matrix, nature of analytes and their concentration, quantity of sample available) and the requirements of the core analytical procedure developed on pure compounds (selectivity and sensitivity of the method). Sample preparation also depends on the type of detection used in the HPLC (UV, MS, MS/MS, etc.). The procedure must be adjusted based on the requirements of the core analytical method. Sample preparation can be a time-consuming step, but it can be very important in order to clean-up the sample matrix, concentrate the analytes, improve detectability, and make the sample amenable for the projected core analytical procedure. Only a summary discussion of sample preparation is given in Chapter 6, while the main subject of this book is related to the selection and implementation of HPLC as the core analytical procedure. The initial period of planning is typically followed by the collection of the necessary resources, the necessary instrumentation, and collection of materials such as various chemicals including solvents, reagents, standards to be utilized in calibrations, certified reference materials (CRMs), etc.

Collection of chemicals and certified reference materials Specific attention before starting the development of an analytical method is the collection of necessary chemicals and reference materials. The chemicals that are to be purchased must fulfill specific criteria of purity such as A.C.S. (chemical grade that meets or exceeds purity standards set by American Chemical Society), reagent purity (generally equal to A.C.S. grade and suitable for use in many laboratory and analytical applications), U.S.P. (chemical grade of sufficient purity to meet or exceed requirements of the U.S. Pharmacopeia), N.F. (sufficient purity to meet or exceed requirements of the National Formulary), Lab (relatively high quality with exact levels of impurities unknown; usually pure enough for food, drug, or medicinal use), purified (practical grade, meeting no official standard), technical (used for commercial and industrial purposes). Other purity classifications are also known [55]. In addition to the necessary chemicals, in many instances for the validation of a method are necessary CRMs [56]. Further discussion on chemicals and certified materials is given in Section 5.4.

Key point •

Collection of as much as possible preliminary information about an analysis is very useful for its further development.

Introductory concepts related to a chemical analysis

1.4

17

Short review of methods used as core analytical procedures

Analytical chemistry as an independent discipline started to be developed in the early 1800s and its first dedicated scientific journal begin its publication in 1862 [57,58]. From the beginning, the analytical procedures were intended to provide information such as qualitative, quantitative, and/or structural for a material or process (material from a process). The same goal remained valid to the present. In this section, a general view on several common core analytical procedures is presented. As previously indicated, almost every method of analysis also contains a sample preparation part which is described in some details in Chapter 6, but the subject is beyond the purpose of present book and detailed information on sample preparation is readily available in the literature (e.g., Ref. [1]). The core analytical procedures can be classified as “nonhyphenated techniques” and techniques including an independent separatory step and a detection in the core analysis. Such techniques can be indicated as “hyphenated”. Nonhyphenated techniques cover a wide area of analytical procedures including wet chemistry, optical spectroscopy, optical nonspectroscopic techniques, MS, X-ray analytical methods, nuclear magnetic resonance (NMR), radiochemical techniques, thermal analysis techniques, electrochemical techniques, etc. In some cases, nonhyphenated techniques can be used only for the analysis of pure compounds or solutions of pure compounds. However, applicability to multicomponent samples is also possible when a specific physical or chemical property of the analyte is not shared with other sample components and does not suffer interference from their presence. Many nonhyphenated analytical techniques are also used for on-line detection of a sample components that were separated in a previous step. This is a common use, for example, for certain optical methods, and for MS. The resulting techniques are made from two hyphenated parts, one performing a separation of sample components and the other part performing detection. Many such techniques include chromatographic separations on-line with various detection procedures. The advantage of including a separation step in the core analysis is that such techniques are in most cases a significantly better fit for the analysis of complex mixtures as compared to nonhyphenated ones. In addition, even when applicable to more complex samples, the use of nonhyphenated techniques is typically limited to the analysis of a single component of the sample. The techniques including a separation can be used for the measurement of many analytes in a unique analysis that requires a small sample size. One additional advantage of most hyphenated methods is related to the addition of automation that allows the analysis of large sets of samples without human intervention (e.g., use of autosamplers).

Wet chemistry Wet chemistry methods of analysis are one of the earliest utilized analytical techniques, and they comprise of a multitude of procedures. For qualitative purposes, various chemical reactions can be used for generating a specific color, precipitate,

18

Method Development in Analytical HPLC

or observable indicator for the presence or absence of a specific analyte. For quantitative purposes, gravimetric methods (using only a balance as an instrument) and volumetric methods were applied early on as analytical procedures and are still very useful today for specific tasks. In addition, some other techniques such as colorimetry, pH/ conductivity measurement, etc. are also classified as wet chemistry although some instrumentation is utilized for measurements. Colorimetry, for example, uses a colorimeter for comparing the color of a sample with that of a standard, usually after the use of a reagent that generates a specific color with the analytes. Compared to many other instrumental analytical techniques, wet chemistry procedures usually have lower sensitivity, and in many instances, they can be more labor intensive. On the other hand, wet chemistry methods can be simple and cheap to perform, and also can be very informative. To enhance productivity for wet chemistry analyses, automation has been developed for some procedures.

Optical spectroscopy based on light absorption Optical spectroscopy includes several techniques based on adsorption, emission/fluorescence (FL), or scattering of light. The process takes place with changes in energy at molecular (atomic) level following Planck’s law E ¼ hn where E is the energy of the absorbed or emitted photon, n is its frequency and h is Planck’s constant (h ¼ 6.626 1034 J/Hz). The range of frequencies in optical spectroscopy is between about 3$1016 to 3$1011 Hz corresponding to a wavelength l ¼ c=n range from about 10 nm to 1 mm (c is speed of light). Optical spectroscopic techniques are characterized by the dispersion of radiation in a spectrum based on its wavelength. This dispersion is followed by the intensity measurement at different wavelengths. One common optical spectroscopic method is ultraviolet-visible (UV-Vis) absorption spectroscopy. The absorption for this technique takes place in the range from about 10 nm to about 200 nm for vacuum UV spectroscopy, and from about 200 to 750 nm for UV-Vis. This type of absorption is produced by electronic transitions of the molecules going from the ground electronic state into excited states, from where the molecular energy is further dissipated by nonradiative processes such as collisions with other molecules. The light absorption in UV-Vis for a molecular species in solution takes place as a relatively wide absorbance band since vibrational and rotational energy levels are superimposed on the electronic energy levels and the systems have a continuous density of states distribution. Because of these wide absorption bands, UV spectra are in general not diagnostic for the identification of a specific compound. However, UV-Vis light absorption is commonly used for quantitative purposes based on the fact that absorbance of a compound X is related to the molar concentration ½X of the analyte by LamberteBeer law described by the following formula: Al ¼ εl ½XL

(1.4.1)

The molar absorbance εl can have very high values for specific compounds (in some cases obtained using a reagent that generates color with the analyte) and the detection can be very sensitive. UV-Vis absorption is commonly used as a stand-

Introductory concepts related to a chemical analysis

19

alone analytical technique but also as detection in other techniques such as flow injection analysis (FIA). UV-Vis absorption is also a useful detection applied in HPLC and this application will be further discussed in more details in Chapter 3. Infrared (IR) spectroscopy is also a commonly used spectroscopic technique with applications in the analysis of both small molecules as well as polymeric ones. The technique has numerous variants and it is most commonly used in absorption mode, but can be used in emission and reflection modes. IR spectroscopy is based on the characteristic energies for vibrational modes of a molecule corresponding to the range of wavelength from about 750 nm to 1 mm (near, mid, and far IR). Changes in rotational energy levels of molecules can also take place corresponding to frequencies in IR, but in far IR or even microwave range. Only specific vibrational modes of the molecule that are associated with changes in the induced molecular dipole moment are “IR active”. These active modes are determined by the symmetry of the molecule (e.g., Ref. [59]). The result is that IR spectrum may have enough details to be diagnostic for the identification of a molecule or of a specific functional group in the molecule. For this reason, IR spectroscopy is very useful for qualitative analysis. The use of IR for quantitative measurements is less common, although utilized in some cases. Various types of instrumentation are used in IR, including some with a monochromator, and other using a Fourier transform system. IR microscopy is also a common technique and in many cases it can be used as a nondestructive analytical procedure. The IR analysis can be applied to gas samples, liquid samples, or solid samples, the liquid samples being usually sandwiched between two plates of material transparent to IR radiation. “Pure” rotational and rotationale vibrational spectroscopy are used mainly for gas analyses. Novel IR techniques are continuously evolving. Among these are the attenuated total reflection IR, twodimensional IR, discrete frequency IR [60], and laser direct IR microscopy which takes advantage of the development of quantum cascade laser technology and is used for chemical mapping of biological and other types of samples. Other variants of IR microscopy with important utilizations are also known. The use of IR absorption as a detection technique in HPLC has been attempted but it is not commonly used. Because water and other solvents such as methanol which are typically used in the mobile phase of HPLC have very strong and broad absorbance bands in IR, the application of IR as a detection technique in HPLC has significant limitations. Both the measurement in a flow cell of the IR absorption of the HPLC effluent and the drying of the mobile phase before detection were attempted, but both procedures have practical difficulties (e.g., Refs. [61,62]). Another radiation absorption technique used only for metallic elements analysis is atomic absorption spectroscopy. This technique covers basically the same type of analytes as those measured by inductively coupled plasma (ICP) which has advantages over atomic absorption and tends to replace it. The interface of atomic absorption spectrometry with HPLC has been attempted in some special applications (e.g., Refs. [63,64]). Other types of spectroscopic methods based on absorption of radiation are also known. Among those are photoacoustic spectroscopy, Mössbauer spectroscopy, photoelectron spectroscopy, etc. These techniques are used for specific types of analyses.

20

Method Development in Analytical HPLC

Optical spectroscopy based on light emission A number of emission spectroscopic techniques including flame photometry, arc and spark emission spectroscopy, ICP, microwave-induced plasma (MIP), etc. are used for elemental analysis, in particular for metals. Both qualitative and quantitative information can be obtained using those techniques. The solution to be analyzed by ICP is introduced through a nebulizer into a plasma generated by radio frequency typically in argon. ICP uses two types of detection, one based on the emission spectrum of atomic species from the sample (ICP-OES), and the other uses the plasma only to generate ions and a mass spectrometer for detection (ICP-MS). Both ICP types are the most important techniques for analysis of metals in various compounds or of some nonmetals such as arsenic. Some attempts were made to use flame photometry coupled with HPLC [65] but the application is not common. ICP-OES was also reported as detection of the compounds containing metal atoms and separated by HPLC (e.g., Ref. [66]). Two important spectroscopic emission techniques are fluorescence (FL) and chemiluminescence (CL). FL is the process in which a molecule, after absorbing an initial radiation, emits a radiation with a different wavelength than the radiation absorbed (differentiation cannot be made between absorbed and emitted radiation that have the same wavelength). Because of the vibrational and rotational energy levels associated with electronic levels, both absorption and emission in FL occur in a specific wavelength range. FL spectroscopy is commonly used for detection in HPLC and the subject will be further presented in Chapter 3. CL is another technique sometimes used in trace analysis, and useful for detection in HPLC. This technique is based on the emission of light as a result of a special chemical reaction that generates molecules in an excited electronic energetic state. CL differs from FL or phosphorescence since the electronic excited state is the product of a chemical reaction rather than of the absorption of a photon. However, the wavelength of the light emitted by a molecule in CL is the same as in its FL. Light scattering is the process of diffuse deviation of light from a straight trajectory in the medium through which it passes. Various analytical techniques are based on light scattering, such as static light scattering (SLS) that is used to determine absolute molecular mass of polymers and nanoparticles by measuring the scattering intensity after exposure to a high-intensity monochromatic light, usually from a laser. A specific type of light scattering is Raman scattering which is an inelastic type of scattering at molecular level with the change in the wavelength of the incoming light. In this type of scattering, a photon interacts with a molecule by changing its initial vibrational energy state followed by the molecule relaxation to a different vibrational energy level and emission of a photon of different energy from the incoming light. The light frequency in Raman scattering is caused, similar to IR, by changes in vibrational energy states. However, the dependence of Raman scattering on induced molecular dipole moment is different from that in IR spectroscopy. This difference allows vibrational transitions that might not be active in IR to be intense in Raman, making the two techniques complementary. The use of Raman scattering for detection in HPLC has been attempted but it is not common [67].

Introductory concepts related to a chemical analysis

21

Optical nonspectroscopic techniques Among the optical nonspectroscopic methods of analysis can be listed refractometry, polarimetry, circular dichroism, nephelometry, turbidimetry, etc. In some instances, these techniques can be accompanied with a spectroscopic component such as a monochromatic light (usually with l ¼ 589 nm, the sodium D line) used to measure refractive index, or the use of circular dichroism spectroscopy. Among these techniques, the monitoring of the refractive index of a solution passing through a flow-cell can be used as a detection technique in HPLC (see Chapter 3). Turbidimetry and nephelometry are used to measure the “cloudiness” of a solution. In turbidimetry, the loss of intensity of a transmitted light is measured while passing through the cloudy solution while in nephelometry the intensity of the scattered light is measured. The measurement of the light scattering has applications such as for the measurement of concentration of polymer solutions, and in HPLC a special type of light scattering technique indicated as evaporative light scattering (ELS) can be used for detection. The technique can be applied for compounds that do not have good light absorbance in UV, are not fluorescent, and may be difficult to ionize. In ELS detectors, the effluent from the HPLC is injected in the form of a spray from a nebulizer into a heated tube (drift tube), where also a nebulizer gas is introduced. Volatile molecules in the drift tube form vapors, while nonvolatile molecules form a fine mist and the scattered light from this mist is recorded and used for detection. Further discussion on ELS detectors for HPLC is given in Chapter 3.

X-ray spectroscopy X-ray is an electromagnetic radiation with frequencies between 3$1016 and 3$1019 Hz. X-rays are generated when matter is irradiated by a beam of high-energy charged particles such as electrons. The result of irradiation is the production of a continuous spectrum of X-rays known as white radiation, and of characteristic X-rays for the irradiated material (e.g., a metal). The characteristic X-rays consist of specific frequencies of high intensity and are the result of transitions of an electron from a higher atomic level dropping to a vacant inner electronic level (with principal quantum numbers 1, 2, 3). X-ray spectroscopy involves methods similar to other optical methods based on absorption or emission. In addition, diffraction of X-rays and wavelength dispersive X-ray spectroscopy are used to determine crystallographic structure of materials. An important analytical application of X-ray is X-ray FL which is based on the emission of characteristic “secondary” X-rays from a material that has been excited by being bombarded with high-energy X-rays. This technique is a nondestructive method for elemental analysis. Besides X-ray diffraction and X-ray FL, other X-ray techniques are utilized in practice for analytical purposes. Among these are several electron spectroscopy techniques (X-ray photoelectron spectroscopy or ESCA, electron impact spectroscopy, Auger spectroscopy). The use of X-ray as a detection technique for chromatography is uncommon. One example of use of X-ray FL is the identification of phosphorus, bromine, and sulfur containing compounds separated by thin layer chromatography (TLC) [68]. X-rays were also used as an ionization source in atmospheric pressure photoionization for MS detection used in liquid chromatography [69].

22

Method Development in Analytical HPLC

Mass spectrometry Mass spectrometry (or spectroscopy) (MS) is an analytical procedure in which the analyte molecules are converted into gas-phase ions, those are separated based on their mass-to-charge ratio (M/z), and the ion abundance is measured. The plot of the ion abundance as a function of the mass-to-charge ratio generates a mass spectrum that can be used for compounds identification. The total ion abundance, or the abundance of a specific ion from the spectrum, can be used for quantitative purposes. Ion separation is achieved using mass spectrometers that can be of different types such as quadrupole, ion trap, time-of-flight, magnetic sector, Wein filter, ion cyclotron resonance, etc. The gas-phase ions in a mass spectrometer are formed in an ion source which can also be of various types. For gases and vapors, for example, electron bombardment ionization (EI) and chemical ionization (CI) are used. For liquid samples as those generated in HPLC, two techniques indicated as electrospray ionization and atmospheric-pressure chemical ionization are used. Also, for liquid and solid samples, a technique indicated as matrix-assisted laser desorption/ionization (MALDI) is utilized. Other procedures to generate ions include ICP, fast atom bombardment, photoionization, thermospray, etc. MS is frequently used hyphenated with a separation technique such as gas chromatography, liquid chromatography, supercritical fluid chromatography (SFC), or capillary electrophoresis (CE). Due to the exceptional sensitivity that can be achieved by MS detection, and the additional capability of MS to separate the analyte ions by their (M/z) values, the chromatographic techniques using MS for detection generated some of the most powerful known analytical methods. In gas chromatography coupled with mass spectrometry (GC-MS), for example, the molecules of the sample are carried in gas form to the MS detector where ions are formed using either EI or CI. In the type of ionization indicated as EIþ, a molecule X typically undergoes a reaction of the type: X þ e / X •þ þ 2e

(1.4.2)

This type of ionization is the most common, although negative radical ions can be generated from some compounds (EI-ionization). The energy typically used for ionization is 70 eV and the ions X •þ are in an excited energetic state. Being unstable, these • ions decompose with formation of fragment ions Aþ i and radicals Bi . The fragment þ þ ions Ai are commonly even electron ions EE (odd-electron fragments OE•þ are also possible), the fragmentation taking place with the formation of a number of fragment ions as indicated by the reaction: • X •þ / Aþ i þ Bi

(1.4.3)

More than one fragmentation path is common for a given molecular ion, and the nature and abundance of the set of fragments is characteristic for a given compound. The ions generated from this fragmentation are separated in the mass spectrometer and recorded. The mass spectrum generated in this way is diagnostic for many compounds.

Introductory concepts related to a chemical analysis

23

The separation step of the compounds produced by the GC is, however, essential for identification since overlaid spectra of the compounds from a mixture is too complex and cannot be used for identification. MS is also a detection technique of exceptional utility in HPLC. The use of tandem mass spectrometers (MS/MS) and of highresolution MS (HR-MS) brings significant advantages for the application of these techniques as detectors for liquid chromatography. MS has also some uses without being hyphenated with a separation part. One such application is the direct connection of a pyrolyzer with MS in a technique indicated as Py-MS (see e.g., Ref. [70]). Of high utility in particular for the analysis of biomolecules is MALDI technique. This technique uses a laser beam interacting with an energy-absorbing matrix where the analyte is present. This process creates ions from the analyte with minimal fragmentation. The ions are further analyzed using a mass spectrometer. MALDI and MALDI imaging found many applications in microbiology and medicine [71]. MALDI has also been used successfully as detection for gel electrophoresis and CE [72]. MS is also used for the analysis of ions generated by ICP. The development of ICPMS improved the ICP sensitivity such that the technique can detect traces of metallic compounds. Although ICP-MS provides excellent selectivity by itself, the addition of HPLC for sample components separation has been utilized and it provided a powerful technique for elemental speciation studies [73,74]. The technique is even capable of detecting carbon thus providing a universal technique for detecting organic compounds eluting from the chromatographic column [75].

Electrochemical methods Electrochemical methods include a range of techniques that use electrical properties of a solution for the measurement of analytes. These include methods where an electrolysis takes place and either the weight of deposited material is measured (electrogravimetric) or the amount of current in measured (coulometric), methods based on current-voltage dependence (polarography, voltammetry, oscillography, stripping methods), methods based on the measurement of current intensity during an oxidation or reduction (amperometric methods at fixed or pulsed potential), and methods based on measurement of electrical conductivity of a solution (conductometry). One very useful type of electrochemical analytical procedure is based on potentiometric measurements using ion selective (or specific) electrodes (ISE). In this technique, a specially constructed electrode generates an electrical potential dependent on the concentration of a specific ion in a solution. A common type of such electrode in the one used for the pH measurement (sensitive to H3Oþ ions), but similar electrodes have been developed for many ions (both cations such as Kþ, Caþ2, NHþ 4 etc., and anions such as Cl, CN, SCN, etc.). Through the utilization of enzymes and a double mechanism of functioning, these electrodes can be used for the detection of neutral molecules such as in the case of glucose selective detector (e.g., Ref. [76]). Electrochemical methods of analysis are widely utilized being typically rapid and requiring inexpensive instrumentation. They can be utilized stand alone, but are also used as detection techniques for HPLC (as well as for electrophoresis). Further discussion on the use of electrochemical detection in HPLC is presented in Chapter 3.

24

Method Development in Analytical HPLC

Nuclear magnetic resonance NMR is a spectroscopic technique applied mainly for structural elucidations. NMR can be used for the study of molecules containing atoms with an odd number of protons and/or neutrons, typically 1H, or 13C, but also for atoms such as 14N, 19F, etc. When placed in a magnetic field, NMR active nuclei absorb electromagnetic radiation at a specific frequency. The energy of the radiation absorbed is caused by the difference DE between energies of the nuclear spin states, and the frequency of the signal is proportional to the strength of the magnetic field. The field strength reaching the nucleus is affected by the electronic environment of the nucleus which is producing what is known as a chemical shift, d. The chemical shift d depends on the molecule structure and as a result it provides structural information for the molecule. Similar to other spectroscopic techniques, for the utilization of NMR spectrum in compound identification and structure elucidation, the spectrum must be obtained for a pure compound. Compared to MS, this limitation of applicability to pure compounds is difficult to overcome for NMR and the attempts to couple NMR with HPLC encounters a number of problems [77]. One of these problems is the much lower sensitivity of NMR compared to MS, even for 1H NMR which is about 5000 times more sensitive than 13C NMR. Another problem in coupling NMR with HPLC is related to the content of 1H (in the case of 1H NMR) in common HPLC solvents such as water, methanol, or acetonitrile when the mobile phase would generate a huge signal compared to the analytes. For 1H NMR, the problem can be overcome using fully deuterated solvents as mobile phase for the HPLC, but this is still is a major inconvenience [78]. Some partial success in hyphenating NMR with HPLC has been obtained by using various techniques such as “stop flow” mode that allows a longer time for the NMR signal acquisition which increases sensitivity, and by the use of special data processing techniques for the suppression of the solvent signal. HPLC-NMR coupling was also attempted by using collection of fractions of HPLC eluate that contain the analyte in special loops that are subsequently analyzed by NMR, or using solid phase extraction traps for specific eluates that are later eluted with deuterated solvents and analyzed, etc. [79]. Other applications were also reported in the literature [80].

Radiochemical methods Radiochemical methods are based on the measurement of various types of radiation such as alpha (a, helium nuclei carrying a 2þ charge), beta (b, consisting of electrons), gamma (g, electromagnetic radiation having the wavelength less than 103 nm), and neutrons. Radioactivity can be detected and measured at extremely low intensities. Radiochemical methods have a variety of applications. For example, neutronactivation analysis can be used for the quantitation of elements in many materials independent of the chemical form of a sample. However, the method being based on neutron activation requires a source of neutrons (e.g., a nuclear reactor or a special source containing a radioactive element emitting neutrons). Numerous other utilizations of radiochemical methods are known, including the tracing of the presence of a specific analyte when the sample is spiked with a small quantity of the analyte in

Introductory concepts related to a chemical analysis

25

radioactive form. This tracing procedure can be used in connection with chromatographic separations. b-Rays absorption is used in the electron capture detector (ECD). This is a device used in gas chromatography, and detects molecules in a gas by their property of attaching electrons via electron capture ionization. Radiometric detection has also been used after chromatographic separation of compounds containing radioactive nuclides.

Thermal methods of analysis Thermal methods are designed to study various properties of samples as a function of changes in temperature. Among these are thermogravimetric analysis in which the mass variation of a sample is monitored as sample temperature is increased, differential thermal analysis in which the temperature difference is measured between a sample and a standard during heating, differential scanning calorimetry (DSC) in which the difference in the heat absorption or emission of the sample compared to a standard is measured, thermomechanical analysis in which a mechanical property variation with temperature is measured, etc. Due to its high sensitivity, DSC became useful for the characterization of phase change in materials [81] and has been used for the characterization of the stationary phases in HPLC and their interaction with the analytes [82,83]. Thermal methods of analysis are not otherwise directly associated with separation methods except for the use of the thermal conductivity detector (TCD) in gas chromatography. TCD is based on the variation in thermal conductivity of the gas effluent from a GC column when analytes are eluting. A special thermal method of analysis can be considered pyrolysis associated with some type of separation and detection which is in many cases GC or GC-MS. In this technique, a compound is subject to flash heating at a specific high temperature (e.g., between 350 and 900 C) generating a number of smaller molecules. The decomposition products usually form a complex mixture but the nature of components of the mixture depends on the pyrolyzed compound and can be used for its identification. Pyrolysis is commonly utilized in the analysis of polymers for changing a nonvolatile material in small molecules that can be analyzed using, for example, GC-MS (e.g., Refs. [70,84]) and be characteristic for the polymer. Pyrolysis is also useful for the understanding the formation by thermal decompositions of compounds that can be toxic or environmentally harmful (e.g., Ref. [85]).

Other nonhyphenated analytical techniques A number of other techniques are also used in chemical analysis. One example is electron paramagnetic resonance (EPR), which is a spectroscopic method utilized for the analysis of molecules having unpaired electrons. The basic concept of EPR is analogous to those of NMR, but the spins excited are those of the electrons instead of the atomic nuclei. EPR spectroscopy is used in particular for studying organic radicals. EPR coupled with HPLC has also been experimented. Several other analytical techniques such as Mössbauer spectroscopy, secondary-ion mass spectrometry, etc. are used in chemical analysis.

26

Method Development in Analytical HPLC

Chromatographic methods of analysis Chromatographic techniques are typical examples of hyphenated methods. These techniques include a number of procedures allowing the separation and also detection/measurement of different molecular species from a sample. For a chromatographic separation, the sample is introduced (injected) in a flowing mobile phase that passes a stationary phase. The stationary phase retains stronger or weaker different passing molecular species and releases them back into the mobile phase, separated in time. The mobile phase in chromatography can be a gas, a liquid, or a supercritical fluid. In chromatographic techniques, the eluted molecules from the stationary phase (usually in the form of a column) differ from the mobile phase components by certain physico-chemical properties, which makes them detectable. The hyphenation of separation with various types of detection led to the very useful analytical techniques. Gas chromatography (GC) uses a gas as mobile phase (usually H2 or He). The stationary phase is placed in a column and the most common chromatographic columns in GC are capillary columns, in which the stationary phase is coated on the inner wall of a long capillary tube made of fused silica. Columns containing particles (packed columns) were initially utilized in GC separations, but their utilization became infrequent. GC is applicable for the analysis of compounds that have volatility such that they can be placed in the form of gas/vapor phase in the mobile phase. This is achieved by different procedures depending on the nature of the sample. Gasses are usually place in the mobile phase using an injection loop, while liquids are introduced with a syringe into a heated injection port. Other type of placing the sample in the mobile phase are known, such as desorption from a solid phase (e.g., in solid phase microextraction or SPME). Common detectors used in GC include the TCD based on the change in thermal conductivity of the gas that passes the detector when from pure H2 or He starts containing the analyte vapors, flame ionization detector (FID) based on the quantity of ions formed in a flame from the analytes, thermionic type detectors (such as nitrogen phosphorus detector or NPD), mass spectrometer (MS or MS/MS) based on the type of ions formed in an MS, ECD based on the capability of analytes to capture b rays, IR detector based on IR absorption of the analytes (either direct or after using cryodeposition), etc. The sensitivity of the detectors used in GC is compound dependent, and also detector dependent. TCD type detector has the limit of detection in the range 2$105 to 108 g analyte (for 50 mL/min gas flow) FID in the range 2$108 to 1011 g, ECD for selected compounds in the range 1010 to 1013 g, MS in the range 109 to 1011 g, MS/MS in the range 1011 to 1014 g, etc. The most useful detection for the GC separations is very likely MS, the technique being known as GC-MS. Because the mass spectrum generated by EIþ is diagnostic for most compounds, the technique is widely applied for qualitative analysis. The compound identifications are usually achieved by comparisons with standard spectra found in mass spectral libraries. The development of large libraries with standard spectra (over 675,000) and of algorithms for automatic searches in mass spectral libraries such NIST [86] made the use of GC-MS an invaluable tool for compound identification. In addition, GC-MS can be easily used for quantitative purposes based on the measurement of ion abundance and corresponding calibrations. GC-MS methods can be used for simultaneous

Introductory concepts related to a chemical analysis

27

analysis of many components in a sample, and the use of automation (autosamplers) allows sequential analysis of many samples without human intervention. In some GC applications hundreds of compounds can be separated and this high separation efficiency is even higher when using GC x GC systems (comprehensive two-dimensional GC). The main limitation for GC is however the requirement for the analytes to have volatility and this restricts the utilization of gas chromatography such that it cannot be applied to nonvolatile molecules or those which decompose when heated. Effort has been made to extend the volatility of certain molecules, for example by derivatization (e.g., Ref. [1]). The run time in GC can be as fast as less than 1 min or as long as 80e90 min depending on the sample components and GC separation parameters (column type, temperature gradient in GC). Liquid chromatography (LC) includes several chromatographic procedures where the mobile phase is a liquid. The most commonly utilized type of LC is high performance (or pressure) liquid chromatography (HPLC). HPLC includes a set of techniques with different separation principle and using a variety of detection techniques. For this reason, HPLC has high versatility, can be used for the analysis of nonpolymeric and polymeric molecules, organic and inorganic, polar and nonpolar, ionic, etc. and it is applicable to the analysis of almost any type of material except those that cannot be solubilized in any solvent or conditions. The diversity of sample may include environmental, pharmaceutical, biological, related to food and agricultural products, etc. HPLC is very likely the most utilized technique in chemical analysis in particular for quantitation purposes. HPLC uses a stationary phase in the form of a column (or cartridge) packed with very small porous particles (1.7e10.0 mm in diameter), or a porous monolithic material, with the liquid mobile phase (or eluent) moving through the column with the help of a high-pressure pump. The most common procedure to inject the sample in the liquid flow is using an injection loop, and the process can be automated. Much less utilized techniques to include the sample in the HPLC flow are the septum injector or stop flow injector. The mobile phase in HPLC delivered by the pumps is made from a solvent or a mixture of solvents to which some additives may be added, such as buffers. The composition of the mobile phase can be kept constant (isocratic) during the separation or they can be changed following a specific program. In this case, the separation is indicated as taking place in gradient conditions. The detection for the analytes in HPLC can be based on refractive index, UV absorption, FL, type of ions formed in a mass spectrometer, etc. The different types of detection are usually indicated as LC-UV, LC-FLD, LC-MS, LC-MS/MS, etc. HPLC is used in particular for quantitative measurements because of its sensitivity, excellent reproducibility, and robustness (the quality of the analysis to not be influenced by small experimental modifications). The sensitivity of detection in HPLC is different for different detectors, and is also compound dependent. One of the most sensitive detectors for LC is MS/MS that can measure for specific compounds as low as 1014 g material. Besides the specific time of retention of different compounds, (retention time) HPLC can also be used in various ways for qualitative analysis. Depending on the detection used, the qualitative information from HPLC can be or not sufficient for compound identification. However, even with the use of MS/MS detection it may be difficult to identify an unknown compounds and LC-MS/MS is more frequently

28

Method Development in Analytical HPLC

utilized for positive confirmation of the presence or absence of a known compound. Similar to GC, HPLC is capable of simultaneous analysis of many components in a sample, can handle samples with a complex matrix, and the use of automation (autosamplers) allows sequential analysis of numerous samples without human intervention. The run time for HPLC separation is usually is in the range between 2e3 and 15e20 min, although shorter or longer separation times are utilized in certain applications. A discussion on various aspects of HPLC and its utilization is given through the whole present book. Besides HPLC other forms of liquid chromatography are known, such as open column LC and TLC. Open column chromatography is used mainly for semipreparative purposes, but TLC has useful analytical applications. The stationary phase in TLC consists of a thin layer of a solid adsorbing material coated on an inert plate of glass, plastic, or metal, and a liquid mobile phase. The technique is mainly used for quick qualitative analysis of mixtures. Quantitation using TLC is less precise compared to other chromatographic techniques, but can be achieved using densitometry (measurement of optical density of spots created on a TLC plate by the analytes). Another chromatographic method used for analytical purposes is SFC. This technique uses as mobile phase a supercritical fluid and the stationary phase either in a packed column similar to HPLC or a capillary column. The properties of supercritical fluids are intermediate between gases and liquids and can be varied by small changes in the temperature and pressure. The changes in these properties also modify the solvating properties of the mobile phase and influence the separation. The most commonly used fluid in SFC is CO2, which is not polar [87]. Modifiers, such as small amounts of alcohols, can be added to the mobile phase for changing its polarity. SFC is used mainly in analytical applications for nonpolar compounds. This is one of the limitations of SFC contributing to its less frequent utilization compared to HPLC. The detection in SFC can be done with flame-based detectors such as FID and NPD (similar to GC) or based on UV light absorption, FL, or MS (similar to LC). One other type of chromatographic technique sometimes used for analytical purposes is countercurrent chromatography. This technique uses a liquid stationary phase and has several variants. However, this technique is used mainly for preparatory purposes.

Other methods of analysis which are including an independent separation step Besides chromatography, an important group of hyphenated techniques involving a separation are the electro separations. This group includes several electrophoretic techniques, electrochromatography, electrodialysis, electrofiltration, and a special technique indicated as ion mobility (IM). Some of these techniques are used for qualitative and quantitative analysis. A common such technique is electrophoresis. Electrophoresis uses a uniform electric field (at high voltage) to produce in a fluid the migration the molecules or particles (nanoparticles) of a sample. The migration in the electric field is caused by the presence of a charged interface between the

Introductory concepts related to a chemical analysis

29

migration molecules or particle surface and the surrounding fluid. The typical supporting media for the fluid include paper or various gels (agarose, polyacrylamide, starch, etc.). The analytes are separated based on their ionic mobility and partitioning into the supporting phase via noncovalent interactions. The process is controlled by the size and electric charge of the molecule. The supporting medium can be placed in the form of a thin layer on a plate. At the end of the separation, the analytes are visualized for example by staining. The molecules are identified by their migration distance and can be quantitated using densitometry. Electrophoresis is usually applied for the analysis of macromolecules such as proteins and nucleic acids. A special group of electrophoretic techniques is known under the general name of CE. This group includes capillary zone electrophoresis (CZE) capillary gel electrophoresis, capillary isoelectric focusing, capillary isotachophoresis, micellar electrokinetic chromatography. These techniques are using for molecular migration a medium placed in a capillary tube. Similar to planar electrophoresis, in CE, the migration of the analytes is caused by an electric field that is applied between the two ends of the capillary. In CE, the analytes ions, positive or negative, are pulled through the capillary in the same direction by an electroosmotic flow. The analytes separate as they migrate due to their electrophoretic mobility, and are detected near the outlet end of the capillary. Detection in this case can be performed using UV or UV-Vis absorption, laser-induced fluorescence (LIF), MS, or Raman spectroscopy and can generate very informative results. For example, CZE-LIF can detect specific compounds at levels as low as 1018 to 1021 mol. Instruments having an array of capillaries can be used to analyze simultaneously a number of samples. Electrochromatography is a technique that combines the separation obtained from a flow of a mobile phase through a stationary phase, and the migration of the analytes in an electric field. The technique is usually applied for the separation of macromolecules, the chromatographic part being based on size exclusion, and the electrical migration being similar to electrophoresis. The technique can be applied using capillary columns (capillary electrochromatography [CEC]) when the flow through the “active” stationary phase is not pressure driven but based on an electro-osmotic flow. A number of variants of electrochromatography have been experimented. The technique has various applications many similar to those of CZE. Another electro separation technique that can be used for analytical purposes is IM. The technique is also known as ion mobility spectroscopy. This technique can be hyphenated with HPLC at the front end and with MS at the other end providing an additional stage of separation between HPLC and MS. In this technique, the ions of the analyte are moved in an electrical field against the flow of a carrier buffer gas. Collisions of the analyte and gas molecules have higher probability for the larger molecules of the analyte than for the smaller ones. Because of that, larger molecules will migrate slower than the smaller ones and a separation is achieved based on the molecular size of analyte molecules. The separation in IM occurs at a time scale of milliseconds (in the range 10e1 to 10e3 s). The use of IM associated with HPLC will be further discussed in Chapter 3.

30

Method Development in Analytical HPLC

Key points • •

Chemical analysis can be performed using a wide variety of methods, and some core analyses may include only measurement while other have two hyphenated parts one involving compounds separation and the other measurement. The methods involving separation can be applied for the analysis of mixtures of compounds.

1.5

Introductory information regarding HPLC

The first study on liquid chromatography can be considered the work performed by Tsvet [88], although a few earlier attempts had been made to use liquid chromatography [89]. Tsvet separated the plant pigments xanthophylls, chlorophylls, and carotenoids on a column having calcium carbonate as stationary phase and petrol ether/ ethanol as mobile phase. The foundation of liquid chromatography as a new research field was established later by the contributions of Martin and Synge [90] followed by their development of a new principle of separating mixtures of compounds using new materials as solid supports for separations [91]. Further evolution of liquid chromatography is due to the contributions brought by Kirkland, Snyder, Horwath, and many others, such that the liquid chromatography has been transformed into a performant analytical techniquedHPLC [92e95]. In time, HPLC became a common core analytical technique used for compound separation and measurement (it is a hyphenated technique). HPLC type separations are also used for preparative or semipreparative purposes, as well as for some nonanalytical applications. Practical applications of analytical HPLC cover important human activities such as pharmaceutical development, medical diagnosis, medical research, food and agricultural products characterization, environmental studies, various consumer products evaluation, support for legal concerns, etc. The subject of this book is related to analytical HPLC, while other applications of HPLC are beyond the purpose of this book. Analytical HPLC includes a number of related techniques that use for the compound separation a chromatographic column (or cartridge) containing a stationary phase, and a liquid mobile phase (eluent) that is forced to pass through the column being driven by a pump at high pressure. As previously indicated, the mobile phase in HPLC can have a constant composition during the separation (isocratic separation), or can be changed (gradient separation). The stationary phase in HPLC can be in the form of porous particles (1.7e10.0 mm in diameter), superficially porous particles, or as porous monolithic rods. A variety of high-porosity materials with a large surface area and having many chemically active sites can be used as stationary phase. Depending on the type of chromatography these materials can be bare silica gel, silica gel with bonded octyl groups on its surface, silica gel with bonded octadecyl groups, silica gel with bonded propylamino groups, inorganic or organic porous polymers having ion exchange capability, porous materials having a specific pore dimension, etc. (see Chapter 7). Also, a variety of solvents or solvent mixtures can be used as mobile phase. Common solvents used in HPLC are

Introductory concepts related to a chemical analysis

31

water, methanol, acetonitrile, hexane, etc. as well as mixtures of those (see Chapter 8). The sample components that are together in a liquid-solution (the solutes) are typically injected in the mobile phase and eluted from the chromatographic column based on an equilibrium of the type: Xmo $ Xst

(1.5.1)

10.201

In Eq. 1.5.1 Xmo indicates the analyte in the mobile phase, and Xst the analyte in the stationary phase. Because the molecules of the analyte are not present all the time in the mobile phase, they will be eluted from the column later than a molecule of mobile phase, and different analytes being retained differently will be eluted at different times, indicated as retention times tR . The stationary phase is usually contained in a chromatographic column and it is made from small particles. The eluted molecules from the chromatographic column are detected and measured by one or more detectors. The detectors produce an (electrical) signal that is translated into a graphic output known as a chromatogram. The separated components are displayed as peaks in the chromatogram. Although the initial sample is introduced as a narrow “plug” in the mobile phase flow, due to diffusion and other broadening effects the peaks are wider than the initial plug and have a Gaussian shape (ideally). The signal generated by the detectors may provide only information about the presence and concentration of a sample component, but some detectors also provide structural information. An example of an HPLC chromatogram of a plant extract having some values of the retention times tR marked on the graph is given in Fig. 1.5.1. A plant extract usually contains a mixture of compounds and as shown in Fig. 1.5.1, the chromatogram shows some well-defined peaks but also merged peaks. In a “good” chromatographic separation, a well-defined peak corresponds to a unique compound,

9.621

700

600

100

11.223

200

10.436

8.323

4.752

5.121

300

1.693

400

7.184

2.027

500

0 2

4

6

8

10

12

14

Figure 1.5.1 Example of an HPLC chromatogram of a plant extract showing the retention times tR for selected peaks.

32

Method Development in Analytical HPLC

but this may not be always the case since some compounds in a separation may elute at the same retention time. For some detectors, additional information regarding the peak, such as the associated UV spectrum or mass spectrum, may provide verification for the “purity” of the chromatographic peak. In some cases, spectral information may also be used for the identification of the chemical nature of the eluted peaks. In this way, the chromatographic results are used for qualitative and quantitative analysis as further discussed.

Qualitative analysis based on HPLC Qualitative analysis in HPLC is based on two types of information generated by this analytical method. One type of information is generated by the separation process where the components of a sample are separated based on their nature, and the retention times of each analyte can be used as a parameter specific for a compound. However, retention time also depends on separation conditions and this parameter can be used only for the verification of the presence of a known compound when the separation conditions are not modified, but cannot be used for the unknown identification. The subject of separation in HPLC is reported in many publications (e.g., Ref. [96]), and is also be presented in some details in the present book. The second type of information regarding the nature of the analytes in a sample is generated by some specific detectors used in HPLC. The detectors used in HPLC can be classified as universal detectors, which have the same type of response for any compound, and such detectors cannot be used for compound identification. However, other detectors also contain qualitative information regarding the nature of the analytes. This information is in some cases still not sufficient for the identification of unknown compounds but sufficient for compound “recognition” and in other cases, such as from MS/ MS detectors, the information can be diagnostic for compound identification. The utility of different detectors for qualitative analysis is further discussed in Chapter 3.

Quantitative analysis based on HPLC Quantitative analysis is the main use of HPLC. Once separated, the concentration of the analytes in the sample or their amount in the injected sample can be obtained from the chromatographic peak area (or height). Peak areas (or peak heights) in the chromatogram are dependent on the concentration or the amount of analyte, and quantitation is done by using calibration curves with standards, or by other procedures. The accuracy of quantitation is frequently improved by using an internal standard. The internal standard responds proportionally to changes in the analysis and provides a similar, but not identical, measurement signal. Taking the ratio of analyte signal to internal standard signal and plotting this ratio against the analyte concentrations in the calibration solutions will result in a more accurate calibration curve (see Section 4.5). Quantitative analysis is characterized by several parameters including accuracy of the results, precision of measurements, lowest concentration of an analyte that can be measured, etc. These aspects will be further presented in detail in Sections 4.5 and 4.6. As a general comment, the HPLC methods offer a wide range of capabilities and can be

Introductory concepts related to a chemical analysis

33

developed for the accurate and precise analysis of many types of samples, with analytes content starting from ultralow trace level (e.g., below ng/mL) to high component concentrations. A special type of measurement using HPLC (see Section 4.6) is the measurement of molecular weight of polymers. The purpose of this book is to provide guidance regarding the development of various types of methods used in HPLC analysis.

Key points • • •

HPLC is a hyphenated technique performing both compound separation and detection. HPLC includes a number of related techniques that use for separation a chromatographic column (or cartridge) containing a stationary phase, and a liquid mobile phase (eluent) that is forced to pass through the column being driven by a pump at high pressure. HPLC can be used for the analysis of a wide type of molecules.

1.6

Decision to use an HPLC analytical method

The decision to use an HPLC analytical method should be based on the collected information regarding the purpose of analysis and analysis requirements (precision, requested time for generating results, specific protocol), properties of the sample (analytes and matrix), as well as the available resources necessary for the analysis (see Section 1.3). In some situations, the decision must consider the information reported in the literature for the analysis of the same type or similar type samples. An analytical technique is initially considered as candidate for the core step in an analytical method when it satisfies two conditions: (a) it is appropriate for the analysis of the pure compounds of interest, and (b) it has the sensitivity necessary for the determination of the lowest quantity of analyte estimated to be transferred in a processed sample. Further selection from a number of possible techniques is done based on the chances to achieve a good separation of the analytes and other sample components. Once the potential capability of a technique to analyze the sample is established, a sample preparation is selected such that the matrix of the sample is simplified or eliminated and does not produce interferences in the chosen analysis. Because HPLC covers in fact a set of methods with the separation adaptable to various classes of molecules and with a variety of detection techniques, it is possible to successfully use HPLC to analyze a very large variety of samples. Also, having the capability to analyze samples with a complex matrix, to analyze more than one analyte in a sample, being suitable for quantitation purposes even at trace level concentrations, and being amenable to automation, HPLC is frequently the best technique to solve an analytical problem. However, HPLC has its own limitations such as the need to have the sample in solution and not always having a good capability to identify unknown compounds. Therefore, in order to make the correct decision to use HPLC, it is important to know if this is the best choice, if other analytical techniques are also suitable for a specific analysis or even more appropriate than HPLC. This decision is not always simple since there is more than one aspect that

34

Method Development in Analytical HPLC

characterizes an analytical method. When comparing one method with another, it is possible that not all qualities are in favor of the same method. The comparison may show that one method provides only limited information about the sample but does not require an elaborate sample preparation, or may provide less precise results but requires very short time to perform, may be more labor intensive but does not require expensive instrumentation. One other aspect regarding the choice of an analytical method is that in some instances one single analytical method is not sufficient to solve a problem. For example, because HPLC does not have a high capability to identify the nature of unknown compounds, other methods must be used for this purpose ahead of HPLC use. After the identification of the nature of the compounds, HPLC can be used for quantitation purposes while the other methods may not be able to provide this information, or generates information of lower accuracy or precision. For example, if the sample has adequate volatility, GC-MS is more adequate for the identification of unknown compound, or after the separation and collection of specific fractions from the sample, IR or NMR analysis may provide the answer regarding the nature of the sample components. In some instances, because of unsurpassed sensitivity of HPLC with MS/MS detection, the LC-MS/MS technique can be used first for the detection of specific ions and only after this detection other techniques must be involved for the identification of the corresponding compounds. This procedure can be used for nontargeted analysis for which HPLC may not be able to identify the compound but can detect a specific ion with high sensitivity. The HPLC core analytical method is usually developed initially on pure compounds. In this stage, the verification of separation capability of analytes can be achieved, and if information about the range of concentrations in the sample is available, the test if the detection selected for the analysis are sensitive enough. In this first stage, also some potential interfering components of the matrix are tested and verified if they require or not a sample preparation for their elimination. The true verification that a developed method is adequate comes only when real samples are analyzed, after the sample preparation step.

Comparison of HPLC with several nonhyphenated analytical methods HPLC, although commonly used as method of analysis, is not always the optimum choice. Some comparisons of HPLC with other analytical methods are further presented, in general showing the advantages of HPLC. Because of the complexity and large number of analytical techniques, the presented comparison is far from covering the multitude of possibilities of performing an analysis. In some analysis, even a combination of HPLC with an additional analytical technique can be necessary. Consulting the published information may provide important guidance for the best approach for solving a particular analytical task. Wet chemistry can be selected versus HPLC for same simple analyses that do not require advanced separations. For example, gravimetric and volumetric methods can

Introductory concepts related to a chemical analysis

35

be selected when a single analyte must be measured, the analyte is present at a relatively high concentration and it is present in a simple matrix. The advantage of these methods compared to HPLC is that they are simple, require only an analytical balance as instrumentation, and do not use specific solvents that may be needed for an HPLC analysis. Colorimetric type methods can be chosen versus HPLC, for example, when an analyte is capable of forming an intense color with a specific reagent and colorimetry can show very good sensitivity. Besides utilizing more expensive instrumentation for a similar analysis, for such analysis HPLC may require postcolumn derivatization that complicates the analysis. Wet chemistry methods are in general difficult to apply on complex mixture and when more than one analyte must be analyzed for a sample. Also, wet chemistry methods usually require more intensive manpower, but do not need high qualification of the operators. Optical spectroscopy based on light absorption as stand-alone analytical methods have many applications and they may be preferred for certain tasks to HPLC. The UVVis absorption spectroscopy can be used similarly to colorimetry when an analyte is capable of forming an intense color (or UV absorption) with a specific reagent. For multisample analysis, autosamplers can be adapted and as indicated in Section 1.4, UV-Vis light absorption is used for detection in FIA [97]. However, for the analysis of compounds that have intrinsic UV or visible absorbance the use of HPLC (with UV-Vis detection) may be a much more convenient alternative. Without HPLC, the stand-alone UV-Vis method has strong limitation when more than one analyte must be measured in the same sample, or when samples have a complex matrix. IR spectroscopy does not usually overlap with HPLC applications. The important requirement for HPLC to have the sample in solution is not imposed for IR that can be used on solid materials. This feature makes IR a very useful tool for the analysis of polymeric materials that are difficult to place in solution. For this reason, qualitative analysis of numerous polymers is usually performed using IR and not HPLC. On the other hand, IR spectra are difficult to interpret unless applied on pure compounds and HPLC has significant advantages over IR when quantitative measurements are required by providing much better sensitivity, capability of analyzing samples with complex matrix, multianalyte measurements, etc. IR spectroscopy has in itself numerous other variants and applications. For example, IR microscopy is an invaluable technique for the analysis of various types of samples being a nondestructive technique as indicated in Section 1.4. In cases when analysis must be performed for surface analysis, IR techniques can be applicable but not HPLC. Also, IR may provide qualitative information on unknown compounds and functional group identification that cannot be obtained using HPLC even with MS or MS/MS detection. Only for the measurement of molecular weight (MW) of polymers IR is not applicable and either size exclusion HPLC or SLS are used for this purpose. Optical spectroscopy based on light emission includes a number of useful standalone analytical techniques. Among these, ICP-OES is commonly used for the analysis of compounds containing metals and also some containing nonmetals. Being based on the measurement of light intensity for specific wavelengths that are characteristic for a specific metal, ICP-OES can be used for the analysis of complex multielement samples, without the need for a preliminary separation. Depending on the atom nature

36

Method Development in Analytical HPLC

in the analyte, the sensitivity of ICP-OES can be exceptionally high in the range of parts per billion. This sensitivity is significantly higher than the usual sensitivity obtained for ion chromatography type HPLC with conductivity detection (e.g., Ref. [98]). FL as a stand-alone technique can be the best choice for analysis when an analyte is capable of produce intense FL following the addition of a specific reagent, similar to the use of stand-alone UV-Vis spectroscopy. However, for compounds with intrinsic FL, HPLC (with FL detection) is the preferred method of analysis since it brings the same advantages as it does compared to stand-alone UV-Vis. The use of stand-alone CL is rather similar to the case of the use of FL [99]. However, in many cases, the use of HPLC is necessary for improving selectivity and CL is generated by postcolumn reagent addition (e.g., Ref. [100]). Several light scattering techniques are also useful stand-alone analytical methods. These methods do not overlap too frequently in purpose with HPLC, although in some cases they do, such as in the case of MW measurement that can be done either using size exclusion HPLC or SLS. Raman spectroscopy for example, is used mainly for compound identification and the study of chemical bonding and it is not utilized similar to HPLC for quantitative analysis of multicomponent samples and for measuring many analytes at low concentration levels. Raman spectroscopy also has applications in solid-state physics and in some “fingerprint’ type studies used in biochemical characterizations. Optical nonspectroscopic techniques are successfully used in some analytical procedures without involving separation. For example, both refractometry and polarimetry are used in gemmology for assessing the true nature of gemstones. Also, the techniques can be used for rapid measurement of the concentration of certain solutions in food processing. However, these applications are specialized and do not overlap with the use of HPLC. The use for detection in HPLC of refractive index monitoring and of ELS was indicated in Section 1.4. X-ray spectroscopy practically does not have many applications overlapping with those of HPLC. The only similar goal may be considered related to elemental analysis. The use of X-ray FL is applicable to solid samples and is a nondestructive procedure while ion chromatography HPLC requires a solution for the analysis of certain cations and anions. Mass spectrometry is a very important technique used for detection in HPLC and GC, but as showed in Section 1.4 it is also used without having a separation step before the source of ions of the mass spectrometer. In this category can be included ICP-MS and MALDI, two techniques of high utility that typically do not have the same area of applications as HPLC. The use of ICP-MS for the elemental analysis has excellent selectivity and it is not surpassed by any other analytical method in sensitivity for this purpose. Only for the differentiation of the type of ions (specific ionic group) in which an element is present, HPLC analysis is necessary and ICP-MS is not providing 2  an answer, such as for the differentiation of SO2 4 from SO3 [101], or of ClO2 from  ClO3 [102], etc. MALDI has some utilization similar to HPLC in protein analysis but is mainly geared for detection without separation of larger biogenic molecules having important utilization in proteomics, the identification of biological materials, and tissue imaging.

Introductory concepts related to a chemical analysis

37

Electrochemical methods as group of analytical techniques are sometimes used for the same type of analysis as HPLC. Compared to HPLC, electrochemical methods typically lack the capability of multicompound analysis, may show interferences when applied on samples with a complex matrix and may have lower sensitivity than HPLC with specific detectors. However, they are in general very rapid, simpler, and much less expensive. In the comparison of the group of HPLC technique with the group of electrochemical ones, each group comprised from a variety of techniques, it is not possible to account for every possible application, and there are common examples where an electrochemical technique is more appropriate for a specific analysis compared to an HPLC one, or the other way around. NMR is typically utilized for different applications compared to HPLC. NMR (usually 1H or 13C NMR) is mainly used for structure elucidation (or confirmation) of a compound, while HPLC is used for quantitative analysis of multicomponent samples. NMR is applied on pure compounds and at relatively high concentration, while HPLC, depending on the associated detection technique, can be used for certain compounds at levels as low as 10e20 pg/mL. The utilization of NMR as detection technique in HPLC is commented in Section 1.4. Radiochemical methods may offer extremely sensitive and selective methods of analysis. However, the concerns related to health and environmental issues limit the utilization of radiochemical methods. Thermal methods of analysis are not frequently overlapping with HPLC in their applications. These methods are geared toward evaluation of material properties and not toward quantitative or qualitative analysis of samples.

Comparison of HPLC with other hyphenated analytical methods Gas chromatography (GC) is a common hyphenated analytical technique with a wide range of practical applications. The GC separation can be hyphenated with various detectors, the one providing the most information about a sample being the MS. As indicated in Section 4.3, GC-MS has excellent capability of identification of the nature of compounds in a sample. However, the requirement for GC to analyze only compounds capable to exist in gas/vapor phase makes this technique unusable for a wide range of molecules that are not volatile or cannot be chemically modified (e.g., by derivatization) to become volatile. Unfortunately, in this category are many biogenic compounds that are not volatile and cannot be modified to become volatile. The restriction of volatility is not imposed to HPLC which can be utilized to a much wider range of types of molecules than GC. Regarding sensitivity, there are highly sensitive detectors for GC (see Section 1.4), and quantitation can be performed using calibrations. For the analysis of gases, GC is usually the preferred method. For many volatile molecules, the selection between GC or HPLC as a method of analysis depends on the details of analysis requirements or in some instances it can be performed on either method. For the analysis of nonvolatile molecules, GC is not applicable. SFC has a number of applications that are overlapping with those of HPLC. SFC has certain advantages compared to HPLC such as not using a liquid solvent that

38

Method Development in Analytical HPLC

can represent an environmental hazard, and in some instances providing a better separation than HPLC and in a shorter period of time. On the other hand, the hydrophobic character of supercritical CO2 (even when using polar modifiers) represents a major limitation of the type of sample that can be analyzed by SFC and the technique is usually limited to the analysis of nonpolar materials such as fats, and to the chiral separations that do not require water in the mobile phase. Also, some problems regarding reproducibility were noted when using SFC [103]. Electrophoreses (planar) providing a different type of separation than HPLC also have usually a different range of applications. While HPLC is a universal technique, electrophoresis is geared toward much more specific tasks such as the separation of macromolecules that can bare electric surface charges. The precision and sensitivity of quantitation using planar electrophoresis is much lower than that of HPLC. CE on the other hand can provide exceptional sensitivity for certain compounds, can analyze very small quantity of sample, and provide very good separation. The precision for quantitative analysis of CE is in many instances significantly lower than for HPLC and for this reason the technique has applications that are usually different from HPLC. Depending on the purpose of analysis, nature of the analytes, required precision of results, etc., CE still can be a technique of choice for some analyses. Among the main applications of CE is the analysis of amino acids [104] and DNA sequencing. The utilization of smaller volumes of mobile phase makes CE a preferred alternative technique in many applications where restricting solvent utilization is desired (green chemistry) [105,106]. Electrochromatography is a technique with applications similar to electrophoresis, and similar characteristics with this technique. For example, CEC is using a liquid mobile phase that is driven by electroosmosis through a capillary containing a chromatographic stationary phase. In this way, CEC is a combination of CE and HPLC. The technique encounters similar difficulties as electrophoresis such as Joule heating of the migration media, generation of gases by electrolysis, irreproducibility of the injection volume, etc.

The goals of developing an HPLC analytical method In developing an analytical method, several goals must be attained (see Section 9.1). One goal is obviously to develop an analytical method that resolves correctly the given analytical problem. Several aspects regarding what means “correctly” are further discussed in relation to the validation of an analytical method presented in Chapter 11. The standard steps regarding the validation including selectivity and specificity, precision, reproducibility, accuracy, range of calibrations, linear range, limit of detection (LOD), limit of quantitation (LOQ), recovery of the analyte(s), robustness, ruggedness, and stability of results will be detailed in that discussion. However, other goals in developing an analytical method must be attained. One such goal is to develop a method that is efficient requiring limited sample preparation, and having high productivity in the sense that can process a large number of samples in a short period of time (high-throughput feature). Another goal is related to the cost of the method. The cost of

Introductory concepts related to a chemical analysis

39

reagents, mobile phase price and consumption, cost of new analytical columns, and the need for the acquisition of new equipment should be as low as possible. However, in the goal of cost reduction, an appropriate balance should always be considered between the quality of the results and the cost saving. For example, buying new guard columns that would add some cost for a new method may be more appropriate than using the column without guard column and altering the analytical column quality after running a small number of samples. In addition, the method must be adjusted to the capabilities, equipment, and expertise existent in the laboratory. For example, MS detectors, in particular MS/MS, is usually offering an excellent alternative for HPLC detection, however, such equipment is usually expensive. If such equipment is available in the laboratory, its use is recommended, but if new such equipment would be necessary, a different, simpler method may be the recommended alternative. Further discussion can be made regarding the method optimization (additionally presented in Chapter 10) and the cost of manpower and utilized funds to optimize an analytical method. An optimized method can be a better alternative to a method that “just works”, but in some instances the optimization may not be essential, and improvements on a method that “just works” can be done during the application of the method when analyzing real samples. The time required for developing a method, as well as the urgency of having the method functional at a specific date should always be considered when developing an analytical method.

Key points •



Compared with other core analytical techniques HPLC has many advantages including capability to be applied to multi component samples, to analyze more compounds in one analysis, to be used for a wide variety of molecular types, can have very high sensitivity and precision, and it is amenable to automation. In the development of an analytical method, the main goal is to have a method that resolves correctly the given analytical problem, but some other aspects related to cost, efficiency, and timing must also be considered.

1.7

Introduction to data processing in HPLC analysis

Most applications of HPLC are related to the generation of quantitative data regarding various analytes. These data are based on measurements of the analytical signals such as peak areas in the chromatograms, or of other parameters characterizing the HPLC process. This section presents some introductory concepts related to data processing commonly utilized in the analytical laboratory. The quantitative data are typically evaluated using statistical concepts. This is possible because it is common to obtain the quantitative data from a number of measurements. Statistics is the topic of many publications (e.g., Refs. [107,108]), and only a few elementary concepts on this subject as applied to analytical data are further discussed.

40

Method Development in Analytical HPLC

Average and standard deviation

  The results xj obtained for a variable x when it is measured a number of times are typically scattered around a specific value, the measurements being always affected by errors. The errors of measurements are classified as systematic (determinate) or as random (see, e.g., Ref. [109]), both types of errors being at the same time possible for measurements. Systematic errors are generated by a specific cause, and they affect the accuracy of the results. They are typically detected by the comparison of the results to be evaluated with the results from different methods of analysis or with known values for the measured analytes. Random errors do not have an assigned cause and they are usually studied using statistical procedures. For the random errors, it can be assumed that they are scattered within a continuous range of values. Therefore, the measurement of one variable x can generate any values in this continuous range. However, only specific values fx1 ; x2 ; x3 ; .:xn g are obtained when a set of measurements is performed, this set being indicated as a sample of the random variable x. All measurements (which are infinite in number) would generate an ideal measurement set, this set being called a population. (To differentiate the statistical terms sample and population from their use in the common language, they will be italicized). For a sample fx1 ; x2 ; x3 ; .:xn g which contains n measurements of x, two important parameters can be indicated. One such parameter is the average (or mean) m of the measurements and the other is the standard deviation s. Standard deviation characterizes the distribution of measurements about the mean. The expressions for m and s are given by the formulas: n P



xj

j¼1

(1.7.1)

n

and:



vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uP 2 un  xj  m u tj¼1

(1.7.2)

n1

Both m and s are important parameters for a set of data, the value of s, for example, characterizing the precision of a set of data and ultimately of an analytical procedure used to generate those data. From the values of m and s a common parameter is generated, indicated as relative standard deviation RSD, which is frequently expressed in percent. RSD and RSD% are expressed by the formulas: RSD ¼

s m

and

RSD% ¼

s $100 m

(1.7.3)

Introductory concepts related to a chemical analysis

41

Similar to the average m and standard deviation s for a sample, equivalent parameters can be defined for the infinite set population. However not being possible to know all the values in the population, the mean of the population m and the standard deviation of the population s are defined as limits by the expressions: n P

m ¼ lim

xj

j¼1

n vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 un  uP u tj¼1 xj  m

n/N

s ¼ lim

n1

n/N

(1.7.4)

(1.7.5)

Since an infinite number of measurements cannot be performed, the true values for m and for s are not known, and they are approximated with m and s for large number of measurements n. The value s2 is indicated as variance. The results for a population are frequently obtained from a number of samples (e.g.,       p samples) xA1 ; xA2 ; .xAn , xB1 ; xB2 ; .xBn , .. xP1 ; xP2 ; .xPn , each one with its own mean mA , mB , .. mP . The set of p samples are characterized by an average mean m given by the following formula: p P



mj

j¼1

(1.7.6)

p

Because of associativity of addition, m for p samples is equal with m for a single sample with n$p measurements and both can be used for the approximation of m. A standard deviation of the mean m can also be defined and the variability across multiple samples of a population is characterized by a standard error “se” given by the following formula: s s se ¼ pffiffiffi z pffiffiffi n n

(1.7.7)

The errors affecting each measurement in a population (and therefore in a sample) are not equal some errors being larger and other smaller. It is expected that larger random errors will be less frequent than small errors. The result is that the distribution of measurements around their mean is dependent on the magnitude of their error. The relative frequency of occurrence of random errors in large sets of measurements is described by the following expression: f ðxÞ ¼

1 ð2ps2 Þ

 2

2 2s exp  x  m = 1=2

(1.7.8)

42

Method Development in Analytical HPLC

Eq. 1.7.8 is known as Gaussian density function and it describes the probability of obtaining an error s for a measurement with mean of the population m. This frequency function f ðxÞ shows that the point of maximum frequency is obtained for the mean, when x ¼ m, the distribution of positive and negative errors is symmetrical, and as the magnitude of the deviation from the mean increases, frequency of occurrence decreases exponentially. The area under the curve f ðxÞ for x < a gives a cumulative frequency distribution expressed by the formula: Z FðaÞ ¼

a N

f ðxÞdx

(1.7.9)

The cumulative frequency distribution is equal to the probability P ¼ FðaÞ for x to have a value below a in any measurement (P has values between 0 and 1, but it can also be indicated as % with values between 0% and 100%). The integral of f ðxÞ over the whole space for values of x gives P ¼ 1 and for a limited region of values, P < 1. The errors with the relative frequency of occurrence given by rel. 1.7.8 have a socalled normal distribution N ðm; sÞ. The replacement of variable x with a new related variable defined by the formula:   z¼ x  m =s

(1.7.10)

will generate the normal distribution N ð0; 1Þ. With the help of variable z it is possible to evaluate how close the values of m are to any measured x for a certain population. For the variable z given by Eq. 1.7.10, two values za=2 and z1a=2 (with za=2 ¼ ⎼z1a=2 ) can be found such that the probability for z of being inside the interval (za=2 , z1a=2 ) is equal with P and outside the interval the probability is a. This interval (za=2 , z1a=2 ) is indicated as confidence interval. This is translated into having a probability P ¼ 1  a for having za=2 < z < z1a=2 or using Eq. 1.7.10 for having: za=2
s2 cannot indicate that second procedure is more precise than the first. For a proper comparison, the variable F defined as: F¼

s21 s22

(1.7.32)

that has an F-distribution must be utilized for comparison. Taking d1 ¼ n1  1 and d2 ¼ n2  1, and selecting a probability P and a resulting a ¼ 1  P, two points Fðd2 ; d1 ; 1  a =2Þ and Fðd1 ; d2 ; 1  a =2Þ can be found in tables for Fðd1 ; d2 ; aÞ for given d1 ; d2 , and a (e.g., Ref. [107]) such that if the following expression is fulfilled: 1 < F < Fðd1 ; d2 ; 1  a = 2Þ Fðd2 ; d12 ; 1  a=2Þ

(1.7.33)

it can be concluded that the two procedures (analytical methods) have the same precision and otherwise, the methods are different. The comparison must be further made between s21 and s22 •Fðd1 ; d2 ; 1  a =2Þ to decide which one generates a lower value. Manual statistical evaluation of data is usually laborious, and computer programs such as ANOVA (available in the statistics package in Microsoft Excel) are usually applied for comparing results. ANOVA package allows, for example, to determine if two or more sets of data are significantly different or not based on the F-test. In ANOVA the variance s2 for two or more groups of data are compared (for a specified probability P ¼ 1a). If the variance within groups is smaller than the variance between groups, the F-test will find a higher F value, and therefore a higher likelihood that the difference observed is real and not due to random errors. ANOVA can evaluate the data dependence on one variable (one way ANOVA) or two variables (two-way ANOVA) (e.g., Ref. [115]).

Least square regression The linear dependence of the concentration (or the amount) of an analyte on the signal generated by the analyte is a common procedure for quantitative analysis. A common type of dependence has a linear form of the type: x ¼ a þ by

(1.7.34)

Eq. 1.7.34 which is usually applied for analytical calibrations is an empirical relation (indicated as measurement function), which permits the calculation of the values

48

Method Development in Analytical HPLC

of the amount (x-variable) of a substance in a sample, from the measured values of the analytical signal (y-variable). In Eq. 1.7.34, the values of a and b must be obtained from the calibration data such that to minimize (in absolute value) the differences between experimental known data fxi g and calculated data based on the signal fa þbyi g. Introducing the residual parameter ri with the expression: ri ¼ xi  ða þ byi Þ

(1.7.35)

the values for a and b are obtained by minimizing the expression: Eða; bÞ ¼

Xn

ðx  a  byi Þ2 ¼ i¼1 i

Xn

r 2 ¼ min i¼2 i

(1.7.36)

The minimum is achieved for E(a,b) when its partial derivatives with respect to a and b are set to zero. This is expressed by the formulas: vEða; bÞ ¼0 va

vEða; bÞ ¼0 vb

(1.7.37)

Expressions 1.7.37 allow the calculation of the values for a and b. In addition, a correlation coefficient R2 (indicated as coefficient of determination) describing the closeness to linearity of the dependence of the two sets fxi g and fa þbyi g can be generated. The correlation coefficient R is known as Pearson correlation coefficient and takes values between 0.0 and 1.0 with 1.0 for perfect correlation. Capability of generating parameters a, b, and R2 is available in Microsoft Excel. (By using the least 0 0 square regression on the inverse function i.e., the equation y ¼ a þ b x, parameters a'

0 0 0 and b' can be obtained and these will generate parameters a ¼ a b and b ¼ 1= b , 2 that fit Eq. 1.7.34, but the value for R for the dependence y ¼ FðxÞ and the dependence x ¼ F1 ðyÞ are not usually identical because the minimized differences as in Eq. 1.7.36 are not the same). The dependence of an analytical signal on the concentration (or of the concentration vs. signal) is not always linear. In some cases, a better fit of the data is obtained using a quadratic dependence of the form: x ¼ a þ by þ cy2

(1.7.38)

The best fit for the data following certain dependencies (linear, polynomial, exponential, logarithmic, power) can also be generated, and such calculations are significantly facilitated by the capabilities offered by the statistical packages and the graphics capabilities such as those from Microsoft Excel. For example, LINEST function in Microsoft Excel allows the calculation of parameters for linear, quadratic or even higher polynomial order dependencies and offer the capability to include weighing the data importance in generating the best fit. Data processing capabilities of some analytical instruments also provide various procedures for linear, nonlinear (and weighed or not), and least square calibrations (e.g., MassHunter package from Agilent).

Introductory concepts related to a chemical analysis

49

Elimination of suspect experimental data In analytical practice it is common that in a set of data some are “outliers”. A criterion for the elimination of such data based on probability theory is “Peirce’s criterion”. In this procedure, the mean m and standard deviations s are first calculated for a set of data fx1 ; x2 ; . xi ; .; xn g. depending on the number n of the data in the set and the number of suspected outliers, a number R is indicated in a “Peirce Criterion Table” that is available in the literature (e.g., Ref. [116]). The value of R is further used to calculate a maximum allowable difference generated by the expression (vertical bars indicate absolute value):  xi  mj

max ¼ R$s

(1.7.39)

For any suspected value xj such that:  xj  mj > R$s

(1.7.40)

the value xj is considered an outlier and should be eliminated. Several other procedures were reported in the literature for the elimination of outliers (e.g., Ref. [117]).

Other mathematical procedures used in data processing In addition to the processing of quantitative data, qualitative data can also be processed for various purposes. Examples of such purposes include evaluation of qualitative and quantitative similarity, and classifications. The evaluation of similarity is necessary in various situations, for example, when multicomponent samples must be overall compared (e.g., Ref. [118]). Such problems occur frequently in practice in evaluating various results, such as from nontargeted analysis. A similarity index comparing two samples X and Y can be obtained, for example, from their chromatograms, each having a number of peaks at retention times tR ðiÞ (in practice within a narrow window around tR ðiÞ), the peaks having areas Ai ðXÞ and Ai ðYÞ. The chemical nature of the peaks may be known, for example, from the MS spectrum and previous identifications, or may remain unknown. When the nature of the peaks cannot be verified, e.g., by their mass spectra, special attention must be given to peak alignment in order to not compare as “ i ” peaks of different compounds. Also, in order to eliminate instrument variability, the areas may be normalized using an internal standard, and average of areas for the same sample analyzed a few times may be used as values for Ai ðXÞ and Ai ðYÞ. A similarity index can be generated by obtaining initially the ratios: Ri ¼

Ai ðXÞ for Ai ðYÞ > Ai ðXÞ Ai ðYÞ i ¼ 1; .::n

or

Ri ¼

Ai ðYÞ for Ai ðXÞ > Ai ðYÞ Ai ðXÞ (1.7.41)

50

Method Development in Analytical HPLC

A resulting similarity index can be obtained by taking the average of Ri values using the expression: n P

SI % ¼

Ri

1

n

$100

(1.7.42)

For all peaks showing equal areas, all Ri values are equal to 1.00, and SI % ¼ 100%. The more differences are between the samples, the lower are the Ri values and SI % < 100%. This index of similarity accounts for both quantitative differences between the analytes (when Ri < 1) as well as qualitative differences since if one peak is absent in one sample and present in the other, Ri ¼ 0. Various other indices for evaluating similarity were described and used for different purposes, such as for the comparison of the UV spectra of peaks [119] process useful in identifying peak purity. A common classification procedure for differentiating samples is based on principal component analysis (PCA). In this procedure, a set of samples can be separated in groups that can be considered similar when for each sample is available a large number of descriptor parameters, for example, related to the sample composition (concentration, peak area, etc.). Assuming for example, that for each sample were measured n analytes, the problem is the classification of the sample based on n descriptors (e.g., n peak areas, or concentrations of n analytes). Such problems are very common, for example in comparing consumer products, types of food, and other multicomponent samples. In PCA (and in the related technique linear discriminant analysis), linear combinations of part of the available descriptors are obtained such that the data are the best differentiated using the first combination of descriptors, followed by the second combination, then by the third one. In this procedure, the n number of descriptors are reduced for sample characterization, for example, to three (orthogonal) variables x, y, z, which are the most relevant for sample differentiation (canonical variables). The components of each variable include only a subset of all descriptors that are detected to indicate differences. Mathematical base for PCA is readily available in the literature (e.g., Refs. [120,121]), and computer packages are available for the processing of the data (e.g., Statistica from StatSoft). As an example of PCA use, a number of tobacco samples were measured 41 analytes including sugars (e.g., glucose, fructose, sucrose, xylose, etc.), organic acids (e.g., malic, citric, lactic, quinic, trihydroxybutanoic, chlorogenic, stearic, etc.), nitrogenous compounds, alcohols, etc. [122]. The target was the differentiation of the tobaccos in groups such as flue-cured, burley, oriental, and reconstituted (manufactured material from tobacco waste). A “training set’ with samples being known to be flue-cured, or burley, or oriental, or reconstituted was selected and analyzed. From all the measured analytes some were selected by the PCA computer package to generated the “canonical variables” x, y, z capable to differentiate the groups. A tridimensional plot showing the separation of the samples in the canonical variable space of each type of tobacco is shown in Fig. 1.7.1. For an unknown sample, based on the same measured analytes, by using the resulting canonical variables the unknown can be placed in the canonical variables space and its

Introductory concepts related to a chemical analysis

51

Figure 1.7.1 Classification of tobacco types based on 41 analytes measured and composed in three canonical variables using principal component analysis.

nature can be established. In addition, a “distance” to the center of each specific class can be calculated for an individual sample (the distance between flue-cured and reconstituted groups in shown in Fig. 1.7.1). A sample containing a mixture of different tobacco types will be placed at intermediary position in the canonical variable space.

Key points • • •

The numerical data related to quantitative measurements are evaluated using statistical techniques. Statistical evaluations are important for the evaluation of precision and accuracy for a set of data, and in general for an analytical method. Statistical evaluations also allow to compare results from different methods and to eliminate potential outliers.

References [1] S.C. Moldoveanu, V. David, Modern Sample Preparation for Chromatography, second ed., Elsevier, Amsterdam, 2019. [2] J.M. Bayona, Sampling theory and methodology, in: J. Pawliszyn (Ed.), Comprehensive Sampling and Sample Preparation, vol. 1, Elsevier, Amsterdam, 2012.

52

Method Development in Analytical HPLC

[3] M. Thomson, M.H. Ramsey, Quality concepts and practices applied to sampling-an exploratory study, Analyst 120 (1995) 261e270. [4] L.H. Keith (Ed.), Principles of Environmental Sampling, ACS Professional Ref. Book, ACS, 1988. [5] G.D. Wight, Fundamentals of Air Sampling, Lewis Pub., Boca Raton, 1994. [6] F. Vincenti, C. Montesano, A. Ciccola, I. Serafini, G. Favero, M. Pallotta, F. Pagano, G. Di Francesco, M. Croce, M.L. Leone, I.M. Muntoni, M. Sergi, Unearthed opium: development of a UHPLC-MS/MS method for the determination of Papaver somniferum alkaloids in Daunian vessels, Front. Chem. 11 (2023) 1238793. [7] S.K. Thompson, Sampling, third ed., Wiley, Hoboken, 2012. [8] C. Kadilar, H. Cingi, Sampling theory. Theoretical approaches, in: J. Pawliszyn (Ed.), Comprehensive Sampling and Sample Preparation, vol. 1, Elsevier, Amsterdam, 2012. [9] P. Gy, A.G. Royle, Sampling for Analytical Purposes, J. Wiley & Sons, Chichester, 1998. [10] C. Zhang, Fundamentals of Environmental Sampling and Analysis, second ed., John Wiley & Sons, Hoboken, 2024. [11] E. Popek, Sampling and analysis of environmental chemical pollutants, in: A Complete Guide, second ed., Elsevier, Amsterdam, 2018. [12] B. Kratochvil, J.K. Taylor, Sampling for chemical analysis, Anal. Chem. 53 (1981) 924Ae938A. [13] C. Ort, M.G. Lawrence, J. Rieckermann, A. Joss, Sampling for pharmaceuticals and personal care products (PPCPs) and illicit drugs in wastewater systems: are your conclusions valid? A critical review, Environ. Sci. Technol. 44 (2010) 6024e6035.14. [14] E.N. Jonsson, J.R. Wade, M.O. Karlsson, Comparison of some practical sampling strategies for population pharmacokinetic studies, J. Pharmacokinet. Biopharm. 24 (1996) 245e263. [15] M.H. Ramsey, A. Argyraki, M. Thompson, Estimation of sampling bias between different sampling protocols on contaminated land, Analyst 120 (1995) 1353e1356. [16] M.H. Ramsey, S. Squire, M.J. Gardner, Synthetic reference sampling target for the estimation of measurement uncertainty, Analyst 124 (1999) 1701e1706. [17] K.H. Esbensen, C. Paoletti, P. Minkkinen, Representative sampling of large kernel lots I. Theory of sampling and variographic analysis, TrAC - Trends Anal. Chem. 32 (2012) 154e164. [18] P. Minkkinen, K.H. Esbensen, C. Paoletti, Representative sampling of large kernel lots II. Application to soybean sampling for GMO control, TrAC - Trends Anal. Chem. 32 (2012) 165e177. [19] K.H. Esbensen, C. Paoletti, P. Minkkinen, Representative sampling of large kernel lots III. General considerations on sampling heterogeneous foods, TrAC - Trends Anal. Chem. 32 (2012) 178e184. [20] S. Huang, M. Fan, N. Wawryk, J. Qiu, X. Yang, F. Zhu, G. Ouyang, X.-F. Li, Recent advances in sampling and sample preparation for effect-directed environmental analysis, TrAC - Trends Anal. Chem. 154 (2022) 116654. [21] Y. Lai, L. Dong, Q. Li, P. Li, J. Liu, Sampling of micro- and nano-plastics in environmental matrixes, TrAC - Trends Anal. Chem. 145 (2021) 116451. [22] https://search.epa.gov/epasearch/?querytext¼SAMPLING&areaname¼&areacontacts¼ &areasearchurl¼&typeofsearch¼epa&result_template¼2col.ftl#/. [23] B. Dellinger, G. Grotecloss, C.R. Fortune, J.L. Cheney, J.B. Homolya, Sulfur dioxide oxidation and plume formation at cement kilns, Environ. Sci. Technol. 14 (1980) 1244e1249.

Introductory concepts related to a chemical analysis

53

[24] J.H. Margeson, J.E. Knoll, M.R. Midgett, G.B. Oldaker, W.E. Reynolds, Determination of sulfur dioxide, nitrogen oxides, and carbon dioxide in emissions from electric utility plants by alkaline permanganate sampling and ion chromatography, Anal. Chem. 57 (8) (1985) 1586e1590. [25] C.H. Risner, High-performance liquid chromatographic determination of major carbonyl compounds from various sources in ambient air, J. Chromatogr. Sci. 33 (1995) 168e176. [26] B. Liu, H.-X. Wang, Determination of atmospheric hydroxyl radical by HPLC coupled with electrochemical detection, J. Environ. Sci. 20 (2008) 28e32. [27] L. Linxiang, Y. Abe, Y. Nagasawa, R. Kudo, N. Usui, K. Imai, T. Mashino, M. Mochizuki, N. Miyata, An HPLC assay of hydroxyl radicals by the hydroxylation reaction of terephthalic acid, Biomed. Chromatogr. 18 (2004) 470e474. [28] F.K. St Charles, S.C. Moldoveanu, Partial vapor pressures, activity coefficients, Henry volatility, and infinite dilution activity coefficients of nicotine from binary mixtures with glycerol and with 1,2-propanediol at 298.15 K, ACS Omega 8 (2023) 7749e7756. [29] P. Forbes, Atmospheric chemistry analysis: a review, Anal. Chem. 92 (2020) 455e472. [30] S.C. Moldoveanu, Analysis of acrylonitrile and alpha-methacrylonitrile in vapor phase of mainstream cigarette smoke using a charcoal trap for collection, Beitr. Tabakforsch. 24 (2010) 145e156. [31] B.C. Turner, D.E. Glotfelty, Field air sampling of pesticide vapors with polyurethane foam, Anal. Chem. 49 (1977) 7e10. [32] R.G. Lewis, A.R. Brown, M.D. Jackson, Evaluation of polyurethane foam for sampling of pesticides, polychlorinated biphenyls and polychlorinated naphthalenes, in ambient air, Anal. Chem. 49 (1977) 1668e1672. [33] J.-Z. Dong, S.C. Moldoveanu, Gas chromatography-mass spectrometry of carbonyl compounds in cigarette mainstream smoke after derivatization with 2,4dinitrophenylhydrazine, J. Chromatogr. A 1027 (2004) 25e35. [34] S.T. Oyama, M. Yamada, T. Sugawara, A. Takagaki, R. Kikuch, Review on mechanisms of gas permeation through inorganic membranes, J. Jap. Petrol. Inst. 54 (2011) 298e309. [35] M. Monteleone, A. Fuoco, E. Esposito, I. Rose, J. Chen, B. Comesa~ na-Gandara, C.G. Bezzu, M. Carta, N.B. McKeown, M.G. Shalygin, V.V. Teplyakov, J.C. Jansen, Advanced methods for analysis of mixed gas diffusion in polymeric membranes, J. Membr. Sci. 648 (2022) 120356. [36] P. Winkler, M. Schulz, W. Dannecker, Problems with the determination of heavy metals in precipitation, Fresenius’ J. Anal. Chem. 340 (1991) 575e579. [37] K.E. Creasy, T.W. Francisco, E.H. Baughman, Sampling and sample preparation, in: R.A. Meyers (Ed.), Process Analysis, in Encyclopedia of Analytical Chemistry: Applications, Theory and Instrumentation, J. Wiley & Sons, New York, 2001. [38] K. Pittertschatscher, N. Inreiter, A. Schatzl, H. Malissa, A multiresidue method with mobile on-site sampling based on SPE for ultratrace analysis of plant protectants in environmental waters, Fresenius’ J. Anal. Chem. 365 (1999) 338e350. [39] W. Mendenhall, Introduction to Probability and Statistics, Duxbury Press, Belmont, 1971. [40] M.R. Carter, E.G. Gregorich, Soil Sampling and Methods of Analysis, CRC Press, Boca Raton, 2008. [41] W. Salomons, W.M. Stigliani, Biogeodynamics of Pollutants in Soils and Sediments, Springer, Berlin, 1995. [42] L. Martín-Pozo, B. de Alarcon-Gomez, R. Rodríguez-G omez, M.T. García-C orcoles, M. C¸ipa, A. Zafra-Gomez, Analytical methods for the determination of emerging contaminants in sewage sludge samples. A review, Talanta 192 (2019) 508e533.

54

Method Development in Analytical HPLC

[43] https://www.gmpsop.com/pharmaceutical-sampling-procedures-for-non-sterile-products/. [44] K.-S. Seo, R. Bajracharya, S.H. Lee, H.-K. Han, Pharmaceutical application of tablet film coating, Pharmaceuticals 12 (9) (2020) 853. [45] P. Abu-Rabie, N. Spooner, Dried matrix spot direct analysis: evaluating the robustness of a direct elution technique for use in quantitative bioanalysis, Bioanalysis 3 (2011) 2769e2781. [46] P. Yin, L. Zou, X. Zhao, G. Xu, Sample collection and preparation of biofluids and extracts for liquid chromatography-mass spectrometry, Methods Mol. Biol. 1277 (2015) 51e59. [47] K. Lew, Blood sample collection and handling, in: J. Pawliszyn (Ed.), Comprehensive Sampling and Sample Preparation, vol. 3, Elsevier, Amsterdam, 2012, pp. 95e121. [48] Q. Zang, C.F. Guigan, X.C. Le, Urine sampling and handling, in: J. Pawliszyn (Ed.), Comprehensive Sampling and Sample Preparation, vol. 3, Elsevier, Amsterdam, 2012, pp. 123e142. [49] I. Markeviciute, W. Frenzel, Evaluation of a flow-through dialysis probe for sampling and sample preparation: hyphenation to ion chromatography, Talanta 186 (2018) 576e583. [50] N. Chen, S. Wu, Y. Zhu, An electrodialytic device for automated inorganic anion preconcentration with determination by ion chromatography-conductivity detection, J. Chromatogr. A 1638 (2021) 461898. [51] Y.-Y. Chao, Y.-T. Wei, C.-T. Lee, H.-S. Kou, Y.-L. Huang, Membrane sampling with microdialysis coupled to HPLC/UV for on-line simultaneous determination of melamine and cyanuric acid in non-dairy coffee creamer, Anal. Sci. 27 (2011) 1025e1030. [52] Y.-L. Wang, J.-F. Yuan, W. Shang, J. Zhang, Z.-Q. Zhang, Dialysis sampling on-line coupled with high-performance liquid chromatography for simultaneous investigation of the interactions between multi-components in herbs and the albumin, Analyst 136 (2011) 823e828. [53] J.B. Vaught, M.K. Henderson, Biological sample collection, processing, storage and information management, IARC Sci. Publ. 163 (2011) 23e42. [54] H.G. Gika, G.A. Theodorakis, I.D. Wilson, Liquid chromatography and ultraperformance liquid chromatography-mass spectrometry fingerprinting of human urine: sample stability under different handling and storage conditions for metabonomics studies, J. Chromatogr. A 1189 (2008) 314e322. [55] http://www.info.dent.nu.ac.th/chemistry/files/chemical%20grade%20diagnosis.pdf. [56] International Standards Organization (ISO), General Requirements for Competence of Reference Material Producers, ISO Guide, 2016, 17034. [57] J. von Liebig, Zur Guano-Analyse, Fresenius’ Z. f€ur Anal. Chem. 1 (1862) 114. [58] G. Kirchhof, R. Bunsen, Die Spectren der Alkalien und alkalischen Erden, Fresenius’ Z. f€ur Anal. Chem. 1 (1862) 1e2. [59] S. Moldoveanu, Aplicat¸iile Teoriei Grupurilor in Chimie, Editura Stiintifica si Enciclopedica, Bucuresti, 1986. [60] K. Yeh, D. Lee, R. Bhargava, Multicolor discrete frequency infrared spectroscopic imaging, Anal. Chem. 91 (2019) 2177e2185. [61] I. Surowiec, J.R. Baena, J. Frank, T. Laurell, J. Nilsson, M. Trojanowicz, B. Lendl, Flowthrough microdispenser for interfacing m-HPLC to Raman and mid-IR spectroscopic detection, J. Chromatogr. A 1080 (2005) 132e139. [62] C.S. Robb, S.E. Geldart, J.A. Seelenbinder, P.R. Brown, Analysis of green tea constituents by HPLC-FTIR, J. Liq. Chromatogr. Relat. Technol. 25 (2002) 787e801.

Introductory concepts related to a chemical analysis

55

[63] J.-S. Blais, A. Huyghues-Despointes, G.M. Momplaisir, W.D. Marshall, Highperformance liquid chromatographyeatomic absorption spectrometry interface for the determination of selenoniocholine and trimethylselenonium cations: application to human urine, J. Anal. At. Spectrom. 6 (1991) 225e232. [64] G.R. Ricci, L.S. Shepard, G. Colovos, N.E. Hester, Ion chromatography with atomic absorption spectrometric detection for determination of organic and inorganic arsenic species, Anal. Chem. 53 (1981) 610e613. [65] T.L. Chester, Dual flame photometric phosphorus-selective detector for high performance liquid chromatography, Anal. Chem. 52 (1980) 1621e1624. [66] J. Proch, P. Niedzielski, Iron species determination by high performance liquid chromatography with plasma based optical emission detectors: HPLCeMIP OES and HPLCeICP OES, Talanta 231 (2021) 122403. [67] Y.-H. Lo, H. Hiramatsu, Online liquid chromatographyeRaman spectroscopy using the vertical flow method, Anal. Chem. 92 (2020) 14601e14607. [68] R.A. Libby, Quantitative thin layer chromatography with X-ray emission spectrometry, Anal. Chem. 40 (1968) 1507e1512. [69] J.-P. Hieta, R. Vesander, M. Sipil€a, N. Sarnela, R. Kostiainen, Soft X-ray atmospheric pressure photoionization in liquid chromatographyemass spectrometry, Anal. Chem. 93 (2021) 9309e9313. [70] S.C. Moldoveanu, Analytical Pyrolysis of Natural Organic Polymers, Elsevier, Amsterdam, 1998. [71] A.S. Dhillon, S. Hagan, O. Rath, W. Kolch, MAP kinase signaling pathways in cancer, Oncogene 26 (2007) 3279e3290. [72] J. Hanrieder, A. Zuberovic, J. Bergquist, Surface modified capillary electrophoresis combined with in solution isoelectric focusing and MALDI-TOF/TOF MS: a gel-free multidimensional electrophoresis approach for proteomic profiling - exemplified on human follicular fluid, J. Chromatogr. A 1216 (2009) 3621e3628. [73] Y.-J. Hsieh, S.-J. Jiang, Application of HPLC-ICP-MS and HPLC-ESI-MS procedures for arsenic speciation in seaweeds, J. Agric. Food Chem. 60 (2012) 2083e2089. [74] M. Montes-Bayon, K. DeNicola, J.A. Caruso, Liquid chromatographyeinductively coupled plasma mass spectrometry, J. Chromatogr. A 1000 (2003) 457e476. [75] C. Smith, B.P. Jensen, I.D. Wilson, F. Abou-Shakra, D. Crowther, High-performance liquid chromatography/inductively coupled plasma mass spectrometry and tandem mass spectrometry for the detection of carbon-containing compounds, Rapid Commun. Mass Spectrom. 18 (2004) 1487e1492. [76] V.V. Cosofret, R.P. Buck, Pharmaceutical Applications of Membrane Sensors, Taylor & Francis, Abingdon, 2017. [77] M. Sandvoss, L.H. Pham, K. Levsen, A. Preiss, C. M€ ugge, G. W€ unsch, Isolation and structural elucidation of steroid oligoglycosides from the starfish Asterias rubens by means of direct online LC-NMR-MS hyphenation and one- and two-dimensional NMR investigations, Eur. J. Org. Chem. 2000 (2000) 1253e1262. [78] U.G. Sidelmann, I. Bjørnsdottir, J.P. Shockcor, S.H. Hansen, J.C. Lindon, J.K. Nicholson, Directly coupled HPLC-NMR and HPLC-MS approaches for the rapid characterization of drug metabolites in urine: application to the human metabolism of naproxen, J. Pharm. Biomed. Anal. 24 (2001) 569e579. [79] R. Brkljaca, S. Urban, Recent advancements in HPLC-NMR and applications for natural product profiling and identification, J. Liq. Chromatogr. Relat. Technol. 34 (2011) 1063e1076.

56

Method Development in Analytical HPLC

[80] M. Wang, H. Wang, H. Zheng, D. Uhrin, R.J. Dewhurst, R. Roehe, Comparison of HPLC and NMR for quantification of the main volatile fatty acids in rumen digesta, Sci. Rep. 11 (2021) 24337. [81] H. Fatahi, J. Claverie, S. Poncet, Thermal characterization of phase change materials by differential scanning calorimetry: a review, Appl. Sci. 12 (2022) 12019. [82] S.-H. Son, J. Jegal, Synthesis and characterization of the chiral stationary phase based on chitosan, Appl. Polym. 106 (2007) 2989e2996. [83] J. Flieger, H. Trebacz, M. Pizon, A. Plazinska, W. Plazi nski, A. Kowalska, A. Szczesna, T. Plech, Thermodynamic study of new antiepileptic compounds by combining chromatography on the phosphatidylcholine biomimetic stationary phase and differential scanning calorimetry, J. Separ. Sci. 42 (2019) 2628e2639. [84] S.C. Moldoveanu, Analytical Pyrolysis of Synthetic Organic Polymers, Elsevier, Amsterdam, 2005. [85] S.C. Moldoveanu, Pyrolysis of organic molecules, in: Applications to Health and Environmental Issues, second ed., Elsevier, Amsterdam, 2019. [86] https://chemdata.nist.gov/. [87] P. Raveendran, Y. Ikushima, S.L. Wallen, Polar attributes of supercritical carbon dioxide, Acc. Chem. Res. 38 (2005) 478e485. [88] M.S. Tswett, On a new category of adsorption phenomena and on its application to biochemical analysis, Proc. Warsaw Soc. Nat. 14 (1905) 20e39. [89] J. Farradane, History of chromatography, Nature 167 (1951) 120. [90] A.J.P. Martin, R.L.M. Synge, A new form of chromatogram employing two liquid phases, Biochem. J. 35 (1941) 1358e1368. [91] G.A. Howard, A.J.P. Martin, The separation of the C12-C18 fatty acids by reversed-phase partition chromatography, Biochem. J. 46 (1950) 532e538. [92] C. Horvath, S.R. Lipsky, Use of liquid ion exchange chromatography for the separation of organic compounds, Nature 211 (1966) 748e749. [93] C. Horvath, B.A. Preiss, S.R. Lipsky, Fast liquid chromatography. Investigation of operating parameters and the separation of nucleotides on pellicular ion exchangers, Anal. Chem. 39 (1967) 1422e1428. [94] J.J. Kirkland, Superficially porous supports for chromatography, 1970. US Patent, 3,505,785. [95] J.J. Kirkland, J.J. DeStefano, Controlled surface porosity supports with chemicallybonded organic stationary phases for gas and liquid chromatography, J. Chromatogr. Sci. 8 (1970) 309e314. [96] S. Moldoveanu, V. David, Essentials in Modern HPLC Separations, second ed., Elsevier, Amsterdam, 2022. [97] M. Trojanowicz, K. Kołacinska, Recent advances in flow injection analysis, Analyst 141 (2016) 2085e2139. [98] J.S. Fritz, D.T. Gjerde, Ion Chromatography, Wiley-VCH, Weinheim, 2009. [99] G. Xu, S. Dong, Electrochemiluminescent detection of chlorpromazine by selective preconcentration at a lauric acid-modified carbon paste electrode using tris(2,2‘-bipyridine) ruthenium(II), Anal. Chem. 72 (2000) 5308e5312. [100] A. Medvedovici, A. Farca, V. David, Derivatization reactions in liquid chromatography for drug assaying in biological fluids, Adv. Chromatogr. 47 (2009) 283e322. [101] Y. Zuo, H. Chen, Simultaneous determination of sulfite, sulfate, and hydroxymethanesulfonate in atmospheric waters by ion-pair HPLC technique, Talanta 59 (2003) 875e881.

Introductory concepts related to a chemical analysis

57

[102] B. Zhu, Z. Zhong, J. Yao, Ion chromatographic determination of trace iodate, chlorite, chlorate, bromide, bromate and nitrite in drinking water using suppressed conductivity detection and visible detection, J. Chromatogr. A 1118 (2006) 106e110. [103] J.D. Pinkston, Advantages and drawbacks of popular supercritical fluid chromatography/ mass spectrometry interfacing approaches - a user’s perspective, Eur. J. Mass Spectrom. 11 (2005) 189e197. [104] H.Y. Ta, F. Collin, L. Perquis, V. Poinsot, V. Ong-Meang, F. Couderc, Twenty years of amino acid determination using capillary electrophoresis: a review, Anal. Chim. Acta 1174 (2021) 338233. [105] M. Ranasinghe, J.P. Quirino, Can we replace liquid chromatography with the greener capillary electrophoresis? Curr. Opin. Green Sustainable Chem. 31 (2021) 100515. [106] N.M. Porpiglia, S.A. Savchuk, S.A. Appolonova, F. Bortolotti, F. Tagliaro, Capillary Electrophoresis (CE) vs. HPLC in the determination of asialo-Tf, a crucial marker for the reliable interpretation of questioned CDT increases, Clin. Chim. Acta 486 (2018) 49e53. [107] D.L. Massart, A. Dijkstra, L. Kaufman, Evaluation and Optimization of Laboratory Methods and Analytical Procedures, Elsevier, Amsterdam, 1978. [108] C. Liteanu, I. Rica, Statistical Theory and Methodology of Trace Analysis, E. Horwood, Chichester, 1980. [109] J.C. Miller, J.N. Miller, Statistics for Analytical Chemistry, third ed., Ellis Horwood, PTR Prentice Hall, New York, 1993. [110] P.E. Green, Analyzing Multivariate Data, The Dryden Press, New York, 1978. [111] R. Larson, E. Farber, Elementary Statistics: Picturing the World, sixth ed., Pearson Education Inc., Boston, 2015. [112] https://en.wikipedia.org/wiki/List_of_statistical_software. [113] A.C. Olivieri, Practical guidelines for reporting results in single- and multi-component analytical calibrations: a tutorial, Anal. Chim. Acta 868 (2015) 10e22. [114] F. Grémy, D. Salmon, Bases Statistiques, Dunod, Paris, 1969. [115] E.L. Crow, F.A. Davis, M.W. Maxfileld, Statistics Manual, Dover Pub. Inc., New York, 1960. [116] W. Chauvenet, B. Pierce, A Treatise on the Method of Least Squares: Or, the Application of the Theory of Probabilities in the Combination of Observations, Franklin Classics, 2018. [117] J.R. Taylor, An Introduction to Error Analysis, second ed., Univ. Sci. Books, Sausalito, 1997. [118] F. Ortega-Gavilan, A.M. Jiménez-Carvelo, L. Cuadros-Rodríguez, M.G. BagurGonzalez, The chromatographic similarity profile e an innovative methodology to detect fraudulent blends of virgin olive oils, J. Chromatogr. A 1679 (2022) 463378. [119] J. Li, D.B. Hibbert, S. Fuller, G. Vaughn, A comparative study of point-to-point algorithms for matching spectra, Chemometr. Intell. Lab. Syst. 82 (2006) 50e58. [120] P.E. Green, Analyzing Multivariate Data, The Dryden Press, Hinsdale, 1978. [121] G.H. Dunteman, Principal Components Analysis, SAGE Pub, New York, 1989. [122] S. C. Moldoveanu, (unpublished data).

Retention and elution processes in high-performance liquid chromatography 2.1

2

Equilibrium types in HPLC

The chromatographic process takes place as an equilibrium of molecules X between the stationary phase (st) where X is retained, and the mobile phase (mo) where X is eluted. The molecule X can be the analyte or another sample component. A number of model equilibria can be considered for the explanation of this process. The model equilibria are determined by the nature of the components involved in the process: stationary phase, compound X, and mobile phase. Each type of equilibrium depends on the molecular interactions between the participating components. Because the molecular interactions in the chromatographic process are complex, more than one type of equilibrium can take place at the same time, and in some cases, it is not clear what type of equilibrium can be considered dominant. The model equilibria for the chromatographic process are the same as the known interphasic types of equilibria such as partition (similar to liquid-liquid partition), adsorption (similar to adsorption of a gas on a solid), ion exchange, size exclusion.

Partition equilibrium model In partition type of equilibrium, both stationary and mobile phases are assumed to be liquid, but one is assumed to be immobilized on a solid support. This assumption allows to extend the theory of liquid-liquid extraction to the chromatographic process and to consider that the molecule X is subject to a simple equilibrium shown in Eq. 2.1.1. Xmo $ Xst

(2.1.1)

Indicating with ½Xst the molar concentration of X in stationary phase, and with ½Xmo the molar concentration of X in mobile phase, the equilibrium is controlled by the distribution constant KðXÞ with the formula: KðXÞ ¼

½Xst ½Xmo

(2.1.2)

At equilibrium the chemical potentials mst ðXÞ and mmo ðXÞ of the component X in each of the two phases st and mo must be equal, such that the following expression characterizing the partition process 2.1.1 can be written (notation “ln” is used for natural logarithm and “log” is used for logarithm in base 10) Method Development in Analytical HPLC. https://doi.org/10.1016/B978-0-443-29849-3.00004-X Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

60

Method Development in Analytical HPLC

m0mo ðXÞ þ RT ln amo ðXÞ ¼ m0st ðXÞ þ RT ln ast ðXÞ

(2.1.3)

In Eq. 2.1.3 m0st ðXÞ and m0mo ðXÞ are the standard chemical potentials of compound X in the two phases, ast ðXÞ and amo ðXÞ are the activities, T is the absolute temperature, and R is the gas constant (R ¼ 8.31451 J deg1 mol1 ¼ 1.987 cal deg1 mol1). From Eq. 2.1.3, the ratio ast ðXÞ=amo ðXÞ representing the thermodynamic distribution constant (or thermodynamic partition constant) K therm for the partition process can be written in the form: K therm ðXÞ ¼

 0  ast ðXÞ m ðXÞ  m0mo ðXÞ ¼ exp  st amo ðXÞ RT

(2.1.4)

The dependence of activities aðXÞ on molar concentration ½X and on activity coefficients gðXÞ is given by the formula: aðXÞ ¼ gðXÞ$ ½X

(2.1.5)

Using formula 2.1.5 and the notation m0st ðXÞ  m0mo ðXÞ ¼ Dm0 ðXÞ, Eq. 2.1.4 can be written as follows: K therm ðXÞ ¼

  gst ðXÞ½Xst g ðXÞ Dm0 ðXÞ KðXÞ ¼ exp  ¼ st RT gmo ðXÞ½Xmo gmo ðXÞ

(2.1.6)

For a constant pressure and temperature Dm0 ðXÞ ¼ DG0 ðXÞ, where DG0 ðXÞ is the variation in the standard free enthalpy (Gibbs free energy) during the transfer of the analyte X from the mobile to the stationary phase. With this substitution, the dependence of KðXÞ on the variation of free enthalpy can be written in the form: KðXÞ ¼

  gmo ðXÞ DG0 ðXÞ exp  gst ðXÞ RT

(2.1.7)

In diluted systems the activity coefficients g are very close to one and as a result,  gmo ðXÞ gsy ðXÞ z 1 and using the expression DG0 ðXÞ ¼ DH 0 ðXÞ  T DS0 ðXÞ where DH 0 ðXÞ is the standard enthalpy and DS0 ðXÞ the standard entropy change for the transfer of the analyte mo/ st, the formula for the distribution constant can be written in the form (known as van’t Hoff equation):     DG0 ðXÞ DH 0 ðXÞ þ TDS0 ðXÞ KðXÞ ¼ exp  ¼ exp RT RT

(2.1.8)

This enthalpy changes during the distribution (partition) process appears as a result of molecular interactions taking place between the analyte X, stationary phase, and mobile phase molecules. The detailed energetics of this process is not taken into

Retention and elution processes in high-performance liquid chromatography

61

consideration in the present treatment, and Eq. 2.1.8 is used just as a general expression for the description of the separation in HPLC based on partition. The formula indicates that larger DH 0 in absolute value (since DH 0 is negative) and larger DS0 for the transfer to the stationary phase of compound X generate a larger KðXÞ and therefore a stronger retention of the analyte on the stationary phase. The enthalpy change DH 0 ¼ H final  H initial is negative when the final enthalpy of the system is lower than the initial one. Therefore, for the displacement of the equilibrium 2.1.1 toward the right, DH 0 must be negative. Values for DH 0 in chromatographic separation vary significantly depending on the separation system (analyte, stationary phase, mobile phase) and common valued are between 1.0 kcal mol1 and 30 kcal mol1 or even larger (in absolute value). On the other hand, although positive values for DS0 would contribute to the increase of KðXÞ, the retention of a molecule on the stationary phase is associated with a decrease in entropy and DS0 becomes negative. Common values for DS0 are between 1.0 and 7.0 cal mol1 deg1. The contribution to the value of KðXÞ of enthalpic component is about 10 time higher than the one of entropic component, in common partition equilibria encountered in HPLC. The contribution to the free enthalpy changes for the transfer of X from the phase mo into phase st comes from the changes suffered by the entire system, and not only from the interactions generated by the analyte X with the mobile or stationary phase. These changes must also include any energetic modification in the mobile phase upon the transfer of X from the mobile phase into the stationary phase. As a result, the following formula can be written: DG0 ¼ DG0analyte þ DG0eluent

(2.1.9)

With the help of Eq. 2.1.9, the partition model of the chromatographic equilibrium provides a means to estimate the dependence of its equilibrium constant on the organic content in the mobile phase when this phase is made, for example, from a mixture of water and an organic solvent. For this purpose, it should be assumed that the free energy DG0eluent can be separated into the contributions of the two solvents DG0H2 O and DG0org as follows: DG0eluent ¼ ð1  xÞDG0H2 O þ xDG0org

(2.1.10)

In Eq. 2.1.10, x is the molar fraction of the organic component in the mobile phase. (Molar fraction for a component Y in a solution is the number of moles nY of Y divided by the sum of the total.number of moles for the solution components, and it is given by Ptotal ni ). Including Eqs. 2.1.9 and 2.1.10 in Eq. 2.1.8, the the formula xY ¼ nY i following expression can be written: ln KðXÞ ¼ 

DG0analyte þ ð1  xÞDG0H2 O þ xDG0org RT

(2.1.11)

62

Method Development in Analytical HPLC

  The terms in Eq. 2.1.11 can be grouped as the free term  DG0analyte þDG0H2 O = RT  . and the coefficient for x giving the term DG0H2 O  DG0org RT, both terms being constant at a constant temperature. As a result, the dependence of K on the content of organic solvent can be written for partition distribution in the form: ln KðXÞ ¼ a  bx

(2.1.12)

The previous discussion regarding distribution of component X between the two nonmiscible liquid phases st and mo is based on the assumption that the analyte X participates in the partition as a sole species. However, it is not uncommon that one analyte X is present in a solution in more than one form, due to tautomerism, dimerization, ionization (electrolytic dissociation), ion-pairing, or complexation. For a compound X present in solution in a number of forms, X1 ; X2 ; .; Xn , all being identified as compound X, the equilibrium between phase st and mo for X is described by the parameter DðXÞ, indicated as distribution coefficient (ratio): DðXÞ ¼

½X1 st þ ½X2 st þ .::½Xn st ½X1 mo þ ½X2 mo þ .::½Xn mo

(2.1.13)

An expression similar to Eq. 2.1.8 is valid for DðXÞ, considering all the corresponding values for the enthalpies and entropies changes of each component, and DðXÞ ¼ KðXÞ for n ¼ 1.

Adsorption equilibrium model Adsorption and desorption process on the solid surface of stationary phase is another possible type of equilibrium describing the separation process in HPLC. In this case, the model assumes the existence of a number of adsorption sites on the surface of stationary phase, where the solute molecules X and also the molecules of the organic component of the mobile phase S can be reversibly bound during the process. Overall, the molecule X is considered involved in a displacement equilibrium of the form: Xmo þ nSst % Xst þ nSmo

(2.1.14)

In this equilibrium, both X and S can be either adsorbed on the stationary phase ðstÞ or can be present in the mobile phase ðmoÞ. The equilibrium described by Eq. 2.1.14 therm given by the ratio: can be characterized by a thermodynamic constant Kad therm ðXÞ ¼ Kad

  ast ðXÞ amo ðSÞ n amo ðXÞ ast ðSÞ

(2.1.15)

Retention and elution processes in high-performance liquid chromatography

63

Replacing the activities in Eq. 2.1.15 with concentrations, equilibrium 2.1.14 is described by the constant Kad ðXÞ with the formula:   ½Xst ½Smo n Kad ðXÞ ¼ ½Xmo ½Sst

(2.1.16)

Imposing some requirements for the adsorption process, from the equality of chemical potentials m for the components of the two sides of the equilibrium 2.1.14, the thermodynamic constant can be written as follows: therm Kad ðXÞ ¼ exp

 0  mmo ðXÞ þ nm0st ðSÞ  m0st ðXÞ  nm0mo ðSÞ  RT

(2.1.17)

For all activity coefficients g ¼ 1, and assuming that standard chemical potentials in the mobile phase m0mo ðXÞ and n$m0mo ðSÞ are equal and cancel each other, the expression for Kad ðXÞ can be simplified as follows:   nm0st ðSÞ  m0st ðXÞ Kad ðXÞ ¼ exp  RT

(2.1.18)

In the adsorption process of analyte X can be also independently considered in its own equilibrium Xmo % Xst , which can be described by its equilibrium constant KðXÞ ¼ ½Xst =½Xmo . The relation between the equilibrium constant KðXÞ and the adsorption equilibrium constant Kad ðXÞ can be obtained using Eq. 2.1.16. As a result, the following expression is valid:  KðXÞ ¼ Kad ðXÞ

½Sst ½Smo

n (2.1.19)

The equilibrium constant KðXÞ for adsorption is therefore given by the expression: 

½Sst KðXÞ ¼ ½Smo

n

  nm0st ðSÞ  m0st ðXÞ exp  RT

(2.1.20)

Similar to the case of partition, for a constant pressure and temperature, Dm0 ðXÞ ¼ DG0 ðXÞ. Eq. 2.1.20 indicates that in the case of adsorption, the process of analyte retention depends on the difference between the molecules of analyte and the molecules of mobile phase in terms of the free energy of interaction DG0 with the stationary phase, but instead of depending only on the free energy, Eq. 2.1.20 indicates that the concentration in the stationary phase of the molecules S of the mobile phase, also plays a role in this equilibrium.

64

Method Development in Analytical HPLC

Eq. 2.1.20 also allows to evaluate the variation with mobile phase composition. In the adsorption process, not all molecules of the mobile phase are adsorbed on the stationary phase. Assuming that x is the molar fraction of the component in the mobile phase the concentration ½Smo of adsorbing molecules is proportional to x, and assuming all the other parameters in Eq. 2.1.20 as constants, the logarithm of Eq. 2.1.20 leads to the formula: 0

0

ln KðXÞ ¼ a  b ln x

(2.1.21)

Eq. 2.1.21 shows a different dependence between equilibrium constant and mobile phase composition in adsorption mechanism, which is In KðXÞ versus ln x, unlike the dependence ln KðXÞ versus x in case of retention based on pure partition mechanism. This difference can be used in identification of the type of retention mechanism that takes place during an HPLC separation. Deviation from both dependences can be interpreted as a mixed mechanism, or the presence of other processes interfering in the retention process

Equilibrium model involving ionic molecules Ionic equilibrium is another type possible in HPLC separations, involving analytes that are ionic species. In this case, the equilibrium is not partition or adsorption being similar to the one taking place during the ion exchange process. The exchange mechanism is based on the substitution of an ion from the stationary phase with one from mobile phase. The equilibrium process is governed by the affinity of ionic species involved in this substitution. The substitution process is influenced by many factors depending on the nature of exchanged ions, the stationary phase (the nature of functionality exchanging ions) and the mobile phase (pH, ionic strength, temperature, and others). However, some formal descriptions of the previous mechanisms can still be applied to ionic exchange in order to characterize the equilibrium. The ion exchange process is subdivided into cation exchange and anion exchange depending if the exchanged ions have a positive charge (cations) or a negative one (anions). In cation exchange, the cations M zþ in solution are exchanged with the cations C þ (e.g., Hþ or Naþ) that were initially retained by the stationary phase with counterionic groups of the type resX  . In anion exchange the anionic species B in solution are exchanged with the anions A (e.g., OHe, Cle) that were initially retained by the stationary counterionic groups of the type resY þ . In cation exchange chromatography, for example, for a given ion M zþ in solution and a resin in Cþ form, the exchange equilibrium is the following: z resX  C þ þ M zþ % ðresX  Þz M zþ þ z Cþ

(2.1.22)

Retention and elution processes in high-performance liquid chromatography

65

The constant describing this equilibrium can be written as follows:  þ z ½M zþ res ½C res KðMCÞ ¼ zþ = ½M mo ½C þ mo

(2.1.23)

where the index res indicates the resin stationary phase and mo indicates mobile phase. The equilibrium described by Eq. 2.1.22 can be viewed as equivalent with two independent equilibria: þ Cmo % resX  C þ

(2.1.24)

zþ % ðresX  Þz M zþ Mmo

(2.1.25)

Each of these equilibria are governed by the corresponding constants Kres ðCÞ and Kres ðMÞ described by the formulas:



Kres ðCÞ ¼ Cþ res = Cþ mo

(2.1.26)



Kres ðMÞ ¼ M zþ res = M zþ mo

(2.1.27)

These two constants lead to the following expression for the total equilibrium constant KðMCÞ: KðMCÞ ¼

Kres ðMÞ ½Kres ðCÞz

(2.1.28)

The exchange constant KðMCÞ indicates the degree to which an ion M zþ is preferred in the exchange process, compared with the ion Cþ . Larger constants for KðMCÞ indicate higher affinity for the resin of species M zþ . The difference in the values of exchange constant KðMCÞ for different ions is usually explained by the Gibbs-Donnan effect [1]. Eq. 2.1.23 shows that the equilibrium in the retention process of ions depends on both the equilibrium constant and the concentration ½C þ mo of the replaced ions in solution.

Equilibrium model in size exclusion processes Size exclusion (SEC) is a process in which the separation takes place based on the differences of the molecular size of sample components, or more correctly based on their hydrodynamic volume. In solutions, macromolecules are usually associated with solvent molecules and the resulting overall volume is indicated as hydrodynamic volume. In this separation the stationary phase consists of a porous structure in which small

66

Method Development in Analytical HPLC

molecules can penetrate and spend time passing through the long channels of the solid material, while large molecules cannot penetrate the pore system of the stationary phase and remain in the mobile phase being eluted earlier. The equilibrium for a molecule retained in the stationary phase and the one present in the mobile phase can be described by the equilibrium constant KSEC , which represents the ratio of the concentrations of the sample component in the pore (½Xpore ) and the concentration in the solution from the interstitial volume of the stationary phase (½Xinter ): KSEC ¼

½Xpore ½Xinter

(2.1.29)

The evaluation of KSEC can be obtained similarly with the previously described cases of equilibriums and is based on the evaluation of free enthalpy during the transfer of the molecules X from the interstitial space of the column into the pore. Assuming for the sample components (e.g., analyte X) that no attractive or repulsive interaction exists with the stationary phase except for the effects caused by the imperviousness of the pore walls, the free energy for the process is generated only from the difference in the entropy DS0 of molecules X outside the pore and inside the pore and DH 0 ¼ 0. Based on Eq. 2.1.8, this generates the expression: KSEC ¼ exp DS0 = R

(2.1.30)

As indicated in Eq. 2.1.30, only the entropic component of the interaction is involved in the value of equilibrium constant. Outside the pore, macromolecules are assumed expanded and having a higher entropy level than in the pore of stationary phase. Inside the pore, the macromolecules are contracted and lose part of their conformational entropy and for the process DS0 ¼ S0pore  S0inter < 0 where S0pore is the entropy of the molecule inside the stationary phase pore, and S0inter the entropy outside the pore with DS0 values between - 4.0 and 10.0 cal mol1 deg1. As a result, for the macromolecules KSEC  1 and they are very little retained or not retain at all by the stationary phase. The value of DS0 is in fact determined by the hydrodynamic volume of the macromolecules since in a solution, associations with the solvent molecules take place and the larger is the hydrodynamic volume of the macromolecule the larger is the negative value of DS0 . For small molecules, no difference in the entropy exists between the molecules inside the pores or in the interstitial volume and as a result DS0 ¼ 0 and KSEC ¼ 1. Therefore, the nonpolymeric molecules are retained longer in the separation process. In practice, the size exclusion mechanism previously described is frequently associated with a mechanism in which energetic interactions also take place (DH 0 s 0). As a result, the separation in SEC can be affected not only by the hydrodynamic volume of the separated molecules, but also by some molecular interactions between the analyte and the stationary phase. Experimentally, the distribution constant KSEC for macromolecules depends on their shape. It has been shown that for linear chains the values of KSEC are better correlated with the molecular radius of gyration, while for branched chains the correlations are better with the hydrodynamic volume [2].

Retention and elution processes in high-performance liquid chromatography

67

Other types of equilibrium models In addition to the previously described main equilibria, some processes in HPLC may be considered as based on a different type of equilibrium. One example is the equilibrium based on affinity interactions that are typical for protein binding. This type of equilibria allows very specific separations. Examples of protein binding interactions are protein-antibody, avidin-biotin, etc. Affinity chromatography is more frequently used at low pressure for protein purification, but is also applied in HPLC.

Key points • •

The model equilibria for the chromatographic process are adopted from the known interphasic types of equilibria such as partition (similar to liquid-liquid partition), adsorption (similar to adsorption of a gas on a solid), ion exchange, size exclusion. Equilibrium constants are determined by the free enthalpy (Gibbs free energy) DG during the transfer of the analyte X from the mobile to the stationary phase

2.2

Intermolecular interactions involved in HPLC separations

The interactions taking place during an equilibrium can be evaluated at both molar level (gram quantities), as well as at the molecular level. For example, the thermodynamic functions such as energy E, enthalpy H, free energy A, free enthalpy G, chemical potential m, etc. are indicated for a mole of compound, while interactions such as electric charge to charge, or charge to dipole moment are evaluated at molecular level. The change from a property at molecular level to one at molar level, is obtained by multiplication with Avogadro number N ¼ 6.02214179  1023 mol1. Also, for interactions at molecular level, instead of a molar constant such as the gas constant R, the Boltzmann constant kB ¼ R=N ¼ 1.3806504  1023 J K1 should be used. Many properties of molecules are referred to both molecular level and molar (bulk) level and for simplicity, notation such as E; H; A; G; m; etc. will be used for quantities at both molar and molecular level. Also, identical dependencies between thermodynamic functions at the molar level are found at the molecular level. For example, the relation between the energy E and free energy A is given by E ¼ A  TðvA=vTÞV . Assuming A not temperature dependent, the result is E ¼ A, and therefore the two thermodynamic functions can be used interchangeably. Also, expression DG ¼ DA þ pDV indicates DG ¼ DA at constant volume, and DG ¼ DE at constant temperature and volume. Molecular interactions will be considered initially only for two molecules in a dielectric medium with the dielectric constant ε (relative to vacuum with ε ¼ 1 for vacuum). However, the interactions in a bulk suffer the influence of all the surrounding molecules, and the extension to the bulk of the findings for a molecular pair must be further corrected for the effects of the environment.

68

Method Development in Analytical HPLC

A common property indicated for both molar (bulk) and molecular level is polarity. At molecular level, the asymmetrical charge distribution in the molecule is referred as polarity. While in molecules containing ionic bonds the charges in the molecule are localized on the atoms that attract each other, this is not the case for molecules containing covalent or coordinative bonds. In these molecules the bonds are formed when valence electrons are shared between two or more atoms and the electrons forming the bond. The shared electrons can be in the “bond center” or can be displaced from the bond center, when the charge distribution in the molecule is not uniform creating polarity. A compound should be indicated as polar when opposite partial charges are known to be present in their molecule, but this information is not always available and the compounds are considered polar when specific functional groups known to be “polar” such aseOH, eCOOH, oreNH2 are present in the molecule. Also, specific properties of the compound are an indication of polarity. One bulk property related to polarity is the solubility in water or in solvents miscible with water (such as methanol). Based on this property, polar compounds are indicated as hydrophilic while nonpolar compounds are indicated as hydrophobic. The hydrophobic character is sometimes indicated as lipophilic character. However, the two terms should not be used interchangeably, hydrophobicity being the physical property of a compound to be repelled from water, while lipophilicity referring to the ability of a chemical compound to dissolve in fats, oils, lipids, and nonpolar solvents. For example, silicones and fluorocarbons have hydrophobic character but they are not lipophilic. An empirical bulk parameter commonly used for evaluating the hydrophobic/hydrophilic character (and polarity) of a compound is the octanol/water partition constant (also indicated as partition coefficient). This constant indicated as Kow ðXÞ or PðXÞ is given by the formula: Kow ðXÞ ¼

½Xoctanol ½Xwater

(2.2.1)

As shown by Eq. 2.2.1, octanol/water partition constant is the ratio of the concentrations of a compound in octanol versus that in water, when the octanol and the water are in contact to each other and the octanol is saturated with water and the water with octanol. Experimental values for Kow are known for many compounds and these are given in extensive tables [3]. Also, several procedures were developed for log K ow estimation including some using computer programs [4e6]. Positive values for log K ow indicate a hydrophobic character for the compound X, larger values indicating higher hydrophobicity. Compounds with low or negative values for log K ow are the polar. In the separation process taking place in HPLC, depending on the structure of participating molecules different types of interactions may be involved. The separation process is based on these interactions and a short survey on types of interactions between molecules and ions is given in this section.

Retention and elution processes in high-performance liquid chromatography

69

Ionic interactions or charge to charge interactions In molecules containing ionic bonds, the bonds involve the electrostatic attraction between ions having opposite charges þq and eq. The charge value is given by the formula: q ¼ ze

(2.2.2)

where z is the numerical charge and e ¼ 1.602 1019 C (Coulomb). Ionic bonds are formed when valence electrons are transferred from one atom (or atom group) to another and the two generated ions are attracted one to the other. The free energy A of the interaction between two charges q1 and q2 separated by the distance r is described by the Coulomb law: A¼

1 q1 q2 z1 z2 e2 ¼ ke 4pε0 εr εr

(2.2.3)

In Eq. 2.2.3, ε0 is the vacuum permittivity (ε0 ¼ 8.854 1012 C V1 m1 where V indicates volt, and C coulomb), q1 and q2 are the charges of particles given in terms of elementary charge, r is the distance in m, and ε is the dielectric constant. The constant ke is Coulomb constant (ke ¼ 8.98755 10þ9 N m2 C2 where N are Newtons). Since all the interactions are assumed to take place at constant volume (in condensed phase), the free energy of interaction can be assumed to be equal with the free enthalpy of interaction A ¼ G for the specified system. Also, since no temperature T appears in Eq. 2.2.3, it can be concluded that A ¼ E, the total energy of the system. The range of energies involved in ionic interactions is indicated in Table 2.2.1.

Table 2.2.1 Typical values for different interaction energiesa.

a

Interaction type

Energy (kJ molL1)

Energy (kcal molL1)

Dispersion Dipole e induced dipole Dipole e dipole Hydrogen bonding Hydrogen bonding for HF-:FHalogen bonding Ion-dipole Ionic Covalent

8e30 4e8 4e13 10e40 160 10e20 60e130 200e800 200e400

2e7 1e2 1e3 2.5e10 38 2.5e5 15e30 50e200 50e100

The energies involved in the formation of the bonds must be considered with negative values for their formation and with positive value for breaking them.

70

Method Development in Analytical HPLC

Ion to dipole interactions At molecular level, the polarity of a molecule can be characterized by its dipole moment ! m defined by the formula: ! ! m ¼qd

(2.2.4)

! where d is a vector with the length equal to the distance (in nm) between the two separated point charges þq and eq directed from the negative charge to the positive charge. The vector character of the dipole moment (with components mx ; my ; and mz ) is frequently neglected (! m is usually indicated as m). The unit of dipole moment is the debye (D), and 1 D ¼ 3.336 1030 C m. The concept of polarity is not applied only to the molecules of an analyte but can be extended to the molecules of a mobile phase and stationary phase. The interaction between an ion and a dipole can be reduced to the interaction between a charge ze and the two partial charges þq and eq of the dipole. The Coulomb energy of the ion-dipole interaction can be calculated by taking d the length of the dipole and r the distance between the charge ze and the middle of the dipole as shown schematically in Fig. 2.2.1. This free energy of interaction is obtained by considering the Coulomb interactions between the charges ze and þq, and ze and eq. Following all the calculations, the value for the free energy A is given by the expression: A¼

  ze$ q d cos q 4pε0 ε r 2  0:25 d2 cos2 q

(2.2.5)

Since r[d, Eq. 2.2.5 can be reduced to the following expression: A¼

ze$m cos q 4pε0 ε r 2

(2.2.6)

Eq. 2.2.6 allows the calculation of the energy of interaction between a dipole and an ion. This energy is attractive if the negative partial charge of the dipole points toward a positive ion (0 < q < 90 ) showing a maximum for q ¼ 0 , and is repulsive if the positive charge points toward the positive partial charge of the dipole (90 < q < 180 ), showing a minimum for q ¼ 180 . The calculations leading to Eq. 2.2.6 makes however the assumption that the dipole-ion interaction is present in

ze

d

r

+q

-q Figure 2.2.1 Schematic drawing of the interaction between a charge ze and a dipole generated by the partial charges q.

Retention and elution processes in high-performance liquid chromatography

71

a fixed position, but this is a strong simplification of the reality. The correct values for A should be obtained using a Boltzmann angle-averaged expression (e.g., Ref. [7]). In this case the free energy of interaction between an ion and a dipole has the expression: Az 

ðzeÞ2 $m2 6kb T ð4pε0 εÞ2 r 4

(2.2.7)

Eq. 2.2.7 indicates that the expression of free energy A for ion dipole interaction depends on temperature. As a result, the total energy Eid for the ion-dipole interaction is not anymore equal with A. From the formula:  E¼A  T

vA vT

 (2.2.8) V

the result for the total energy ion-dipole is given by the expression: Eid z 

ðzeÞ2 $m2 3kb T ð4pε0 εÞ2 r 4

(2.2.9)

The range of energies involved in ion-dipole interactions are indicated in Table 2.2.1.

Dipole to dipole interactions Two polar molecules interact through Coulomb type forces, similarly to the interaction of ions and dipoles. In this case, the two dipole values are indicated as m1 and m2 , and they are considered as separated by the distance r. The values for the free energy for dipole-dipole interaction is obtained by considering all the Coulomb interactions between the charges þq1 , q1 and þq2 , q2 followed by the application of Boltzmann angle-averaging. The final result of the calculation gives the following expression for the free energy: A¼ 

m21 m22 3kB T ð4pε0 εÞ2 r 6

(2.2.10)

This angle-averaged energy between two dipoles is usually referred as Keesom interaction. Similar to the case of ion-dipole interaction, the angle-averaged expression for the free energy of two dipoles is temperature dependent. Using Eq. 2.2.8 that relates E and A, it can be obtained that the total energy of the interaction Edd is given by the formula: Edd ¼ 

2 m21 m22 3kB T ð4pε0 εÞ2 r 6

(2.2.11)

72

Method Development in Analytical HPLC

Ion to polarizable molecule interactions Besides the permanent dipole moment, molecules also have the capability to generate an induced dipole moment ! m ind . This is caused during molecular interactions, when the charges from the interacting molecules create an electric field. This field induces modifications in the charge distribution and as a result an induced dipole moment. The capability of a molecule to change its polarity generating an induced dipole ! moment ! m under the influence of an electric field E , is indicated as polarizability ind

a and is defined by the formula: ! m ind a¼ ! E

(2.2.12)

The SI units of polarizability a are C2 m2 J1 (C2m2J1 ¼ C m2 V1), but more  3 A ¼ 4pε0 1030 m3 ¼ 1.11 commonly the values are expressed in units of 4pε0   3 A the symbol 1040 C2 m2 J1 (sometimes for polarizability expressed in 4pε0  0

a is used, and the polarizability is indicated as polarizability volume). (Polarizability is in fact a tensor with nine components, each corresponding to the three components ! of ! m and the three components of E , but usually a is taken as the determinant of the ind

3x3 matrix). The influence of an ion placed at the distance r from a neutral molecule with a polar! izability has as a result the induction of a dipole moment ! m ind ¼ a E in the molecule. This induced dipole moment can be assumed as created by the displacement of a charge q at a distance d in the molecule. If the molecule is assumed to also have a permanent dipole m, the calculations for the free energy of this interaction is given by the formula: A¼ 

ðzeÞ2



2ð4pε0 εÞ2 r 4



m2 3kb T

 (2.2.13)

The second term in Eq. 2.2.13 is identical with the value from Eq. 2.2.9, indicating the interaction between an ion and a permanent dipole.

Dipole to molecule interactions The interaction of a dipole with a polarizable molecule can be evaluated using the same procedure as previously used for an ion to molecule interaction, considering that the induced dipole is generated by a charge q of the dipole. For two molecules each possessing permanent dipole moments m1 and m2 and polarizabilities a1 and a2 , the expression for the free energy of interaction becomes: A¼ 

m21 a2 þ m22 a1 ð4pε0 εÞ2 r 6

(2.2.14)

Retention and elution processes in high-performance liquid chromatography

73

This energy is known as Debye interaction energy. Since the free energy in Eq. 2.2.14 does not depend on temperature, formula 2.2.14 is also valid for the energy dipole-induced dipole Edid .

Nonpolar molecule to molecule interactions The interaction between two molecules that do not possess a dipole moment can be approached by considering that one molecule has the polarizability a and the other possesses an instantaneous dipole moment minst created by the orbiting frequency of its electrons. From quantum mechanical evaluations, it can be shown that the value of instantaneous dipole moment is minst ¼ 3=4 a$I for atoms and small molecules having the polarizability a and an ionization potential I. (The ionization potential is typically expressed in eV, 1 eV ¼ 1.60,218 1019 J). As a result, the free energy of interaction between two identical small molecules is given by the expression (ε ¼ 1): A¼ 

3a2 I

(2.2.15)

4ð4pε0 Þ2 r 6

For two dissimilar small atoms or molecules, the formula of free energy interaction becomes: A¼ 

3a1 a2 2ð4pε0

Þ2 r 6

I1 I2 ðI1 þ I2 Þ

(2.2.16)

This energy is known as London dispersion energy. Expression 2.2.16 for the free energy of interaction of two molecules is independent on temperature, and therefore the dispersion energy Ed ¼ A.

Total interaction between two molecules in the absence of ions In the absence of ions, the total interaction of molecules in a medium with dielectric constant ε is obtained by the sum Edd þ Edid þ Ed and is given by the expression: Et ¼ 

 2 2  2m1 m2 m21 a2 þ m22 a1 3a1 a2 I1 I2 1 þ þ 2 3k Tε2 2 6 r ε 2 I þ I B 1 2 ð4pε0 Þ 1

(2.2.17)

The contribution of different components of the total energy Et is displayed in Fig. 2.2.2 for two identical molecules for the distance r in the interval 3.0 to 7.5 Å at 300 K. The parameters used for the molecules are m ¼ 1.6 D, a ¼ 3.8 ($1.11 1040 C2m2J1), ε ¼ 1, and I ¼ 10.9 eV ($1.60218 1019 J). As shown in Fig. 2.2.2, the largest contribution to the total energy of interaction between two molecules in absence of ions is provided by the dispersion energy Ed . Dipole-dipole energy Edd represents about 30%e35% of Et , and the dipole-induced dipole interaction Edid accounts for only about 6% of the total energy.

74

Method Development in Analytical HPLC 0.0000 -2.0000

Ed-id

-4.0000

Ed-d

kJ mol−1

-6.0000

Ed

-8.0000 -10.0000

Et

-12.0000 -14.0000 -16.0000

-18.0000 -20.0000 3

3.5

4

4.5

5

5.5

6

6.5

7

7.5

Distance in A Figure 2.2.2 Values for the energies Edid ; Edd , Ed and total interacting energy Et for an idealized pair of molecules with m ¼ 1.6 D, a ¼ 3.8 ($ 1.11 1040 C2m2J1), ε ¼ 1, and I ¼ 10.9 eV, at T ¼ 300 K.

The total energy of interaction taking place between molecules in the absence of ions are indicated as van der Waals interactions and they correspond to a free energy AvdW ¼ Et .

Lennard-Jones potential The expression 2.2.17 for the total energy of interaction between two molecules accounts only for their attraction that has an energy proportional with 1= r 6 . However, in reality, the attraction between molecules is maximum (in absolute value) to a specific distance r, and at very small molecular distances a strong repulsive force intervenes, not allowing the overlapping of the two molecules. In the approach of considering the atoms incompressible spheres, each atomic sphere having a specific radius known as van der Waals radius of the atom, two molecules cannot approach each other beyond the van der Waals radius of their closest atoms. In order to account for the impossibility of two molecules to become closer to each other than the van der Waals radii of their closest atoms, a repulsion potential must be included in the description of the total intermolecular interaction. A semiempirical formula known as Lennard-Jones potential ELJ has been developed to describe the

Retention and elution processes in high-performance liquid chromatography

75

interaction between molecules as a function of their distance r. The energy for Lennard-Jones potential is given by formula:  ELJ ðrÞ ¼  4Et ðr0 Þ

r0



12 

21=6 r

6 

r0

(2.2.18)

21=6 r

where Et (having a negative value) is calculated using Eq. 2.2.17 and r0 is equal with van der Waals distance between the interacting particles (multiplication with N is necessary to obtain values per mol). In Fig. 2.2.3 is displayed the dependence of ELJ energy on the distance r in the interval 3.5 to 7.5 Å between molecules with the values for Et obtained with Eq. 2.2.17. The other utilized parameters include m ¼ 1.6 D, a ¼ 3.8 ($ 1.11 1040 C2m2J1), ε ¼ 1, and I ¼ 10.9 eV, at 300 K, and r0 ¼ 4 Å. The repulsive forces between molecules, as described by Eq. 2.2.18 are sometimes described together with the attractive forces as van der Waals interactions.

Formation of hydrogen bonds Another type of molecular interaction is the formation of hydrogen bonds. The hydrogen bonds are partly electrostatic and partly covalent and involves an electron donor/electron acceptor interaction. In this bond, a hydrogen atom covalently bonded in the form A⎼H is attracted by an atom : B containing a free electron pair. This leads to

0.0000

-1.0000

-2.0000

kJ mol−1

EL-J -3.0000

Et -4.0000

-5.0000

-6.0000

-7.0000 3.5

4

4.5

5

5.5

6

6.5

7

7.5

Distance in A Figure 2.2.3 Values for the energies Et and EL-J for an idealized pair of molecules with m ¼ 1.6 D, a ¼ 3.8 ($ 1.11 1040 C2m2J1), ε ¼ 1, and I ¼ 10.9 eV, at 300 K and r0 ¼ 4 Å.

76

Method Development in Analytical HPLC

Figure 2.2.4 Examples of formation of hydrogen bonds.

a strong polarization of the A⎼H bond and to electrostatic interactions with the formation of a structure of the type Hdþ : Bd . The hydrogen bonds are typically formed between molecular groups such as OeH or SeH where the hydrogen has a partial positive charge (qþ ) and an electronegative atom such as O or N that has a nonbonding orbital (lone pair) occupied with an electron pair and that can point toward the -H atom. Some examples of formation of hydrogen bonds are given in Fig. 2.2.4. The energy (enthalpy) of hydrogen bonds varies from (absolute) values less than about 20 kJ mol1 for the weak hydrogen, to common hydrogen bonds having between 20 kJ mol1 to 40 kJ mol1 (the values of forming the bonds must be taken with negative sign and of breaking the bonds with positive sign). The enthalpy of hydrogen bonding depends on the nature of the atoms to which the hydrogen in bound and also of the acceptor atom (e.g., OeH.N with about 29 kJ mol1, OeH.O with about 21 kJ mol1, NeH.N with about 13 kJ mol1, NeH.O with about 8 kJ mol1, etc.), but also on the specific molecules that interact, and of temperature. Some uncommon strong hydrogen bonds are known as present in HF, or between compounds having the electron receptor in ionic form. A comparison of the energy involved in various types of bonds is indicated in Table 2.2.1. Hydrogen bonding being created between molecules not having ionic charges, is sometimes included in the generic term of van der Waals interactions.

Halogen bonding (X-bonding) This type of interaction is similar to H-bonding, but involves a halogen atom from a molecule that is electrostatically attracted to the negative charge of an atom from another molecule. Among the halogen atoms from a molecule fluorine atom is the least halogen prone to participate to an X-bonding, being the less polarizable halogen atom and can act only as a donor center in this type of interaction. Thus, the strength scale of X-bonding is F < Cl < Br < I. The range of energies involved in this type of interaction is broader than the energy for H-bonding (Table 2.2.1) [8].

Other types of interaction Several other types of interactions can take place between molecules, besides the main types previously described. One such type of interaction is the charge transfer or donor-acceptor type, which occurs between electron pair donor and electron pair acceptor compounds. Depending on the type of electrons exchange in such interactions, the molecules involved in the charge transfer are indicated n; s; or p donors

Retention and elution processes in high-performance liquid chromatography

77

(Lewis bases) or acceptors (Lewis acids). The interacting energy of charge transfer interaction is usually relatively low, at about 10 kJ mol1, but in specific cases can reach up to 180 kJ mol1. An example of charge transfer interaction with application in HPLC and used for the separation of cis-trans olefins is based on the interaction of silver cations with unsaturated compounds to form weak charge transfer complexes with olefinic double bonds [9]. Another type of interaction is the inclusions in a host molecule. Specific molecules indicated as host, such as cyclodextrins (a, b, and g) and cucurbituril, contain cavities with dimensions comparable with that of small molecules and are able to form inclusion complexes with these small molecules. The interactions in inclusion complexes comprise polar interactions, hydrogen bonding, and possibly charge transfer. The reason for placing inclusions in a host molecule separately from the typical ones is that only the specific shape of the host molecule allows the formation of a stable complex, the other interactions being too weak to justify this stability. Other types of interactions are also known such as stacking.

Solvation energy of an ion The results regarding the interaction between two molecules in a medium with the dielectric constant ε offer only the first step toward the understanding of the interactions as they really take place. For example, for an ion that is not interacting with other ions (or electrical charges) a specific free energy equal to the electrostatic work done for forming that ion must be considered. For an ion with the charge q and radius r, the increase of the charge with dq will require the energy given by the expression: dA ¼

qdq 4pε0 εr

(2.2.19)

The total free energy for charging one ion to the final charge ze will be: Zze A¼ 0

qdq ðzeÞ2 ¼ 4pε0 εr 8pε0 εr

(2.2.20)

The energy given by Eq. 2.2.20 is also known as Born energy. For a mole of ions, expression 2.2.20 will be multiplied by N . The Born energy given by Eq. 2.2.20, but with a negative sign, can be considered as approximating the energy necessary to bring a mole of ions from vacuum into a solvent with a high dielectric constant (ε >>1). The expression of the free energy (at constant volume) for this transfer is given by the formula: DA ¼  N

  ðzeÞ2 1 ðzeÞ2 z N 1 ε 8pε0 r 8pε0 r

(2.2.21)

78

Method Development in Analytical HPLC

For this reason, Born energy is also indicated as solvation energy of ions (not considering the energy necessary to create the cavity in the solvent for accommodating the ions). The comparison of interacting energies for ionic forces with the energies of other forces (polar-polar, between polarizable molecules, etc.) indicates that ionic forces are comparably strong and act at longer distances. For this reason, the ionic interactions in types of chromatography not expected to involve ions may still be affected significantly and in an unpredictable way by ionic interactions.

The effect of a solvent on molecular interactions Solvent effects on molecular interactions are more complicated than simply considering the dielectric constant ε of the solvent in the interaction formulas. One simple procedure for accounting for the solvent contributions is the use of Born model (e.g., Ref. [10]). In this model, a solvation energy Esolv for a molecule X surrounded by a number of molecules Bi is given by formula: 1 Esolv ðXBÞ ¼  4pε0

q2X XqX qBi þ 2rX rXBi i

! 1

1 ε

 (2.2.22)

where qX and qBi are the partial point charges on the molecules, rXBi are the distances between atoms, rX is an effective radius of atom X, and ε is the dielectric constant of the solvent [11]. The problem with Eq. 2.2.22 is that the partial point charges on the molecules are usually not known. The effective atomic radii, which are a measure of the distance from the nucleus to the boundary of the surrounding cloud of electrons, are reported in the literature [12,13]. Numerous other studies attempting to account for the influence of a solvent on the molecular interactions have been published [14e17]. The properties of solutions are complex, and a large amount of information is available on this subject (e.g., Ref. [18]). The differences between ideal solutions and real mixtures, for example, are related to important concepts used in chemistry such as activity (see Eq. 2.1.5) and excess thermodynamic properties, which are defined as the difference between the value of the property in a real mixture and the value that would exist in an ideal solution under the same conditions.

Solvophobic effects Solvophobic effect describes the dissolution of a solute molecule in a solvent bulk as taking place in two steps: the first step consists in the formation of a cavity in the solvent bulk with dimensions suitable for the solute molecule, and the second step consists in the placing of the solute molecule in the created cavity when interactions with the surrounding liquid medium take place. This process is shown schematically in Fig. 2.2.5. Because it can be assumed the dissolution process takes place at constant volume, the change in standard free enthalpy DG0 for the dissolution is equal with the change in 0 free energy DA0 . Therefore, the free energy Asol for X;B (symbol D and the index

Retention and elution processes in high-performance liquid chromatography

79

Figure 2.2.5 Schematic of the dissolution of a nonpolar molecule in a polar solvent (water).

0;sol standard expressions are omitted for simplifying the notation and Asol X;B is DAX;B ) necessary for placing a molecular species X into a solvent formed by molecules B can be expressed by the free energy required for the creation of the cavity in the solvent to accommodate the species X indicated as Acav X;B and the free energy of van der Waals interactions between the molecule X and the surrounding molecules B, indicated as AvdW X;B . The free energy for the creation of the cavity is positive since it requires external energy to create the cavity and Acav X;B > 0. On the other hand, the van der Waals term vdW AvdW X;B is negative (AX;B < 0) since it generates energy due to the interactions between the solute and the solvent molecule. With these notations, the free energy Asol X;B for the dissolution process has the expression: cav vdW Asol X;B ¼ AX;B þ AX;B

(2.2.23)

In the development of Eq. 2.2.23 the term AvdW X;B includes all the possible interactions of the solute X with the solvent B. However, the potential hydrogen bonding between solute and solvent molecules is not seen explicitly the in term AvdW X;B for van der Waals interactions. As described previously, when hydrogen bonds are formed, the same type of interactions of the type of strong polar interactions take place. For this reason, the general term van der Waals interactions is used as an inclusive term for all solutesolvent type of interactions. The total interaction energies have negative values when the dissolution process is favorable and positive ones when it is not. The equilibrium: X % X sol

(2.2.24)

sol is displaced toward placing X in solution when Asol X;B is negative. As a result, AX;B is

vdW negative when Acav X;B < AX;B (the vertical bars indicate absolute value). The

80

Method Development in Analytical HPLC

vdW expressions for the terms Acav X;B ; AX;B were reported in the literature [19]. The values for Acav X;B are higher for polar solvents that have strong interactions and possibly hydrogen

bonds between their molecules. The values of AvdW X;B (in absolute value) are also higher when the compounds X are polar compared with those for X nonpolar. Solvophobic theory can be utilized for establishing an expression for the constant KðXÞ describing equilibrium 2.1.1. The formula developed in this theory (e.g., Ref. [20]) for the value of ln KðXÞ is the following: ln KðXÞ ¼ aA

X

þ bðVX Þ2=3 A

X

þ cX m2X þ dX aX

(2.2.25)

In Eq. (2.2.25), A X is the surface of the cavity in the solvent necessary to accommodate the molecule X, VX is the molar volume of X, mX is the dipole moment, and aX the polarizability of X. Parameters a, b, cX , and dX have rather complicated expres0 sions. Parameter a for example, depends directly on g the superficial tension of the mobile and stationary phase (assumed to be a liquid) by the formula: a¼

 N  0 gmo  gst0 RT

(2.2.26)

The dependence of a on superficial tension indicates that in HPLC retention is favored when the mobile phase as a higher superficial tension. The other parameters in Eq. 2.2.25 also depend on various physical properties of the molecules involved in the equilibrium and their expressions that are more complicated are reported in the literature (e.g., Ref. [19]).

Chaotropic and kosmotropic interactions Significant contribution to the molecular interactions in a solution may also come from the presence of other solutes, besides a specific analyte. The addition of certain solutes that decrease the polarity of a solution may affect significantly other solutes interactions and the interaction of solutes with the solvent. One type of such solutes is making water, for example, to have a lower energy of interaction between its own molecules, and are known as chaotropes. Some inorganic ions can act as chaotropes, their disruptive character growing in the order: CH3SO-3, < CF3COO, < BF-4, < ClO-4, < PF-6. Among chaotropes are also n-butanol, ethanol, guanidinium chloride, NaClO4, CH3COONa, NaCl, phenol, 2-propanol, sodium dodecyl sulfate, thiourea, and urea. In general, large ions or ions with low charge density such as SCN, Br, I, Kþ, Csþ act as chaotropes. In solutions of mixed solvents containing water, the chaotropes also disrupt the solvation shell of other solutes, modifying their interactions [21]. The opposite effect of contributing to the stability and structure of water-water interactions is known as kosmotropic effect. This can be achieved by neutral molecules, such as carbohydrates, proline, tert-butanol, and by small or high charged 2 2 2þ þ 2þ 3þ ions, such as CO2 3 , SO4 , HPO4 , Mg , Li , Zn , or Al , and salts such as (NH4)2SO4. Ionic kosmotropes are characterized by strong solvation energy leading

Retention and elution processes in high-performance liquid chromatography

81

to an increase of the overall cohesiveness of the water solution, which is also reflected by the increase of the viscosity and density of the solution [22]. Kosmotropic anions are more polarizable and hydrate more strongly than kosmotropic cations of the same charge density.

Key points • • •

A number of intermolecular interactions can be involved in the separation process taking place in HPLC. Main molecular parameters influencing interactions are partial charges q, permanent dipole moment m, polarizability a ionization potentials I, for solute molecules but also, including dielectric constants ε, for the mobile phase and stationary phase molecules. Besides intermolecular interactions, the interactions in the bulk of a solvent are important in HPLC.

2.3

A classification of analytical HPLC types

The classification of analytical HPLC types can be obtained using various criteria. For example, analytical HPLC can be differentiated based on the dimensions of the HPLC column in the following subtypes: conventional, narrow-bore, microbore, micro LCcapillary, and nano LC-capillary [23e25]. In the present book, most discussions will refer to conventional, narrow-bore and microbore analytical HPLC. Another classification considers if the mobile phase composition is kept constant during the separation (isocratic separation) or it is modified (gradient separation). This classification does not refer to the principle of separation being just a descriptor for the analytical method procedure. As previously indicated, the stationary phase in HPLC is made from porous particles or porous monolithic rods. A distinction can be made if the particles used in the chromatographic column have a diameter larger than 2.5 mm (e.g., 5.0 mm) or smaller (e.g., 1.7 mm). The use of small particles typically provides a better separation but at the same time requires higher pressures for moving the mobile phase through the column (e.g., higher than 5000 psi or 345 bar). The type of HPLC using smaller particles is indicated as ultra performance liquid chromatographydUPLC or UHPLC. However, in this book, the differentiation HPLC versus UPLC is not emphasized and the term HPLC will be used in the inclusive manner (meaning both HPLC and UPLC). Only when emphasizing differences between the two procedures, UPLC will be differentiated from HPLC that is performed at pressures of the mobile phase below 5000 psi. Another criterion of classification is based on the temperature range in which the separation is performed. The usual temperature for performing HPLC (or UPLC) is between 20 and 60 C. Low temperature (down to 10 C), and high temperature (up to 250 C) can also be used for HPLC, although these utilizations are not common. An useful classification for method development purposes is based on the nature of stationary phase and mobile phase used for the separation. Another useful classification is based on the type of detection utilized. These two types of classification are further commented.

82

Method Development in Analytical HPLC

A classification of HPLC types based on the nature of stationary and mobile phase The stationary phase and mobile phase in HPLC (and UPLC) are selected in connection with the nature of the analytes and with the interactions generated between analytes, the stationary phase and mobile phase. Therefore, the classification further discussed is useful in describing a selected analytical method applied for solving a specific separation/analysis problem. The selection of the most adequate type of HPLC from those further listed types is part of the subject of the present book, and in this section is given only a short overview on the main characteristics of those types. 1) Reversed-phase HPLC (RP-HPLC) is the most common HPLC technique, and a very large number of compounds can be separated by RP-HPLC. This type of chromatography is performed on a nonpolar stationary phase with a polar mobile phase containing water. The stationary phases for RP-HPLC are made, for example, by chemically bonding long hydrocarbon chains (C18, C8, etc.) on a solid surface such as silica. The mobile phase in RP-HPLC is typically a mixture of an organic solvent (CH3OH, CH3CN, isopropanol, etc.) and water. Buffers and other additives can also be added to the mobile phase in RPHPLC. The main equilibrium in RP-HPLC is partition (Eq. 2.1.1), and the main type of molecular interactions between analyte and stationary phase are indicated as hydrophobic interactions. 2) Ion-pair chromatography (IPC) is applied in particular to strongly polar or even ionic compounds. This type of chromatography is in fact a type of RP-HPLC, with the difference of having a mobile phase in which a special additive (ion pairing agent or IPA) is placed to generate a molecular association analyte-IPA amenable to be separated by RP-HPLC. 3) Hydrophobic interaction chromatography (HIC) is a type of RP-HPLC, sometimes indicated as a milder RP-HPLC, applied to the separations of proteins and other biopolymers. The technique is based on interactions between nonpolar moieties of a protein having solvent-accessible nonpolar groups (hydrophobic patches) on the surface of a hydrophilic stationary phase (e.g., hydrophobic ligands coupled on cross-linked agarose). The promotion of the hydrophobic effect by addition of salts (such as ammonium sulfate) in the mobile phase drives the adsorption of hydrophobic areas from the protein to the hydrophobic areas on the stationary phase. The reduction of the salting out effect by decreasing the concentration of salts in solution leads to the desorption of the protein from the solid support. 4) Nonaqueous reversed-phase chromatography (NARP) is an RP-HPLC type utilized for the separation of very hydrophobic molecules such as triglycerides. In this type of chromatography, the stationary phase is nonpolar (similar to RP), while the mobile phase, although with some polarity, is nonaqueous (usually a mixture of nonpolar and some polar organic solvents) and capable of dissolving the strongly hydrophobic molecules. 5) Hydrophilic interaction liquid chromatography (HILIC) is a type of HPLC applied for polar, weakly acidic or weakly basic compounds. In this type of HPLC the stationary phase is polar and the mobile phase is less polar than the stationary phase. For HILIC, the polar stationary phase is typically made by chemically bonding molecular fragments with a polar end group (diol, amino, special zwitterionic, etc.) on a solid support such as silica, or is simply made of silica. The mobile phase in HILIC, although less polar than the stationary phase, contains water soluble solvent such as CH3OH or CH3CN, and also a certain proportion of am aqueous component. The separation is based on the difference in polarity between the molecules of the sample. Ion-polar interactions and even hydrophobic

Retention and elution processes in high-performance liquid chromatography

6)

7)

8)

9)

10)

11)

12)

13)

83

interactions may also play a role for the separation in HILIC. A special type of HILIC (eHILIC or ERLIC) is performed on an ion exchange stationary phase with the mobile phase containing a high proportion of an organic solvent [26]. Normal phase chromatography (NPC or NP-HPLC) is a chromatographic type that uses a polar stationary phase and a nonpolar mobile phase for the separation of polar compounds. The nonpolar mobile phases are solvents such as hexane, cyclohexane, CH2Cl2, tetrahydrofuran, etc. that are not water miscible. In some special separations, the nonaqueous solvent may also contain polar additives such as trifluoroacetic acid. Normal phase chromatography does not have a major difference from HILIC except for the absence of water in the mobile phase. However, instead of being applied to polar molecules, NPC is applied to hydrophobic molecules. Aqueous-normal-phase chromatography (ANPC or ANP) [27] is a technique performed on a special stationary phase (silica hydride) and the mobile phase covers the range including the types used in reversed-phase chromatography and those used in normal-phase chromatography. The mobile phases for ANP are based on an organic solvent (such as methanol or acetonitrile) with a certain amount of water such that the mobile phase can be both “aqueous” (water is present) and “normal” (less polar than the stationary phase). Polar solutes are most strongly retained in ANP, with retention decreasing as the amount of water in the mobile phase increases [28]. Chiral chromatography on chiral stationary phases is a type of HPLC used to separate chiral compounds. The technique requires special stationary phases containing a chiral selector. The mobile phase is frequently nonaqueous (less polar than the stationary phase) and this chiral HPLC is indicated as normal phase (NP) chiral HPLC. Some chiral chromatographic columns may be used with water in the mobile phase and the technique is indicated as reversed phase (RP) chiral HPLC. Chiral chromatography on achiral stationary phases is also possible for some chiral solutes, by using chiral modifiers in the mobile phase, although the stationary phase is not chiral. This type of HPLC separates the diastereoisomers formed between the analytes and the chiral modifier from the mobile phase. Cation-exchange chromatography is a type of HPLC used for the separation of cations (inorganic or organic). In cation exchange chromatography the ionic analytes consisting of the cationic species M zþ are retained by a cation exchange stationary phase containing ionic groups of the type resX  (X covalently bonded to the support res). Two different analyte cations M1zþ and M2zþ , can be separated based on their retention strength. Anion-exchange chromatography is a type of HPLC used for the separation of anions (inorganic or organic). This HPLC is similar in principle with cation exchange type, but the anionic species B from solution are retained by an anion exchange stationary phase containing ionic groups of the type resY þ (Y covalently bonded to res). The mobile phase in ion-exchange chromatography usually consists of buffer solutions. Cation exchange and anion exchange type chromatography are together indicated as ion exchange chromatography or IEC. Ion exchange on amphoteric or on zwitterionic phases is a type of IEC very similar in principle with the cation-exchange or anion-exchange IEC. The stationary phase for this type of IEC contains groups that have an amphoteric character or in the case of zwitterionic phases, both anionic and cationic groups. The mobile phase in these types of chromatography also consists of buffer solutions. Ion exclusion chromatography is an HPLC technique in which an ion exchange resin is used for the separation of neutral species between them and from ionic species. In this

84

14)

15)

16)

17)

18)

19)

20)

21)

22)

Method Development in Analytical HPLC

technique, ionic compounds from the solution are rejected by the selected resin, and they are eluted as nonretained compounds. Nonionic or weakly ionic compounds penetrate the pores of the resin and are retained selectively as they partition between the liquid inside the resin and the mobile phase. Ligand exchange chromatography is a type of chromatography in which the stationary phase is a cation exchange resin loaded with a metal ion (e.g., of a transitional metal) that is able to form coordinative bonds with the molecules from the mobile phase. The elution is done with a mobile phase able to displace the analyte from the bond with the metal, and the separation is based on the differences in the strength of the interaction (of coordinative type) of these solutes with the bonded metal ion [29]. Immobilized metal affinity chromatography or ion chelation chromatography is closely related to ligand exchange chromatography and uses a resin containing chelating groups that can form complexes with metals such as Cu2þ, Ni2þ, Zn2þ, etc. The metal ions loaded on the resin still have coordinative capability for other electron donor molecules such as proteins. The retained analytes can be eluted by destabilizing the complex with the metal, e.g., by pH changes or addition of a displacing agent such ammonia in the mobile phase [30]. Ion-moderated chromatography is an HPLC technique similar to ligand exchange chromatography with the difference that the stationary phase loaded with a metal ion (e.g., Ca2þ, Naþ, Kþ, Agþ, or even Hþ) does not form coordinative bonds with the analyte, the interactions being based mainly on polarity [31]. Gel filtration chromatography (GFC) is a type of size exclusion chromatography (SEC) in which the molecules are separated based on their size (more correctly, their hydrodynamic volume). In gel-filtration an aqueous (or containing a polar solvent and water) solution is used to transport the sample through the column and is applied to molecules that are soluble in water and polar solvents. Size exclusion chromatography uses porous particles with a variety of pore sizes to separate molecules. GFC is sometimes indicated as aqueous SEC. Gel permeation chromatography (GPC) is another type of size exclusion chromatography (SEC), the only difference from gel filtration being the mobile phase, which in this case is an organic solvent usually with hydrophobic character. The technique is used mainly for the separation of hydrophobic macromolecules (such as solutions of certain synthetic polymers). GPC is sometimes indicated as nonaqueous SEC. Interaction polymer chromatography is a type of size exclusion chromatography with an added different separation involving enthalpic separation interactions such that weak attractive enthalpic interaction effect is exactly compensated by entropic SEC exclusion effect [32]. Displacement chromatography is a chromatographic technique where all the molecules of a sample are initially retained on a chromatographic column (loading phase). After the sample is loaded, a “displacement” reagent dissolved in the mobile phase is passed through the column and elutes the specific retained molecule. The method is more frequently applied as a preparative chromatographic technique than as an HPLC analytical method [33,34]. Affinity chromatography is a liquid chromatographic technique typically used for protein and other bio-molecules separation when commonly indicated as bioaffinity chromatography. It can be practiced on a variety of specifically made stationary phases that allow selective retention of the analytes based on affinity interactions [35]. Multimode HPLC (or mixed mode HPLC) is a type of chromatography in which the stationary phase contains by purpose more than one type of functionality, for example, some with bonded nonpolar groups (e.g., -C18), and some with ionic groups (e.g., ¼ SO-3). This type of stationary phases are able to participate to multiple interactions with solutes having

Retention and elution processes in high-performance liquid chromatography

85

different molecular characteristics, thus contributing to their retention on the stationary phase [36]. 23) Other types of chromatography are also practiced in some studies. One such example is biomimetic phospholipid membrane chromatography (BPMC), a type of chromatography in which the role of stationary phase is played by a phospholipid membrane (phospholipid monolayer/bilayer and intact liposomes) that mimics the membrane of a biological cell. Another example is micellar liquid chromatography, which uses a concentration of a surfactant in the mobile phase above the critical micellar concentration. Also, special variants of the previously listed chromatography types are sometimes utilized such as chromatofocusing, which is a type of ion exchange chromatography, dynamic HPLC, etc.

A classification of HPLC types based on the utilized detection Part of an analytical HPLC method is also the detection of the analytes and this detection can be performed using one or more detection techniques simultaneously. The selection of a detection depends mainly on the properties of the analytes, but also must consider the nature of mobile phase, required sensitivity for detection, etc. Depending on the detection techniques, the HPLC can be classifies as: (1) HPLC-RI that uses an refractive index detector, (2) HPLC-UV (HPLC- UV-Vis) that can use a detector for a fixed UV-Vis absorption or an array (multiple wavelength) detector (DAD), (3) fluorescence-HPLC that uses a detector for fluorescence (FLD), (4) LC-MS that uses a mass spectrometer for detection with a number of options regarding the type of mass spectrometer, (5) LC-ELSD that uses an evaporative light scattering detector, (6) LC-ECD that uses an electrochemical detector with a variety of types of electrochemical detection. Some other detectors can be used in HPLC, and the subject of detection in HPLC will be later elaborated in Chapter 3.

Key points • • •

The classification of HPLC types based on the nature of stationary and mobile phase is useful for method development purposes. This classification is also related to the nature of the analytes and the interactions taking place between them. The detection in HPLC is selected based on the properties of the analytes, the nature of mobile phase, the required sensitivity for detection.

2.4

Influence of pH, ionic strength, and temperature on separation equilibria

A series of experimental factors have significant effects on the retention of analytes. These factors can affect directly the equilibrium mo/st, or indirectly by involving analyte molecules in secondary equilibria, which at their turn have effects on the core retention equilibrium. For example, the pH of the mobile phase and the acid-base properties of the analytes to be separated play an important role in

86

Method Development in Analytical HPLC

separation equilibria. Also, as any chemical equilibrium, the separation equilibrium is influenced by temperature. The pH, temperature, as well as the role of additives not directly involved in the equilibrium but influencing it are summarily presented in this section.

The role of pH in separation equilibria The pH is defined as the negative value of the logarithm in base 10 of the activity of hydrogen ions Hþ in a solution, or pH ¼ - log10 aðHþ Þ. Since pH is in most cases determined in water or water containing solutions, the pH is also indicated as the negative value of the logarithm of the activity of hydronium ions H3Oþ, or pH ¼ - log10 aðH3 Oþ Þ. Based on Eq. 2.1.5 defining activity as the product between molar concentration and activity coefficient, the pH can be defined by the expression:

pH ¼  log H3 Oþ  log g H3 Oþ

(2.4.1)

For diluted solutions, g (H3Oþ) z 1 and Eq. 2.4.1 is reduced to the formula:

pH ¼  log H3 Oþ

(2.4.2)

A large group of molecules can dissociate in an aqueous medium by releasing protons and are indicated as (Brönsted-Lowry) acids, or by accepting protons and are indicated as (Brönsted-Lowry) bases, or being able to both accept or donate protons (amphoteric compounds). The dissociation of a simple acid HX, for example, takes place as follows: HX % Hþ þ X 

(2.4.3)

The equilibrium constant Ka for the acid known as acidity constant (usually expressed in logarithm form - log Ka ¼ pKa ) is defined by the formula: Ka ¼

½Hþ ½X   ½HX

(2.4.4)

Water also dissociates following the equilibrium: H2 O % Hþ þ OH

(2.4.5)

For water, the dissociation constant is given by the expression: K¼

½Hþ ½OH  ½H2 O

and



K$½H2 O ¼ Kw ¼ Hþ ½OH 

(2.4.6)

Retention and elution processes in high-performance liquid chromatography

87

The constant Kw is indicated as the ionic product of water and Kw ¼ 1:008$1014 ðat 25o CÞ. A basicity constant Kb can be assigned to a base BOH that is dissociated by the equilibrium: BOH % Bþ þ OH

(2.4.7)

This basicity constant has the following expression: Kb ¼

½Bþ ½OH  ½BOH

(2.4.8)

An acidity constant can also be obtained for a base BOH from the following dissociation: þ BOHþ 2 % BOH þ H

(2.4.9)

In this case, the acidity constant Ka is defined for a base by the formula: ½Hþ ½BOH

Ka ¼ BOHþ 2

(2.4.10)

Since [Bþ ] ¼ [BOHþ 2 ], by multiplying Eqs. 2.4.8 and 2.4.10, the result is the following: Ka Kb ¼ Kw

(2.4.11)

Besides dissociation with the formation of single ions, some molecules can have more than one ionizable group and acidity constants (or basicity constants) can be assigned for each dissociation step. In HPLC, the molecules of a sample are injected in the mobile phase that can be selected as having a specific pH, i.e., a specific concentration of Hþ ions. By imposing this pH to equilibria of ionizable compounds of an analyte (mobile phase is in large excess compared with the analyte), these equilibria can be displaced in a specific direction. Taking as an example the simple equilibrium described by Eq. 2.4.3, an excess Hþ ions (low pH) will displace the equilibrium toward the formation of more HX not ionized molecules, while a higher pH will decrease the formation of HX and produce more X  ions. The pH of the medium is determining the proportion of different forms (acidic, neutral, or basic) of a specific compound, depending on pKa of specific moieties of the molecule. For a simple acid HX, Eq. 2.4.4 can be written in the form: ½HX 10pH ¼ ½X   þ ½HX Ka þ 10pH

(2.4.12)

88

Method Development in Analytical HPLC

100 90

% Composition

80

CH3COO-

CH3COOH

70 60 50 40 30 20 10

0 0

2

4

4.756

6

pH

8

10

12

14

Figure 2.4.1 Composition % of the two forms of CH3COOH as a function of pH.

Expressed in %, the ratio ½HX=ð½X   þ½HXÞ from Eq. 2.4.12 multiplied with 100 will gives the proportion of HX in the mixture of the two components, which is pHdependent. The graph indicating this dependence on pH is shown in Fig. 2.4.1 for CH3COOH, which has pKa ¼ 4.756. More complex acidic, basic, or amphoteric molecules are also present in solution with different components composition as a function of pH. It is common that within 2 pH units, the structure of a compound is changed by 98%, for example, from a completely neutral state into an ionized form. The interactions taking place in the separation process between the molecules of the analyte, molecules of the mobile phase, and those of the stationary phase, are significantly influenced by the charges and polarity of the molecules, as described in Section 2.2. As a result, the pH of the mobile phase has an important effect on separation when the solutes contain ionizable functional groups such aseOH, eSH, eCOOH, eNH2, and others such groups. Taking a simple example of an acid HX that is partly dissociated and takes part in a distribution equilibrium between an organic phase “o” and an aqueous phase “w”, nonionized part of HX will be part of the equilibrium between “o” and “w” while the ionized part X  will remain in “w”. This is illustrated in Fig. 2.4.2.

Organic phase “o”

HX KHX

HX

Ka

H+ + X -

Aqueous phase “w”

Figure 2.4.2 Distribution of a weak acid between an organic and an aqueous phase.

Retention and elution processes in high-performance liquid chromatography

89

The distribution of all the species of the acid between the two phases, will be described by the distribution coefficient DHX given by the formula: DHX ¼

½HXo ½HXw þ ½X  w

(2.4.13)

The distribution of only HX molecules between “o” and “w” will be given by the ratio: KHX ¼

½HXo ½HXw

(2.4.14)

Eq. 2.4.13 for DHX can be used to obtain the relation between DHX and KHX by applying Eq. 2.4.4 for evaluating ½X  w : DHX ¼

½HXo 10pH ¼ K HX ½HXw ð1 þ Ka =½H þ w Þ Ka þ 10pH

(2.4.15)

The dependence given by Eq. 2.4.15 is illustrated in Fig. 2.4.3 for a compound with Ka ¼ 105, (pK a ¼ 5.0). At low pH values, DHX z K HX . At high pH values, DHX z 0; and at pH ¼ pKa , which is the inflection point of the sigmoid curve of dependence DHX z 0:5 K HX . In an HPLC separation, the equilibrium of the analyte between the mobile “w” and stationary phase “o” is determined for a compound capable to form ions by the distribution constant D, and the graph from Fig. 2.4.3 shows how pH can influence this

1

0.8

DHX / KHX

0.6

0.4

0.2

pKa

0 0

2

4

6

pH

8

10

12

14

Figure 2.4.3 The dependence of DHX =K HX as a function of pH for an acid HX (pK a ¼ 5.0).

90

Method Development in Analytical HPLC

distribution for a simple acid HX and using the assumption that species X  will remain only in aqueous phase. Evaluations of the relation between D and K can be made for the case X  also has some distribution between “o” and “w”, as well as for multiprotic acids, for bases, and for amphoteric compounds. All those dependencies show that pH significantly affects the value of D. In practice, the effect of pH of the mobile phase on the separation is usually not subject of precise calculations, but in all applications, it must be taken into account that pH can significantly affect separations. One side concern related to the pH of the mobile phase is related to the fact that the stationary phases of many chromatographic columns have good chemical stability only in a specific pH range. For this reason, strong acidic character or basic character of the mobile phase cannot be used if this deteriorates the chromatographic column.

The role of ionic strength and of additives on separation equilibria As indicated in Section 2.1, as well as in the definition of the pH, it is common to approximate the activities with concentrations by assuming activity coefficients as equal with 1.0 (g (X) z 1). However, this approximation is not very good for solutions containing higher concentrations or salts or even when the solution in which the analyte X is present is not an ideal solution. For example, for solutions containing an ion X z the activity coefficient is given by Debye-H€ uckel formula: log g X z ¼ 

pffiffi Az2 I pffiffi 1 þ as B I

(2.4.16)

In Eq. 2.4.16 I represents the ionic strength of the solution, as is an ion size parameter, and A and B are solvent and temperature-dependent parameters, respectively. The ionic strength in a solution depends on the concentration of species Xi and their net charges, zi , following the formula: I¼

1X ½Xi z2i 2 i

(2.4.17)

For nonionic species, the activity coefficient gðXÞ can be related to the change in the enthalpy of mixing when the compound X is placed in a solvent and does not form an ideal solution but a regular solution (solutions with mixing enthalpy different from zero [37]). For such a regular solution, the expression for the chemical potential of compound X is given by the expression: mX ¼ m0X þ

1 H X  H 0X þ RT ln½X nX

(2.4.18)

Retention and elution processes in high-performance liquid chromatography

91

where nX is the number of moles of compound X and H X  H 0X is the change in enthalpy caused by the interactions during mixing. By taking into consideration that aX ¼ gðXÞ$½X (see expression 2.1.5), the following formula is obtained for the activity coefficient: ln gðXÞ ¼

1 DHmix;X nX RT

(2.4.19)

The change in the enthalpy of mixing depends on the interactions of the solute with the solvent and the solvent can be a pure compound, a mixture of solvents, and also may contain additives. The nature of the solvent (solvents) as well as additives influence the interactions and these affect the activity coefficient and as a result of the effective concentration of the analyte. As a result, separations are affected when the activity coefficients are affected by additives. One type of additives that are known to strongly affect activity coefficients are the chaotropes and kosmotropes. As previously indicated in Section 2.2, chaotropes disrupt the hydrogen bonding, van der Waals forces, hydrophobic effects, and solvation of ions [38]. The kosmotropic ions are favorable for the formation of solvation shell of ionic species by contributing to the stability of water-water interactions [39].

The role of temperature on separation equilibria Both Eqs. 2.1.8 and 2.1.16 for the expression of equilibrium constant KðXÞ controlling the separation equilibrium either in partition model or in adsorption model, indicate that temperature is a parameter involved in this value. For the partition model, for example, the value of KðXÞ has the expression:   DH0 1 DS0 ln KðXÞ ¼ þ T R R

(2.4.20)

In Eq. 2.4.20, DH 0 and DS0 are the standard enthalpy and entropy changes, respectively, for the transfer of the analyte from the mobile to the stationary phase. Since the retention is an exothermic process, DH 0 has negative values and KðXÞ decreases as the temperature increases. Both DH 0 and DS0 can be considered temperature invariants. However, this assumption is only an approximation [40] and the following corrections can be made for the values of DH 0 and DS0 : DH 0T ¼ DH0T0 þ Cp ðT  T0 Þ

(2.4.21)

DS0T ¼ DS0T0 þ Cp lnðT = T0 Þ

(2.4.22)

92

Method Development in Analytical HPLC

In Eqs. 2.4.21 and 2.4.22, Cp is the heat capacity. Heat capacity also varies with temperature but this variation is very small. The correction factor in Eqs. 2.4.21 and 2.4.22 is not large and it is common to assume DH0 and DS0 as constants. Eq. 2.4.20 indicates a linear dependence of ln KðXÞ on ð1 =TÞ. However, in some equilibria the dependence may not be linear because the analyte is present in some separations in more than one form in solution (e.g., tautomers or mixture of ionic species) or because the HPLC process involves more than one mechanism of separation and each mechanism is differently influenced by temperature [41]. For analytical method development it is more convenient to represent the dependence of retention time tR (or ln tR ) on column temperature T. The main consequences of temperature dependence of retention from separation point of view are its influence on selectivity and improvements of peak symmetry. As viscosity of mobile phase changes with temperature, another effect is the decrease of column backpressure, which can be reduced with up to 40% in the normal interval 20e60 C typically varied in HPLC separations [42]. Besides these influences, the diffusion coefficients (see Section 5.1) and the dissociation processes of various species during the separation suffer modifications with temperature changes, which have influences on retention and finally on column stability [43].

Key points • •

The value of pH of the mobile phase, its ionic strength, and temperature are important parameters affecting separations. Additives to the mobile phase also play roles in separation.

References [1] H.P. Gregor, Gibbs-Donnan equilibria in ion exchange resin systems, J. Am. Chem. Soc. 73 (1951) 642e650. [2] T. Sun, R.R. Chance, W.W. Graessley, D.J. Lohse, A study of the separation principle in size exclusion chromatography, Macromolecules 37 (2004) 4304e4312. [3] C. Hansch, A. Leo, D. Hoekman, Exploring QSAR, Hydrophobic, Electronic and Steric Constants, 1995. ACS Washington. [4] G. Klopman, J.-Y. Li, S. Wang, M. Dimayuga, Computer automated log P calculations based on an extended group contribution approach, J. Chem. Inf. Comput. Sci. 34 (1994) 752e781. [5] http://www.epa.gov/oppt/exposure/pubs/episuite.htm. [6] http://www.chemaxon.com. [7] J. Israelachvili, Intermolecular & Surface Forces, Academic Press, Amsterdam, 1991. [8] G. Cavallo, P. Metrangolo, R. Milani, T. Pilati, A. Priimagi, G. Resnati, G. Terraneo, The halogen bond, Chem. Rev. 116 (2016) 2478e2601. [9] B. Nikolova-Damyanova, Retention of lipids in silver ion high-performance liquid chromatography: facts and assumptions, J. Chromatogr. A 1216 (2009) 1815e1824.

Retention and elution processes in high-performance liquid chromatography

93

[10] W.C. Still, A. Tempczyk, R.C. Hawley, T. Hendrickson, Semianalytical treatment of solvation for molecular mechanics and dynamics, J. Am. Chem. Soc. 112 (1990) 6127e6129. [11] G. Hefter, Ion solvation in aqueous-organic mixtures, Pure Appl. Chem. 77 (2005) 605e617. [12] E. Clementi, D.L. Raimondi, W.P. Reinhardt, Atomic screening constants from SCF functions. II. Atoms with 37 to 86 electrons, J. Chem. Phys. 47 (1967) 1300e1307. [13] J.C. Slater, Atomic radii in crystals, J. Chem. Phys. 41 (1964) 3199e3204. [14] M.E. Davis, J.D. Madura, B.A. Luty, J.A. McCammon, Electrostatics and diffusion of molecules in solutions: simulation with the University of Huston Brownian dynamics program, Comput. Phys. Commun. 62 (1991) 187e197. [15] M.E. Davis, J.A. McCammon, Calculating electrostatic forces from grid-calculated potentials, J. Comput. Chem. 11 (1990) 401e409. [16] P. Ferrara, J. Apostolakis, A. Caflisch, Evaluation of a fast implicit solvent model for molecular dynamics simulations, Prot. Struct. Funct. Genet. 46 (2002) 24e33. [17] V. David, N. Grinberg, S.C. Moldoveanu, Long-range molecular interactions involved in the retention mechanisms of liquid chromatography, Adv. Chromatogr. 54 (2017) 73e110. [18] R.S. Berry, S.A. Rice, J. Ross, Physical Chemistry, second ed., Oxford Univ. Press, New York, 2000. [19] O. Sinanoglu, B. Pullman (Eds.), Molecular Associations in Biology, Academic Press, New York, 1968, pp. 427e445. [20] S.C. Moldoveanu, E. Caiali, V. David, Phase ratio and equilibrium constant in RP-HPLC obtained from octanol/water partition constant through solvophobic theory, Chromatographia 80 (2017) 1491e1500. [21] R. LoBrutto, Y.V. Kazakevich, Chaotropic effects in RP-HPLC, Adv. Chromatogr. 44 (2005) 291e315. [22] M. Andreev, J. de Pablo, A. Chremos, J.F. Douglas, Influence of ion solvation on the properties of electrolyte solutions, J. Phys. Chem. B 122 (2018) 4029e4034. [23] D. Ishii, K. Asai, K. Hibi, T. Jonokuchi, M. Nagaya, A study of micro-high-performance liquid chromatography: I. Development of technique for miniaturization of highperformance liquid chromatography, J. Chromatogr. A 144 (1977) 157e168. [24] M.V. Novotny, Development of capillary liquid chromatography: a personal perspective minireview, J. Chromatogr. A 1523 (2017) 3e16.  ak, D. Moravcova, V. Kahle, Instrument platforms for nano liquid chromatography, [25] J. Sest J. Chromatogr. A 1421 (2015) 2e17. [26] D. Dorfel, S. Rohn, E. Jantzen, Electrostatic repulsion hydrophilic interaction liquid chromatography (ERLIC) for the quantitative analysis of polyamines, J. Chromatogr. A 1720 (2024) 464820. [27] J. Pesek, M.T. Matyska, A comparison of two separation modes: HILIC and aqueous normal phase chromatography, LCGC North Am. 25 (2007) 480e490. [28] J.J. Pesek, M.T. Matyska, R.I. Boysen, Y. Yang, M.T.W. Hearn, Aqueous normal-phase chromatography using silica-hydride-based stationary phases, Trends Anal. Chem. 42 (2013) 64e73. [29] G. Lodi, G. Storti, L.A.P.M. Morbidelli, Ion exclusion chromatography: model development and experimental evaluation, Ind. Eng. Chem. Res. 56 (2017) 1621e1632. [30] V. Gaberc-Porekar, V. Menart, Perspectives of immobilized-metal affinity chromatography, J. Biochem. Biophys. Methods 49 (2001) 335e360.

94

Method Development in Analytical HPLC

[31] A. Bhalkikar, C.M. Marin, C.L. Cheung, Method development for separating organic carbonates by ion-moderated high-performance liquid chromatography, J. Separ. Sci. 39 (2016) 4484e4491. [32] A.M. Striegel, Method development in interaction polymer chromatography, Trends Anal. Chem. 130 (2020) 115990. [33] M. Srajer Gajdosik, J. Clifton, D. Josic, Sample displacement chromatography as a method for purification of proteins and peptides from complex mixtures, J. Chromatogr. A 1239 (2012) 1e9. [34] H. Kalasz, Displacement chromatography, J. Chromatogr. Sci. 41 (2003) 281e283. [35] E.L. Rodriguez, S. Poddar, S. Iftekhar, K. Suh, A.G. Woolfork, S. Ovbude, A. Pekarek, M. Walters, S. Lott, D.S. Hage, Affinity chromatography: a review of trends and developments over the past 50 years, J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 1157 (2020) 122332. [36] N.-V.T. Nguyen, Perspective chapter: mixed-mode chromatography, in: S.C. Moldoveanu, V. David (Eds.), Analytical Liquid Chromatography New Perspectives, IntecOpen, London, 2022. [37] J.H. Hildebrand, R.I. Scott, The Solubility of Non-electrolytes, Dover Pub, New York, 1964. [38] Y. Marcus, Effect of ions on the structure of water: structure making and breaking, Chem. Rev. 109 (2009) 1346e1370. [39] C. Florez, Y. Kazakevich, 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. 36 (2013) 1138e1148. [40] M.M. Reif, P.H. H€unenberger, Computation of methodology-independent single-ion solvation properties from molecular simulations. III. Correction terms for the solvation free energies, enthalpies, entropies, heat capacities, volumes, compressibilities, and expansivities of solvated ions, J. Chem. Phys. 134 (2011) 144103. [41] M. Tanase, A. Soare, V. David, S.C. Moldoveanu, Sources of nonlinear van’t Hoff temperature dependence in high-performance liquid chromatography, ACS Omega 4 (2019) 19808e19817. [42] J. Haun, T. Teutenberg, T.C. Schmidt, Influence of temperature on peak shape and solvent compatibility: implications for two-dimensional liquid chromatography, J. Separ. Sci. 35 (2012) 1723e1730. [43] S.J. Marin, B.A. Jones, W.D. Felix, J. Clark, Effect of high-temperature on highperformance liquid chromatography column stability and performance under temperature-programmed conditions, J. Chromatogr. A 1030 (2004) 255e262.

Instrumentation in highperformance liquid chromatography 3.1

3

Overall description of an HPLC instrument

Standard HPLC instrumentation The instrumentation for high-performance liquid chromatography (HPLC) analysis is designed to perform a number of operations necessary for the separation and detection of the components of a sample injected in the instrument. These operations include the capability to mix solvents at a desired mobile phase composition, to pump the mobile phase at high pressure, to load the liquid sample in the mobile phase flow, to separate (at least in part) the sample components by passing the mobile phase loaded with the sample through a chromatographic column, to detect the separated components with detector(s) that generate an electrical signal, to record the signal and display it in the form of a chromatogram. This process is controlled in modern HPLC instruments using a computer with a dedicated software package. The instrumentation made for achieving this process has numerous variations depending on its manufacturer, but the main components of the instruments are about the same. A typical such construction/configuration is illustrated in Fig. 3.1.1 and can be considered as having several modules: solvent supply, high-pressure pump(s), injector (autosampler), column in a column holder, one or more detectors, and the computer controlling the system through a dedicated program.

Figure 3.1.1 Simplified schematics of an HPLC instrument. Method Development in Analytical HPLC. https://doi.org/10.1016/B978-0-443-29849-3.00008-7 Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

96

Method Development in Analytical HPLC

The solvent supply consists of several bottles (reservoirs) of solvents that are delivered to a high-pressure pump. The modern pumps include a degasser component which can also be a separate unit. The high-pressure pump sends the mobile phase of a selected composition through an injector. The injector loads the sample in the mobile phase as a liquid plug. This is done by connecting/disconnecting a loop filled with sample in the flow. The autosampler injectors have the capability of loading sequentially a number of samples, as controlled by the computer program. The mobile phase loaded with sample flows through the chromatographic column (some columns have a precolumn for protection). The column is filled with the stationary phase that has the capability to retain selectively the sample components. The column is usually kept in a column holder that has the capability of maintaining a specific constant temperature of the column. As the components of the sample are eluted at different times (retention times tR ) from the column, they are detected/measured based on a physical characteristic (e.g., ultraviolet (UV) absorption, refraction index, ions formed in a mass spectrometer, etc.) by a detector. The detector generates an electrical signal related to the component detected. It is possible to use only one detector in an HPLC system, but having more detectors connected sequentially is common for many instruments. The computer collects the data from the detector(s) and translates them into a chromatogram. The computer also controls the functions of the pump, injector, temperature of the column holder, and the parameters for the detector. For the development of an analytical HPLC method, the function of each component of the HPLC must be well understood. For this reason, some details on each of the components of an HPLC system are discussed in the following sections.

More complex HPLC instrumentation Besides basic setups in HPLC instrumentation, more complex HPLC setups can be designed for various purposes. These setups can have only some small changes compared to the basic type of instrument previously described. For example, an additional switch valve can be included between the autosampler and the column(s) that will allow to select between two columns placed in the column holder, without physically uninstalling a previously utilized column. Another simple device added after the column(s) and before the detector can be another switch valve that allows sending the flow from the column to the detector only at a specific time during the chromatographic run, and otherwise sending it to the waste. Such valve is common, for example, when the detector is a mass spectrometry (MS) instrument and a portion of the eluate known to not contain the analytes is diverted to the waste. A more complex setup is necessary, for example, for achieving bidimensional separations. Such systems require two different chromatographic columns each one with its own solvent supply and pumping system. In such systems, a portion from the eluate of the first column is selected and transferred to a second column for separation. For achieving a different type of separation compared to the one in the first column, a different type of column having different retention characteristics is selected, and the separation is indicated as orthogonal. When a single time segment from the eluate of the first column is selected to be subject of a second separation, the portion selected is indicated as a heart-cut [1].

Instrumentation in high-performance liquid chromatography

97

It is also possible to submit to the second column the whole eluate from the first column, separated in small sequential heart-cuts and achieve in this way a comprehensive bidimensional HPLC [2,3]. Comprehensive two-dimensional separations are similar to those used to take heart-cuts, but these are taken continuously, each for a short time interval. The schematics of the flow for a comprehensive HPLC with cuts of 100 mL is shown in Fig. 3.1.2. In the system indicated in Fig. 3.1.2, the flow from the first HPLC column after passing a detector (optional) enter a switching valve equipped with two loops (100 mL in the figure). Once a loop is filled with eluate, the switch valve changes the flow to fill the second loop and meanwhile the content of the first loop is introduced in the flow for the second column where it is again separated. If the flow in the first column is, for example, 0.1 mL/min, the loop is filled in 1 min. Every 1 min, the switch valve is changing the flow to the other loop. The separation in the second column should be fast, with a run time of 1 min. The flow from the second column is analyzed by a second detector. A computer program is used for assembling the whole chromatogram to generate unique peaks even when the loop switch occurs in the middle of one peak. The second column in comprehensive two-dimensional systems should be selected to perform an orthogonal separation similar to the case of using a unique heart-cut. Other systems are designed, for example, for sample enrichment in analytes. In this use, the flow with a diluted sample is sent to a column where the analytes are strongly retained for the specific solvent composition and, possibly, the matrix components are less retained. The flow to this column may carry a large volume of the diluted sample. The analytes are accumulated usually at the head of the column and the flow is sent to waste. After the analytes are accumulated, the flow in the column is reversed (with a switching valve), and the solvent is changed such that the analytes are eluted from the head of the first column and sent to a different column where the separation takes place. The flow from this second column with the separated analytes is sent to the detectors.

Figure 3.1.2 Diagram of a comprehensive two-dimensional HPLC system (the valve is switched at regular time intervals).

98

Method Development in Analytical HPLC

In some applications, a postcolumn addition of a reagent may be necessary for reacting with the analytes already separated (analytes derivatization) and produce a specific color, fluorescence, or chemiluminescence (CL) necessary for detection. In such cases, a reagent source and an additional pump are used to deliver the reagent that is mixed in a “mixing T” with the flow from the separating column [4]. In order to minimize the consumption of reagents, the continuous flow of the reagent(s) can be replaced by repeated injections of small volumes of reagent(s) [5]. In some applications, a coiled-shape tubing of a specific length and diameter is added after the mixing T and before the detector. In this coil, the eluting analytes spend enough time for the derivatization reaction to take place. The coil can be heated or for photochemical reactions to take place, the coil can be made from quartz and exposed to light [6]. The detector is placed after the tubing where the chemical reaction takes place. When using this type of setup, attention must be paid to minimizing the volumes that may contribute to peak broadening. In addition to the components previously described for an HPLC system, other devices may be used in an HPLC instrument. One example of such device is a flow splitter that allows only a portion of the mobile phase generated by the pumps to flow into the injector and chromatographic column. Splitters can be used, for example, when a capillary HPLC column is used. Another example of an additional device is that of a fraction collector. After some HPLC separations, some eluate fractions of interest may need to be collected as the effluent exits the detectors. This collection can be done using a fraction collector. Automated systems for fraction collection are available. Such systems collect the flow emerging from the last detector in specified vials either at a given time or upon receiving a signal from the detector (indicating an eluting peak).

Miniaturization in HPLC instrumentation Miniaturization in HPLC is not a widespread practice, but several advantages are associated with miniaturization. One of these advantages is increased portability, use of much smaller volumes of solvents, capability of analyzing smaller samples, lower instrument cost, etc. [7]. One key component of miniaturized HPLC systems is the use of capillary columns, or microfluidic chips. To those capillary columns or chips, the other equipment parts must be adjusted. For example, special pumps capable of very low flow rates must be used, one common type of pump used in miniaturized HPLC being the syringe pump. The injectors must be able to handle very small volumes of sample and the detectors must be able to perform measurements using, for example, flow cells with very small volume [8].

Key points • • •

A typical HPLC system is made of several modules including the solvent supply, highpressure pump, injector (autosampler), column in a column holder, one or more detectors. More complex setups are possible for achieving, for example, bidimensional separations. Miniaturization is a developing field in HPLC instrumentation.

Instrumentation in high-performance liquid chromatography

3.2

99

The separatory section of an HPLC instrument

The separatory section of an HPLC instrument consists basically from the solvent containers, high-pressure pump, injector, and the column in a column holder. Each of these components are further described in short.

Mobile phase supply system The mobile phase for HPLC is obtained by mixing specific solvents to a precise composition. In some separations, a pure solvent is used as mobile phase and no mixing is necessary. The solvents used for making the mobile phase must be miscible. The mobile phase composition can be kept constant during the separation (the time for a chromatographic separation is indicated as run time (trun ) and the separation is indicated as isocratic). However, the composition of the mobile phase can also be changed during the run time, following a specific program. This type of separation is indicated as with gradient. The gradient is used to modify the mobile phase composition such that a desired separation is obtained and/or to have a faster elution of some compounds. The decision about the mobile phase composition and its change if gradient is selected is an important part in developing an analytical method and the subject will be presented in detail in Chapter 9. The solvents necessary to make the mobile phase are typically kept in inert glass bottles. The solvents from the reservoirs must be free of particles, and they are either purchased as HPLC grade or filtered through 0.45 mm filters before use. The filter selected for the filtration must be inert to the solvent and the filtration should not be performed if not necessary since this operation may be a source of contaminations. Also, special attention must be given that no particles are formed when the mobile phase is made by mixing solvents. For example, some HPLC separations require a buffer or other additive in the mobile phase. The buffer is frequently made in only one solvent usually in water. When the buffer solution in water is mixed with an organic solvent, depending on the buffer concentration and the proportion of the organic solvent, precipitation may occur. This effect must be carefully avoided by keeping the concentration of buffers or other additives components low enough (usually less than 100 mmol) to remain soluble in the final mobile phase. Inert plastic tubing (preferably polytetrafluoroethylene although polypropylene is also used) connects the solvents with the pumping system of the HPLC. The solvents as commercially available may contain some dissolved gasses (O2, N2). These gases in the initial solvents may create problems by producing pressure fluctuations, by generating small bubbles because of the drop in pressure when the mobile phase exits the chromatographic column and enters the detectors, by influencing the injection volume (in case of very small injection volumes), or by interfering with the response of electrochemical detectors. For this reason, before reaching the pumping system, it is common that degassing of the solvents is performed. Degassing can be performed by two procedures, a preliminary degassing and an in-line degassing. The preliminary degassing is optional and in some instances is not even recommended. The preliminary degassing

100

Method Development in Analytical HPLC

can be done by sparging the solvents with an inert gas (N2, He) at a low flow rate, or by sonication. However, some solvents used in HPLC may need to contain volatile compounds such as NH3 (NH4OH), (C2H5)3N, or CF3COOH. Sparging of such solvents eliminates the necessary volatile component changing the intended mobile phase composition. In such cases, sparging is not recommended. Even with common solvent mixtures, some composition changes may occur during sparging. For example, degassing with He for 24 h for premixed mobile phase consisting in 50% water and 50% organic solvent (v/v) leads to a loss of 0.005% for methanol, 0.006% for acetonitrile, or 0.05% for tetrahydrofuran [9]. The second type of degassing in an HPLC system is performed in an in-line degasser device based on vacuum applied on one side of a semipermeable membrane, while the other side is in contact with mobile phase that releases the gaseous components. This can be a separate instrument or embedded in the pumping system. The in-line degassers can also pose some problems. They may be a source of contamination when changing from one solvent to another. Also, it should be noted that large bubbles coming from the solvent reservoir cannot be eliminated by the degasser apparatus. A special type of solvent delivery system can be used in ion chromatography. This system is known as eluent generator. When an eluent generator is used, the highpressure pump(s) deliver only water, while the reagents necessary for the elution are generated electrochemically from special cartridges. For this purpose, DC current is applied to a special cartridge to produce either KOH, NaOH, LiOH, or K2CO3, KHCO3 for anion exchange eluents, or methanesulfonic acid (MSA) for cation exchange eluent. By controlling the applied current, for example, the hydroxides for anion exchange are generated at the desired concentration or at a specified gradient (see, e.g., Ref. [10]). These eluent generators are typically offered as a whole assembly, and are installed before the injector in the liquid chromatography (LC) system.

Pumping systems The role of pumping system in HPLC is to deliver a flow of mobile phase through the injector, chromatographic column, and detector (or detectors if more than one is utilized). The mobile phase should have a specified composition obtained by mixing solvents in a desired proportion (this is not necessary if the mobile phase consists of only one solvent). A detailed presentation regarding the mobile phase in HPLC is given in Chapter 8. Also, the pumps must be able to generate the mobile phase at a high pressure which is needed mainly to overcome the resistance to flow of the chromatographic column. This flow is characterized by the volumetric flow rate U. In older conventional HPLC systems, the pumps are usually capable of delivering U between 0.1 and 5 mL/ min (or 10 mL/min for some instruments) and generate up to 6000 psi (about 400 bar, one psi ¼ 0.06894757 bar). Newer instruments can generate up to 17,500 psi (about 1200 bar) at 1.0 mL/min (lower maximum pressures are necessary at higher flow rates) [11]. Such instruments are sometimes specifically indicated as ultra-performance liquid chromatography (UPLC), but in this book, the term HPLC will be used with a general meaning including standard HPLC or instruments capable of delivering higher pressures (UPLC systems). Most high-pressure pumps used in analytical

Instrumentation in high-performance liquid chromatography

101

HPLC are dual reciprocating pumps. These pumps have two pistons working in tandem for delivering the fluid. In such pumps when one piston is filling its cylinder with fluid, the other piston is emptying its cylinder and the fluid is continuously delivered. The flow from the dual pumps (volumetric flow rate U) must be constant, without fluctuations or only with very small ones. This requirement is necessary mainly for the detectors, where the signal may fluctuate when the flow rate varies. To achieve a constant flow, the modern pumps have electronic control of the piston movement and are using pressure pulsation dampers. Various models of dampers are available, and most of them have a volume of around 500 mL, in order to ensure a small delay volume in the delivered fluid. Modern systems are able to deliver flow with the precision for U of about 0.07% relative standard deviation (RSD%), and a flow accuracy of less than 1% from the nominal value. The pulsation with dampening can be reduced to less than 2% variation in pressure. Other types of pumps are also known for HPLC systems such as syringe pumps or hybrid mixing pumps, but they are not common. A dual piston pump can handle only one solvent which is acceptable for isocratic separations that use a pure or a premixed solvent. For being able to handle more solvents (typically two, or four), instruments that can handle more than one solvent were developed. Two types of such instruments are common, indicated as instruments with low pressure mixing, and instrument with high-pressure mixing. In low pressure mixing, usually four solvents can be handled (if necessary) at the desired composition by using a low-pressure pump with several proportioning inlet valves controlled by a computer. The valves open repeatedly for a short period of time (typically less than one second), the duration of time the valve is opened being proportional with the desired mobile phase composition. The mixed solvents are delivered further to one (dual piston) highpressure pump. Low pressure mixing has the advantage of using a single highpressure pump (that is rather expensive), and has more flexibility in choosing more solvents (e.g., four). The changes in the mobile phase composition when using a low-pressure mixing system are taking place more gradually than for the highpressure mixing. In high-pressure mixing, two high-pressure (dual piston) pumps are used, and the ratio of solvents in the mobile phase is controlled by the flow rate of the high-pressure pumps. The mixing of two solvents A and B can be achieved either in a specially designed mixer (typically with volume of less than 500 mL) or in a mixing T (with virtually zero volume). High-pressure mixing provides a more precise control of the composition of the mobile phase (with a typical composition accuracy having less than 0.15% RSD% at 1 mL/min flow). Having two dual piston pumps, the highpressure pumps can handle two solvents at a time, but they also have the capability of switching between two solvent pairs (A1, B1, and A2, B2). The pumping equipment of modern HPLC frequently has the capability to detect any leakage, communicating to the controlling computer of the HPLC to stop the operation when a leak is detected. In both types of pumping systems, there is a specific volume from the point at which the mobile phase solvents are mixed until they reach the head of the chromatographic column. This volume is known as dwell volume VD . The dwell volume VD of the system also creates a dwell time tD related to the dwell volume by the expression: tD ¼ VD =U

(3.2.1)

102

Method Development in Analytical HPLC

The dwell volume in low pressure mixing systems is about 1e1.5 mL and in high pressure mixing systems is 0.07e0.4 mL, depending on the instrument model. Special instruments, such as those used in microscale HPLC may have smaller dwell volumes (less than 30 mL). (Dwell volume VD should not be confused with dead volume V0 which is the volume of the column packed with stationary phase). Also, a small volume exists between the injector and the head of chromatographic column. This volume can be considered part of both dwell volume VD and part of the dead volume V0 , but being very small (5e10 mL) is usually neglected. The pumps delivering the mobile phase with a specific composition are controlled in the modern HPLC instruments by a computer program and using a time table. The time table indicates ranges when the mobile phase composition is constant or is changing following a specific gradient. The application of the time table starts when the sample is injected. The total run time of the chromatogram starting with the moment of injection until the end of the chromatographic run is sometimes referred to as total cycle time. After the cycle time is completed, the HPLC chromatograph is made ready for the next injection. The modification in concentration between two changing points of the gradient can be linear. For a gradient starting at time t1 with the concentration C1 of component A and ending at time t2 with the concentration C2 , at an intermediate point at time t, the concentration of component A in a binary system can be obtained using the expression: CðtÞ ¼ C1 þ

ðt  t1 Þ ðC2  C1 Þ ðt2  t1 Þ

(3.2.2)

The change in the concentration of the two components in a period of time from t1 to t2 is indicated as gradient slope. For a linear change in concentration, the gradient slope can be defined by the expression: D¼

ðC2  C1 Þ ðt2  t1 Þ

(3.2.3)

Some HPLC pumping systems allow both a linear change in the gradient and a nonlinear modification of the concentration. This change can be achieved using a variation in concentration as a function of time given by the expression:  CðtÞ ¼ C1 þ

t  t1 t2  t1

n ðC2  C1 Þ

(3.2.4)

where n is larger than 1 for curves of increase that “holds water” and is between 0 and 1 for “does not hold water” type of curve. A different type of nonlinear curve can be obtained using a variation in concentration with a function given by the expression: 

t2  t CðtÞ ¼ C2  t2  t1

n ðC2  C1 Þ

(3.2.5)

Instrumentation in high-performance liquid chromatography

103

When n is larger than 1, the curve of increase “does not hold water,” and when n is between 0 and 1, the curve of increase “holds water.” In some chromatographic systems, there are used both types of gradient variation and a “type of curve” can be selected from the software controlling the HPLC instrument. As an example, for the instruments manufactured by Waters (Waters Corp., Milford, USA), Eq. 3.2.5 is used for curves 2 (n ¼ 8), 3 (n ¼ 5), 4 (n ¼ 3), and 5 (n ¼ 2) which are convex, and Eq. 3.2.4 for curves 6 (n ¼ 1), 7 (n ¼ 2), 8 (n ¼ 3), 9 (n ¼ 5), and 10 (n ¼ 8) which are concave. These curves are shown in Fig. 3.2.1 for t1 ¼ 0, C1 ¼ 0 and t2 ¼ 5, C2 ¼ 100. Because of the dwell time present in HPLC systems, there is a delay between the change of composition at the point of solvent mixing (set in the time table) and the change in composition at the head of chromatographic column. Therefore, when attempting to modify the retention time of a peak by using a “stronger” solvent, this change should be done in the gradient time table ahead of the peak retention time.

Tubing and connectors (fittings) The flow of the mobile phase from the high-pressure pump to the other modules of an HPLC takes place through special tubing typically made from stainless steel (316 stainless steel), or from polyetheretherketone (PEEK), silica lined PEEK, or even titanium. For use at higher pressures than about 5000 psi, stainless steel is usually required (when used in UPLC mode, PEEK type tubing is not recommended). Tubes of several internal diameters (i.d.) are available, such as 0.12 mm i.d. (0.005 in), 0.17 mm i.d. (0.007 in), 0.25 mm i.d. (0.010 in), 0.50 mm i.d (0.020 in), etc. (for both stainless steel tubing and PEEK tubing a color code is available to designate the i.d.). The choice of 100 90

1 2

Composition %

80

3

70

4 5

60

6

50

7

40 30 20

8

9

10

10

11

0 0

1

2

3

4

5

Time (min) Figure 3.2.1 Gradient variation curves using both Eqs. 3.2.4 and 3.2.5 applied in Waters instruments.

104

Method Development in Analytical HPLC

the tubing in particular after the injector plays a role in the shape of the sample liquid plug and of the shape of the chromatographic peaks. Narrower tubes producing very little diffusion of the sample plug are preferred and tubing with 0.12 mm i.d. (0.005 in) is in general recommended to connect the injector with the chromatographic column. This tubing has a volume of about 0.13 mL/cm such that a sample of 5 mL will spread over about 38 cm, diminishing the effects of sample plug shape modifications. However, due to the friction with the tubing walls of a viscous fluid, downstream of injector (in a laminar flow), the shape of the sample plug is changing and generates a parabolic profile. This process, as well as the diffusion and other convection effects, contribute to deviations of the chromatographic peak from the ideal Gaussian shape. When using more than one detector, they are usually connected in series and the connection is also obtained with similar tubing. In such case, special attention must be given to the backpressure generated by the connecting tubes. Because the tubes with very small diameter add backpressure, and some detectors, such as fluorescence detector and in particular the refractive index detectors contain pressure sensitive flow cells, the backpressure acceptable for those calls should not be exceeded. This is usually achieved by using short tubing length, special order of connecting the detectors (e.g., UV-Vis then fluorescence and at the end refractive index), or even using tubes with slightly larger diameter after a detector sensitive to backpressure. The tubes in HPLC and the HPLC modules are connected with special fittings (made of nut, ferule, and port). The connectors can be made from stainless steel and also from PEEK. The connections must avoid forming void spaces in the mobile phase flow. These void spaces produce peak broadening due to mixing and the turbulent flow they generate.

Injectors and autosamplers The role of the injector is to add in the mobile phase a small, precisely measured volume Vinj of a solution of the sample. The injected sample having an analyte concentration x, (or molar concentration ½X) results that the injector places in the mobile phase the quantity qinj in the mobile phase given by the expression: qinj ¼ Vinj $x

(3.2.6)

The injector can be designed for single injections, but systems with automation capability are common and are indicated as autosamplers (or autoinjectors). The injection must be done reproducibly and accurately. Reproducibility of injection is of particular importance, and modern injectors typically show less than 0.5% RSD% in the injected volume. The accuracy errors in injection volume are important mainly when comparing different instruments since for the same instrument the use of standards for quantitation may compensate small variations from a nominal volume. However, for a specific method, it is recommended to keep the same injection volume when injecting different samples in order to avoid accuracy problems. One common type of injector uses a loop of a precise volume which is filled first with the sample and then connected to the flow circuit using a switching valve as schematically indicated in

Instrumentation in high-performance liquid chromatography

105

Fig. 3.1.1. In order to be able to inject a desired volume of sample and not a fixed volume equal with that of the sample loop, the loop is initially filled with the mobile phase, and then the sample is introduced in the loop, occupying only a portion of its volume. The range of volumes that can be injected depends on the injector model, and capability to inject volumes between 0.5 mL up to 100 mL sample solution is common (most frequently volumes between 1 mL and 20 mL are used for HPLC injections). Special systems have the capability to inject larger volume (up to 1000 mL in special systems) or very low volume (e.g., 0.1 mL). In autosamplers, the samples are placed in a tray containing capped vials with septum (e.g., 2 mL vials) each one containing a solution of the sample. Computercontrolled autosamplers have the capability to inject any desired sample from the tray, as many samples as necessary. For injection, a needle of the injector penetrates the vial septum, draws the desired volume of sample, and loads it in the injector loop. The sample solution is subsequently placed in the mobile phase flow as a solution plug. Modern autosamplers also have the capability to wash the autosampler needle to avoid carryover because of needle contamination from a previous sample. They are also able to keep the tray with samples at a desired temperature (e.g., cooling for improving sample stability in time). Some autosamplers have the capability of mixing the sample with specific reagents from different vials, in case derivatization is necessary before the separation and detection. Unique injection systems are also utilized in some instances for HPLC. For example, in order to use HPLC for the analysis of analytes extracted by solid-phase microextraction (SPME), a special type of injector that contains a desorption chamber instead of a sample loop is utilized. This camber is filled with mobile phase, the analytes adsorbed in the SPME fiber are dissolved in it, and the chamber content is further included in the flow circuit of the HPLC (e.g., Ref. [12]). In-tube SPME (IT-SPME) particularly designed for HPLC analysis uses a special short capillary column for analytes retention, and this system also uses a special type of injection [13,14]. In most HPLC applications, one injection is done for each chromatographic run. However, multiple injections in a single run (MISER) are possible for specific analyses, such that a larger number of samples can be analyzed within a specific interval of time, and with a lower solvent usage [15]. MISER technique can be applied only to special separation where peak interference from one run to another can be avoided.

Column holders The column in modern HPLC systems is usually placed in a column holder that can usually accommodate two columns. The column holders have the role to keep the chromatographic column at a specific temperature, usually at a value from 5 to 65 C. Modern HPLC instruments utilize thermoelectric Peltier technology to cool or heat the chromatographic column. The column holder usually has the additional capability to adjust the mobile phase temperature close to that of the chromatographic column. Most column holders are also equipped with a leak detector. For high temperature HPLC, special column holders must be used which are different from the common column holders. In these special column holders, the temperature can be

106

Method Development in Analytical HPLC

maintained at a specified value up to 200 C [16], the maximum used temperature being also dictated by the boiling point of the mobile phase. For the work at high temperatures, a potential problem is the difference between the temperature of the incoming mobile phase and that of the column, which should not differ by more than 6 C [17]. For high temperature HPLC, postcolumn cooling is also necessary to avoid changes of detector sensitivity at high temperature or due to temperature variations. New technologies offer a full capability of temperature programming and fast cooldown of the mobile phase after a temperature program [18]. Precautions should be given in high temperature HPLC to the stability of analytes as well as of stationary phase used for separation.

Chromatographic columns The chromatographic column is a key component of the HPLC system since the separation process takes place in the stationary phase from the column. External body of most chromatographic columns is made from a tube (cylinder) made from metal (stainless steel) or plastic (e.g., polyetheretherketone, PEEK), which is filled with the stationary phase. Other separation devices, not as common as columns, are also available such as cartridges and microfluidic chips that contain the stationary phase [19,20]. The physical dimensions of common analytical chromatographic columns vary, and values for length (internal) L can be between 30 and 250 mm (common length 50, 100, 150, 250 mm), and internal diameters d for usual columns can be between 1 and 10 mm (e.g., 1.7, 2.1, 3.0, or 4.6 mm). Other dimensions are possible, particularly when the column is designed for special tasks. Based on the internal diameter of the analytical column, they are sometimes classified as follows: (1) standard (3.0e4.6 mm i.d.), (2) minibore (2.0e3.0 mm i.d.), (3) microbore (0.5e2.0 mm i.d.), (4) capillary (0.2e0.5 mm i.d.), and (5) nanoscale (0.05e0.2 mm i.d.). Larger columns are used for semipreparative and preparative purposes. Capillary columns represent a special type as they are made from a silica capillary that contains the stationary phase either as particles or coated on the tubing wall [21]. The empty volume of the column can be calculated as the volume of a cylinder V ¼ ðp =4Þd 2 L, and for analytical columns, V ranges between 0.02 and 20 mL. The stationary phase inside the column is usually made from porous small particles, superficially porous particles (core-shell particles), or even pellicular porous (that are not very common). Porous particles are still the most common type of stationary phase used in HPLC, although the core-shell particles are becoming popular. Porous and core-shell types are made from particles usually of 1.7e5 m m diameter, and for the core-shell, the porous outer shell has only 0.3e0.5 mm thickness. Very frequently for the particles the porous material is used only as a support and on this support is placed the active phase which is bonded, grafted, or coated [22]. Some porous materials such as silica can also be used directly as active phase, due to a layer of water adsorbed on silica surface. Also, for some columns, the stationary phase is in the form of a porous monolith. The chemical nature of the stationary phase is specifically selected based on the type of chromatography that is utilized for the separation

Instrumentation in high-performance liquid chromatography

107

(reversed phase, HILIC, ion exchange, size exclusion, etc.). A large assortment of types of stationary phases (column packings) is available. Besides the chemical nature of the active phase of the chromatographic column, several physical characteristics of the phase are also important. For the fully porous particles, surface area, pore dimensions, the particle size, shape, size distribution, etc. are important for the separation. Besides these properties, for superficially porous particles that have a solid core and only a porous layer on it, the thickness of this layer is important. Since reversephase (RP-HPLC) is the most utilized technique, the largest variety of columns is of RP type. These columns have a hydrophobic active phase, for example, with octadecyl groups (C18), or with octyl groups (C8) bonded on silica. For other types of chromatography, the stationary phase may be made in various forms. For example, ion exchange HPLC can use particles from a substrate inert material that are covered with the active phase, but the stationary phase can be simply an ion exchange resin. Size exclusion chromatography typically uses perfusion particles made from silica or special types of polymers. These particles contain very large pores (400e800 nm) connected with a network of smaller pores (30e100 nm). The selection of the chromatographic column for an analytical method specific for a requested analysis is one of the main subjects of present book, and it will be presented in Chapters 7 and 9. The protection of the stationary phase is commonly done by using in-line filters consisting of small pore frits (e.g., with 0.45 m m pores) at the ends of the column, as well as by the utilization of guard columns (cartridges) placed before the analytical column. The guard columns are selected to match the stationary phase of the analytical column (same active material), but their length is much shorter (e.g., a few mm) and in some cases the stationary phase in the guard column has larger particle size. The guard columns are designed to not affect the separation.

Key points • • • • • • • •

Mobile phase is made in the HPLC pump based on a specific protocol controlled in a computer. The solvents used for making the mobile phase must be miscible and free of particles. Particles in the solvents can be eliminated by filtration, but filtration should be performed only if necessary. When a solvent contains buffers or other additives, they must remain soluble upon mixing with another solvent when making the mobile phase. Two types of HPLC pumps are common, with low-pressure mixing and high-pressure mixing. Some differences between the two types can be seen when transferring an analytical method from one type of instrument to another. Liquid samples are placed as a plug in the flow of mobile phase by an injector that can be designed for single injections, but systems with automation capability are common. Chromatographic columns are filed with a stationary phase that is specific for the type of a chromatographic separation. Besides the chemical nature of the stationary phase in a chromatographic column, several physical properties of the stationary phase and column dimensions also influence a separation.

108

3.3

Method Development in Analytical HPLC

Detectors in HPLC other than mass spectrometers

A number of detectors are very common for HPLC. These include the UV-Vis, refractive index, fluorescence, and mass spectrometric detectors. Other detectors are frequently used only for special applications, such as electrochemical detectors that are common in ion chromatography, and evaporative light scattering detectors used for nonvolatile compounds difficult to analyze with other detectors. Much less common are specialized detectors such as those using infrared (IR) or nuclear magnetic resonance (NMR) spectroscopy. These types of detectors are not commonly available commercially and were developed in research laboratories. Detectors may be of universal type responding equally to any type of component in the column effluent except the mobile phase, or may provide various levels of selectivity to the detection. Selective detectors respond to a specific characteristic of the components in the column effluent, such as a specific absorbance wavelength. Specific detectors, also utilized in practice, respond to a single sample component or to a very limited number of components having similar chemical characteristics.

UV-Vis spectrometric detectors The UV-Vis detectors are used to measure the absorption of light by the analytes in the HPLC eluent. These detectors are basically UV-Vis spectrometers equipped with a flow through cell for measuring the UV-Vis absorption of eluent. These detectors are largely utilized in routine analyses in HPLC and they can be indicated as selective. Two related quantities, transmittance T and absorbance A, are measurable for the light passing through the solution to be analyzed. Transmittance is defined as follows: T ¼ I1 =I0

(3.3.1)

In Eq. 3.3.1, I0 is the intensity of the radiant energy incoming to the sample and I1 is the intensity of the emerging light (T can also be expressed  as percent). As expected, T is a function of the wavelength l (or of frequency v ¼ clight l, where clight is the speed of light) of the radiation that is absorbed. Absorbance Al is defined by the logarithm in base 10 of the inverse of transmittance as follows:     1 I0 Al ¼ log ¼ log T I1

(3.3.2)

Absorbance is related to the molar concentration ½X of the absorbing species X by LamberteBeer law (see also Eq. 1.4.1): Al ¼ εl ½XL

(3.3.3)

where εl is the molar absorption (absorbance) coefficient at the specific wavelength l, (usually taken at the maximum of the absorption band) and L is the path length of the

Instrumentation in high-performance liquid chromatography

109

light through the sample. For quantitation purposes, the absorbance is commonly used, because it is proportional with the concentration. In HPLC, the instantaneous concentration ½X is measured in the flow cell, and the signal Al versus ½X is recorded in the form of a chromatographic peak. Quantitation using UV detection is based on the measurement of the area of this peak and calibrations with solutions of known concentration, as further discussed in Chapter 4. UV-Vis detectors can be classified as fixed wavelength, variable wavelength, and photodiode array detectors (DADs) [23]. There are many variations of the construction of those detectors, depending on the manufacturer and the detector model. Simpler detectors are able to measure Al at a fixed wavelength, for example, generated by a Hg vapor lamp at 365.4 nm and using special filters to select this radiation. However, this type of detector is not common anymore, and detectors with improved capabilities are now common. Variable wavelength detectors have the capability to set a desired wavelength for measuring absorption. The range of values that can be selected for the wavelength varies from model to model. For some detectors, the range is from 190 to 700 nm, for other detectors, the range is extended from 190 to 950 nm, etc. However, a common range of practical utility in UV spectrophotometric measurements starts at about 205 nm. At lower values than this wavelength, many solvents commonly utilized in the mobile phase start absorbing the UV light and impede the measurements. The detectors usually have the capability of a reference beam necessary for the elimination of fluctuations in the light intensity that are not caused by the sample. DADs are capable to record a whole range of the UV-Vis spectrum of the analyte passing the flow cell of the instrument. However, the instruments can also be used for recording only one wavelength, as it is done with a variable wavelength detector and use this type of measurement in quantitative studies. The capability of DAD detectors to record a whole spectral range (e.g., from 190 to 950 nm) can be useful for peak purity evaluation, as well as for determining the maximum wavelength of absorbance for a compound such that the detected maximum can be subsequently utilized for quantitative measurements. The construction of an instrument with DAD detection is similar to the instruments using a specified wavelength detection, except that the light in the DAD is not received by a single photocell, but by an array of photodiodes containing, for example, 1024 elements that can provide a 1.0 nm resolution. A schematic construction of a DAD instrument in shown in Fig. 3.3.1. Flow from the column

Focusing lens

Slit

Concave diffraction grating

c

Vis (tungsten) lamp

UV (deuterium) lamp

Flow cell Shutter n

1

Photodiode array

Figure 3.3.1 Schematic illustration of the diode-array detector used in HPLC.

Electric signal

110

Method Development in Analytical HPLC

Numerous other details differentiate the instruments from model to model. These include, for example, the volume of the flow cell that can be 1.0 mL or 2.0 mL for a microcell, 5 mL for a semimicro, and 14 mL for a standard cell, the path length of the cell that can be 5e10 mm depending on the cell, some cells having a conical shape, the pressure at which the flow cell can stand (low pressure cell 20 bar, typical pressure cells 120 bar, or high-pressure microcells 400 bar). Other details include the presence of only a deuterium lamp as a light source or the addition of a tungsten lamp to extend the radiation range to higher values, the frequency of acquiring signal at 80, 120, or 240 Hz, the use of various filters, etc. The dispersion of light in the modern instruments is usually produced by a diffraction grating, but prisms monochromators were also known to be used for variable wavelength instruments. The quality of the electronics and the sensitivity of UV-Vis detectors also vary considerably from model to model. Depending on the nature of the analyzed material, the detection limit of the UV-Vis can be 0.2e1.0 ng (w0.1 mg/mL in the injected solution in an HPLC), with a linear range of five orders of magnitude (this can also be different from model to model of detector). With an appropriate solvent that does not absorb in the range of UV-Vis measurement, the use of elution gradient can be applied with no problems for the separation.

Fluorescence detectors The use of fluorescence detection (FLD) in HPLC is common when the separated analytes have fluorescence property. In fluorescence, the samples are irradiated with light of a specific wavelength lex and they reemit radiation with a different (typically lower) wavelength lem . The theory of fluorescence emission shows that the intensity of fluorescence Fint at the emission wavelength lem can be expressed as a function of the intensity I0 of the excitation radiant energy with wavelength lex incoming into the sample, by the expression:    Fint;lem ¼ I0;lex 1  exp  εlem ½XL F

(3.3.4)

In Eq. 3.3.4, F is the (quantum) fluorescence yield of the process, a parameter characterizing each compound, and the other parameters beingthe same asdefined for UVVis. For low concentrations, the approximation 1  exp  εlem ½XL z εlem ½XL can be utilized and the intensity of fluorescence Fint;lem is related to the concentration ½X by the relation: Fint;lem ¼ I0;lex εlem ½XLF

(3.3.5)

In reality, only a part of emitted fluorescence is measured in the analytical instru0 ment and the intensity of this measured fluorescence F int;lem is given by the expression: 0

F int;lem ¼ a I0;lex ½X

(3.3.6)

Instrumentation in high-performance liquid chromatography

111

where a is a constant coefficient that incorporates all the constants and accounts for the losses due to partial fluorescence measurement. Measurement of fluorescence intensity (usually at the maximum of the emission band) is the base of quantitation of the fluorescent species. Because the constant a also depends on the instrument used for measurement and its settings, the quantitation using fluorescence measurements is performed using peak areas in the chromatogram and common type calibrations with standards of known concentration. The fluorescence intensity is measured using sensitive light detectors (FLD) that generate an electrical signal of intensity depending on a calibration constant for the instrument. The output is given in luminescence (or light) units (LU) that are arbitrary units proportional with the fluorescence intensity, but also specific for the measuring instrument [24]. Similarly to UV-Vis detection in HPLC, the fluorescence is measured in a flow-through cell that is connected to the flow of the eluent incoming from the chromatographic column. The instantaneous fluorescence intensity in the flow cell is measured and generates the chromatographic peak. A typical configuration of the FLD in HPLC is illustrated in Fig. 3.3.2. The systems are equipped with two diffraction gratings as monochromators, one for the selection of the excitation wavelength (lex ) and another for the selection of emission wavelength (lem ). From the analytical column, the separated species having fluorescent property will emit fluorescence light in all directions from which for the measurement is kept only for the emitted radiation at 90 degrees angle in order to minimize scattering light effects [25]. Modern fluorescence detectors may have the capability of recording a tridimensional emission spectrum of the analyte by stopping the mobile phase flow at a chosen retention time (selected for the analyte) and performing a scan of the entire UV range used for excitation and for the entire emission band. In this way, the tridimensional fluorescence spectrum is displayed as a dependence of fluorescence intensity on emission wavelength and excitation wavelength. It is common in modern fluorescence HPLC detectors that the excitation beam is generated by a high-power lamp that flashes at a specific number of times per second (e.g., 296 times in an Agilent 1200 Ser. detector), such that the signal is modulated in time. Also, the systems typically have a reference detector that measures the excitation light and corrects the flash lamp fluctuations [26]. Flow from HPLC column

Light source ex

Flow cell c

Slit

Diffraction grating for ex

Slit Diffraction grating for em

em

Photomultiplier or photodiode array

Figure 3.3.2 Schematic illustration of a fluorescence detector used in HPLC.

112

Method Development in Analytical HPLC

The detection in fluorescence methods encounters several difficulties because of nonlinearity of fluorescence caused by self-absorption effects, difficulty in discriminating between overlapping broad spectra of interfering molecules, quenching produced by oxygen dissolved with the solvent, etc. Because the intensity of fluorescence increases linearly with the intensity of the initial radiation, laserinduced fluorescence (LIF) detection is a successful technique applied in HPLC. For HPLC, lasers are a convenient excitation source because they have intense light focused into a small volume, they are highly monochromatic, and the associated Raman light has a well-defined wavelength that can be avoided with the monochromator used for observing fluorescence. However, LIF is still affected by background interference commonly arising from the Raman effect in the blank (molecular scattering) or from low level of solid impurities in the solvent producing Rayleigh light scattering. Detectors with constant excitation wavelength and variable absorption or with variable wavelength excitation and absorption are commercially available. Depending on the nature of the analyzed material, the detection limit using FLD can be as low as 10e2e10e3 ng/mL, with a linear range of four orders of magnitude. However, the minimum measurable amount of an analyte is strongly dependent on its nature [27]. When appropriately selected, the use of elution gradient can be applied for separation without interfering with the fluorescence. Different factors related to the mobile phase may influence fluorescence such as pH, solvent nature, temperature (as much as 2%), presence of impurities, as well as the flow rate [28].

Chemiluminescence detectors CL is basically a fluorescence in which the fluorescence of the molecules is generated following a chemical reaction. The CL intensity follows the same law as fluorescence with the difference that quantum yield F from fluorescence must be replaced with a different quantum yield FCL , which is defined as the proportion of analyte molecules that emit a photon during CL. The FCL increases with the efficiency of the chemical reaction producing the excitation (such as an oxidation process). Higher energies required by molecules to achieve the excited state diminish FCL . The time frame of the light emission is dependent on the CL reaction. Certain chemiluminescent systems, although with very good FCL , may emit the light for a period of 40e50 min and in this way the emitting instantaneous intensity is lower. Much shorter times can be achieved using a catalyst. Because no excitation light is needed in CL, the interfering light from Raman effect or light scattering by trace particles is nonexistent. In addition, the development of detectors virtually able to detect single photons makes the technique highly sensitive. Chemiluminescent reactions usually require postcolumn derivatization (see Section 3.1) which may produce some peak broadening and do not allow the use of another detector after using CL [29,30].

Refractive index detectors Refractive index detectors (RI or RIDs) are also commonly used in HPLC. These detectors are based on the change in the refractive index n of the mobile phase when an

Instrumentation in high-performance liquid chromatography

113

analyte is present in it. This modification depends on the analyte concentration and can be used for the quantitation of a variety of analytes. For this reason, the RID is a universal detector. The detector is based on the deviation of the direction of a light beam when passing under an angle from one medium to a medium with a different refractive index. This deviation depends on the difference in the refractive index between the two media. The flow cell of the RID has two compartments, one reference compartment filled with pure mobile phase and the other compartment having the flowing mobile phase resulting from the chromatographic column. When a separated sample component flows through the second compartment, its refractive index is different from that of pure mobile phase and the beam passing through the cell is deviated. The change in the location of the beam on the (photoelectric) detector is made to modify the detector output proportional to the deviation of the beam from the reference position. This output is electronically processed to provide a signal proportional to the concentration ½X of the solute in the sample cell. This principle is suggested in Fig. 3.3.3. The refractive index depends on the wavelength of the incident beam, and the most accurate RI measurements are done with monochromatic light (usually 589 nm, the sodium D line). With optical corrections, white light can still be used for the measurements. The refractive index is also different from compound to compound, but this difference cannot be used for compound identification. However, the calibration of concentration dependence on a measured refractive index cannot be used from one compound to another. RID can be utilized without the need for chromophore groups, fluorescence bearing groups, or other specific properties in the molecule of the analyte. In many cases the sensitivity of RI detection is, however, not as good as that of other types of detection. Also, it is not possible to use elution with gradient for the mobile phase, since this is associated with large variations in the refractive index of the mobile phase. Refractive index is sensitive to temperature changes, and a constant temperature must be maintained during measurements. The response of the detector is given in arbitrary RI units, depending on the detector settings, but proportional with the concentration of the analyte.

Light beam

Pure mobile phase

Reference cell with pure mobile phase

Light beam

Reference cell with pure mobile phase

Mobile phase with analyte

Light detector

Light detector

Figure 3.3.3 Schematic of refractive index measurement of the mobile phase in a refractive index detector.

114

Method Development in Analytical HPLC

Electrochemical detectors Electrochemical detectors are common in particular for the detection of ionic species, but other molecules can also be detected using electrochemical detectors [31]. These detectors are of various types since a number of electrochemical analytical procedures are known (see Section 1.4). Among the techniques more commonly applied in HPLC are the conductometric, potentiometric, and amperometric procedures [32e34]. These techniques may have high sensitivity, and the price of the detectors is relatively low. In ion chromatography, conductometric measurements are the most common. Basically, the conductivity detectors are used for the measurement of concentrations of electrolytes in aqueous solutions based on the following expression: ½X ¼ Constcell

1 1 LX Res

(3.3.7)

where Constcell is a constant depending of the measuring cell, Res is the electrical resistance measured with the instrument and utilized for the calculation of concentration. The measuring cell has two inert electrodes set a constant voltage V. Based on Ohm’s law Res ¼ I=V, the intensity I measurement allows the calculation of Res. Parameter LX is the equivalent conductivity for the ionic species X. Although LX can be taken for practical purposes as constant, it varies to a certain extent with the concentration following Kohlrausch’s law: LX ¼ L0X  Q

pffiffiffiffiffiffi ½X

(3.3.8)

where Q is a constant and L0X is the limiting molar conductivity specific for each ion. Molar conductivity is temperature dependent. Values for limiting molar conductivities were tabulated for certain electrolytes [35]. In ion chromatography, the mobile phase frequently contains acids, bases, or salts that may interfere with the conductometric detection. For this reason, before reaching the detector in IC, it is common to use a chemical suppressor, which can reduce the conductivity caused by the eluent components and virtually eliminating the ions belonging to the mobile phase, and increasing the conductivity due to the analytes. Various models of chemical suppressors are known some based on semipermeable membrane technology [36] and others using an ion exchange cartridge. Both systems have the same role. In the case of anion analysis, the eluent is frequently a solution of NaOH or KOH. The suppressor designed to eliminate the NaOH or KOH from the mobile phase in the form of a resin or a semipermeable membrane provides immobilized (or not permeable) eSO3H groups to the passing solution through the suppressor. In this case, the reaction with, for example, NaOH from the eluent through the suppressor is described by the following scheme: NaOH þ R  SO3 H / R  SO3 Na þ H2 O

(3.3.9)

Instrumentation in high-performance liquid chromatography

115

At the same time, an analyte in the form of a Na salt (e.g., a chloride) passing through the suppressor undergoes the following change: NaX þ R  SO3 H / R  SO3 Na þ HX

(3.3.10)

As a result of suppression, the conductivity caused by the NaOH in the mobile phase is eliminated being changed into H2O, while the conductivity caused by the analyte NaX is not modified being changed into HX which is strongly dissociated. Aqueous buffers containing NaHCO3 and Na2CO3 are also frequently used as eluents in IC. The conductivity caused by such buffers is eliminated using a chemical suppressor by transforming the alkaline carbonates into H2CO3 that decomposes into H2O and CO2 and do not contribute to conductivity. Other anions (e.g., F, Cl, SO2 4 ) generate strong acids that are easily detected based on conductivity. Various other suppression techniques are used in practice [37]. The suppressors must have a small dead volume in order to have a small contribution to the peak broadening in the chromatograms. In the case of a cation analysis, suppression of conductivity from the mobile phase can be achieved with a resin in OH form. For example, H2SO4 or MSA, which are commonly used as additives to the mobile phase, are retained by the anion exchange resin or semipermeable membrane (in OHe form) with the formation of H2O in the solution. The reactions taking place in such cases are the following: H2 SO4 þ R  OH / R  O  SO3 H þ H2 O

(3.3.11)

At the same time, an analyte in the form of a Na salt (e.g., a chloride) passing through the suppressor undergoes the following change: NaX þ R  OH / R  X þ NaOH

(3.3.12)

Semipermeable membrane for cation exchange separations may use a flow of tetrabutyl-ammonium hydroxide as a reagent on the other side of the membrane. The reagent interacts with the mobile phase acids. The OHe ions of the reagent in the semie permeable membrane replace the SO2 4 or CH3SO3 ions from the mobile phase. In this þ way, water is formed with the H of H2SO4 or CH3SO3H. Other ions (e.g., Naþ, Kþ, etc.) generate hydroxides that have high conductivity (NaOH or KOH are highly dissociated). Conductometric detection does not have selectivity, and any ion present in the measuring flow-cell of the detector generates a signal. For this reason, the separation by the HPLC should assure the elimination of interferences [38]. Semipermeable membranes are more common in the chemical suppressors, but the use in the flow of an ion exchange cartridge (e.g., commercially available from Metrohm or Altech Co.) that eliminates the conductive ions of the mobile phase is also practiced. Special devices are commercially available that use a pair of ion exchange cartridges used alternatively in the eluent flow. One cartridge is used for suppressing the eluent ions while the other is regenerated, such that no interruption in the operation is incurred. The reagents in chemical suppressors can be regenerated by electrodialysis. Additional electronic noise reduction is also available in IC-HPLC instruments [39].

116

Method Development in Analytical HPLC

Another electrochemical type of detector is based on potentiometric measurements. Those are performed in an electrochemical flow cell where an oxidation-reduction reaction takes place. An electrochemical cell is typically composed of a working electrode coupled with a nonpolarizable electrode (one that does not modify its potential upon passing of a current). The nonpolarizable electrode is known as the reference electrode, and examples include the saturated calomel electrode (SCE) and Ag/AgCl electrode. A three-electrode cell arrangement can also be used. In this arrangement, the current is passed between a working electrode (made for example from glassy carbon) and an auxiliary electrode, while the potential of the working electrode is measured relative to a separate reference electrode. Considering that any overall cell reaction comprises two independent half-reactions, the cell potential can be broken into two individual half-cell potentials. The half-reaction of interest that takes place at the working electrode surface can be either an oxidation or a reduction. A simple reduction reaction for an oxidized compound Ox (notation not indicating the charge n þ but an oxidized species can be for example an inorganic ion such as Cu2þ) to generate the compound Red is written as follows: Ox þ n e % Red

(3.3.13)

Considering a reversible reduction that has a very rapid electron transfer, and assuming that both Ox and Red are soluble species, the molar concentrations ½Ox and ½Red at the electrode surface (x ¼ 0) are governed by Nernst equation: E ¼ E0 þ

RT ½Oxðx ¼ 0Þ ln nF ½Redðx ¼ 0Þ

(3.3.14)

where E0 is the standard electrochemical potential of the half-cell (expressed in volts), R is the gas constant, T is the temperature (in K deg.), n is the number of electrons involved in the electrochemical reaction, and F is Faraday constant (F ¼ 96,485.332 C/mol). The electrode potential E 0 for this half-reaction is reported to the potential of a reference standard hydrogen electrode (NHE), which is taken as zero. However, experimental measurements are commonly done with a SCE (Hg/Hg2Cl2 or SCE) or with an Ag/AgCl reference electrode. The potential of an SCE electrode versus NHE is þ0.242 V, and the potential of an Ag/AgCl electrode is þ0.197 V versus NHE. If a compound accepts electrons from a standard hydrogen electrode, it is defined as having a positive redox potential, and a compound donating electrons to the hydrogen electrode is defined as having a negative redox potential. A high positive E 0 indicates an oxidant and a low negative E 0 indicates a reducing compound. Expression 3.3.14 is applied, for example, for measurements performed with an ion-selective electrode having an ion-specific membrane separating a reference from the studied solution [40]. The electrical potential difference across the membrane is measured with a special voltmeter and its potential is given by an expression similar to Eq. 3.3.14 where ½X ¼ ½Red=½Ox: 0

E¼E 

RT log½X nF

(3.3.15)

Instrumentation in high-performance liquid chromatography

117

Amperometric detectors are also commonly used in HPLC. In amperometric measurements, the current intensity is measured in an electrochemical cell when a specific potential is applied between two electrodes. The theory of amperometric measurements was initially developed for a static system where no convection takes place (no flow). In such case, for the very rapid electron transfer at the electrode surface, the rate vmass transfer of the electrochemical reaction (expressed in mol1s1cm2) is proportional with the current intensity I and inversely proportional with the electrode area Ael : vmass transfer ¼

I nFAel

(3.3.16)

This rate is determined by the mass transfer in solution and is given by the rate at which the electroactive species are brought to the surface of the electrode. Considering diffusion as the only mechanism for mass transfer, and the rate vmass transfer (flux) can be approximated by the following expression (equivalent with Fick’s first law): vmass transfer ¼ mOx f½Ox   ½Oxðx ¼ 0Þg

(3.3.17)

where mOx is a proportionality coefficient called mass transfer coefficient, and ½Ox  is the concentration of Ox in the bulk solution. The largest rate of mass transfer for Ox occurs from the bulk of solution to the electrode when ½Oxðx ¼ 0Þ ¼ 0. The value of the current in these conditions is called the limiting current Ilim , and its value is given by the following expression: Ilim ¼ nFAel mOx ½Ox 

(3.3.18)

The expression for ½Oxðx ¼ 0Þ can be written now using Eqs. (3.3.16)e(3.3.18) as follows: ½Oxðx ¼ 0Þ ¼

Ilim  I nmOx FAel

(3.3.19)

When the reducing species Red is absent in the bulk solution, ½Red  ¼ 0, and using a relation similar to Eq. 3.3.19 for the reduced species, the expression for ½Red  can be written as follows: ½Redðx ¼ 0Þ ¼

I nmRed FAel

(3.3.20)

With Eqs. 3.3.19 and 3.3.20, Nernst Formula 3.3.15 can be written as follows: E ¼ E0 þ

RT mOx RT Ilim  I ln ln þ nF mRed nF I

(3.3.21)

118

Method Development in Analytical HPLC

Eq. 3.3.21 represents the relation between the potential E and the current intensity I for a reversible redox reaction with very rapid electron transfer at the electrode surface. For the current intensity equal to half of the limiting current, I ¼ 1=2Ilim , the last term in Eq. 3.3.21 is null, and the corresponding potential indicated as half wave E1=2 is independent of the concentrations of the oxidant or reduced species and it is specific for a Red⎼Ox system. This E1/2 potential is known as “half wave potential” and for a diffusion-controlled process (static), a reversible reduction reaction with both Ox and Red species soluble and only with the oxidant initially present in the solution, the variation of the current intensity I as a function of the working electrode potential E follows an equation of the form: E ¼ E1=2 þ

RT Ilim  I ln nF I

(3.3.22)

The graph of Eq. 3.3.22 for a hypothetical reduction with E1=2 ¼ 0.75 V is shown in Fig. 3.3.4. For the case of the electrochemically active species flowing over the surface of an electrode, which is the case of electrochemical detection in HPLC, the currentpotential dependence is determined by a convective diffusion process (not only by diffusion). This makes the limiting current intensity for a Nernstian process dependent on the mobile phase flow rate and on channel and electrode geometry. For a rectangular channel flow electrode in steady-state laminar flow with the working electrode at one wall, the limiting current intensity in Eq. 3.3.22 is given by the relation: 2=3 1=3  Ilim ¼ 1:467 n F ½X   Ael D b1 U

(3.3.23)

2

Current I(A)

1

E 1/2

1

Ilim

0

0

-0.1

-0.3

-0.5

-0.7

-0.9

-1.1

-1.3

-1.5

Potential E(V) Figure 3.3.4 The current-potential curve of a Nernstian reaction involving two soluble species and only with the oxidant present initially. In this example, E1=2 ¼ - 0.75 V.

Instrumentation in high-performance liquid chromatography

119

where ½X   is the bulk concentration of the analyte (equivalent with ½Ox ), Ael is the electrode area, D is the diffusion coefficient of the analyte, b is the channel height, and U is the volumetric flow rate [41e43]. For different channel and electrode shapes, the expression for the current intensity is different [44,45]. In amperometric detection, the current passing through the cell is measured at a fixed potential E, commonly chosen higher in absolute value than E1=2 specific for the analyte. In these conditions, the desired electrochemical process takes place, but also all other species present in solution and having E1=2 lower (in absolute value) than the chosen E value can become electrochemically active species. This may include even the solvent if the working potential E is very high. For eliminating this type of interference, compounds with low electrochemical potentials (in absolute value) are preferred for electrochemical detection. In HPLC, amperometric detection is frequently used for oxidation reactions. The quantitation can be done by calibration of the measured current Ilim versus different concentrations ½X   of analyte while maintaining strictly controlled flow conditions. Also, instead of a constant oxidation potential, a pulsed amperometric detection (PAD) can be used, alternating the oxidation analytical potential with a reducing pulse used for cleaning the electrode (depositions on the electrode may modify the nature of its surface and therefore the cell potential). The application of different working potentials is done at specific time intervals, and the measurement is made only when the active species are oxidized [46]. In addition to amperometric measurement, coulometric methods can also be utilized for detection. In a coulometric method the quantity of electricity Q (in coulombs) is measured, and the value of Q is related to the current intensity I and the period of time of the measurement t are related by the expression: Q ¼ Ilim t

(3.3.24)

For convective diffusion as it takes place in HPLC, Ilim is given by Eq. 3.3.23, which also relates Q with the concentration ½X. Various applications of coulometric detection are described in the literature for HPLC analyses (e.g., Ref. [47]).

Evaporative light-scattering detectors One other detector utilized in HPLC, in particular for compounds that do not have good light absorbance in UV, are not fluorescent and may be difficult to ionize, is the evaporative light-scattering detector (ELSD) (e.g., Ref. [48]). ELSD uses the formation of particles that do not evaporate and can scatter light, while the mobile phase forms a gas by evaporation. These detectors cannot be used for volatile compounds, but have a universal character not differentiating between different types of nonvolatile molecules. In this technique, the eluent is injected in the form of a spray from a nebulizer into a drift tube where also a nebulizer gas is introduced. The drift tube is heated and the solvent is evaporated, while nonvolatile molecules form a fine mist. This mist passes through a cell that is illuminated with a beam of light and the scattered light from the mist is recorded. The gas generated from the solvent does not influence light scattering. The schematic diagram (not to scale) of an ELSD is shown in Fig. 3.3.5.

120

Method Development in Analytical HPLC

Photo multiplier Scattered light

Eluent from HPLC

Analyte and solvent

Assistant gas Heated drift tube

z y

x

.. .. .. .. .. .. ... .. .. .. .. ..

.. . .... .. . Particles of solid analyte ... . . . .. .. Assistant + gas solvent . gas Nebulizer gas

Light source

Drainage

Figure 3.3.5 Schematic diagram of an ELSD setup.

The intensity of the scattered light is dependent on the analyte concentration (within a certain range of concentrations since the linearity is not followed for a wide range). This detector has the advantage over the RI of being usable with gradient elution. However, the presence of any salts or nonvolatile materials in the mobile phase disturbs the measurements. Also, changes in the temperature of the eluent do not affect ELSD, while with RI detectors, a careful temperature control must be applied since the refractive index varies with temperature. ELSD can be more sensitive than RI in specific applications. Modifications of ELSD were attempted to further improve its sensitivity, such as by adding a saturated stream of solvent to the mist of the analyte in order to grow the particles by nucleation and detect them better (the technique is known as condensation nucleation or CNLSD) or to use lasers as light source (LLSD). Light scattering detector does not usually have a linear response to analyte concentration, in particular when the concentration is relatively high. For this reason, in some analyses, the calibration is performed using a log/log dependence. Using the notations ½X for the analyte X molar concentration and y the peak area in the chromatogram generated by the signal from the photomultiplier, this dependence can be written as follows [49,50]: log½X ¼ a þ b log y

(3.3.25)

A multiangle light scattering (MALS) detector is also available for HPLC. MALS detector is successfully utilized for characterizing the molecular weight of polymers (including proteins) by measuring the light scattered intensity IðqÞ by the eluent stream at different angles q. The value of I is a function of polymer Mw, concentration ½X, refractive index increment ðdn =d½XÞ, and a molecule shape factor denoted by PðqÞ [51]. The simultaneous determination of Mw and ½X can be achieved by using SEC separation coupled with multiangle laser light detector and differential refractive index detector.

Instrumentation in high-performance liquid chromatography

121

An alternative to light scattering detection is a corona charged aerosol detector (CAD or cCAD) [52]. This detector is also based on nebulization of column effluent (e.g., with N2) and drying of resulted droplets to remove the mobile phase components, producing analyte particles. A secondary stream of N2 is made positively charged by passing it by a high-voltage platinum corona wire, which is sent in the stream of analyte particles to generate charged aerosols particles. The charged particles obtained by the interaction of the aerosols with the charged nitrogen gas are detected, generating an electric signal with the intensity dependent on the amount of analyte eluted from the column. The typical response of a CAD detector follows a dependence as indicated by Eq. 3.3.25. The CAD is more sensitive than ELSD and has a wider dynamic range [53]. Similarly to ELSD, one of the drawbacks of CAD is its inability to accurately quantify volatile compounds. Because CAD can only measure analytes that form particles, it cannot detect compounds that are volatile or do not readily form particles. In this situation, detection can be achieved by other specific detectors, such as chemiluminescent nitrogen detector (CLND) [54]. CAD detection is reported to be possible in a wide range of concentrations.

Other types of detectors A variety of other detection techniques applicable to HPLC are reported in the scientific literature. For some of these techniques, there are HPLC detectors commercially available. For other techniques, only experimental equipment was created in some laboratories. For example, in case the target compounds contain at least one nitrogen atom, the CLND can be employed. The principle of this detector relies on the combustion of the column effluent in a high temperature furnace that converts the Ncontaining compounds into NO. The dried gas stream is passed into a chamber where it reacts with O3, a reaction that is associated with CL (measured by a photomultiplier). This detector has a high sensitivity, but is not compatible with acetonitrile in the mobile phase [55]. CLND detectors for HPLC are commercially available. Also, inductively coupled plasma-mass spectrometry (ICP-MS) system ready for hyphenation with HPLC are commercially available. Such systems are very useful in measuring analytes containing metals, some nonmetals (S, P, and halogens), and can be used even for measuring different isotopes of the same element [56,57]. Other known detection techniques include Fourier-transform infrared spectrometry (FTIR) [58,59] and Raman spectroscopy [60,61]. The utilization of these techniques in HPLC detection is still in experimental stage. Also, continuous effort is being made to use NMR as detection technique in HPLC. The advantages of NMR in providing structural information makes this technique an attractive alternative for detection, but the hyphenation HPLC-NMR still encounters a number of difficult problems [62,63]. Circular dichroism (CD), optical rotatory dispersion, polarimetry, being able to provide information regarding the stereochemistry of chiral eluates, are sometimes used for HPLC detection of stereospecific compounds [64]. In CD spectroscopy, for example, the molar CD Dεl is measured as the difference of ε for the left circularly polarized light and right circularly polarized light. The value for molar CD Dεl has opposite signs for the enantiomers of a molecule.

122

Method Development in Analytical HPLC

For the detection of specific biological materials, the application of plasmon resonance technique was described [65]. Another detection technique, the MALS which is typically used directly to solution of polymers, can also be used in connection with size exclusion chromatography [66,67]. Examples of applications of MALS hyphenated with size-exclusion chromatography (SEC-MALS) are the determination of mass and size of proteins and the characterization of protein-drug aggregation and antibody-drug conjugates [68,69].

The use of multiple detectors Multiple detectors can be used in an HPLC system. For example, a configuration based on double detection between UV (first) and CLND (second) can be advantageous for comparing quantitative responses for analytes with UV chromophore, such as nucleotides, nucleosides, and their corresponding bases [70]. The detectors are usually connected in series since a parallel connection must consider the specific backpressure generated by each detector such that enough flow should still be delivered to a detector that generates high backpressure. The detector that destroys the analyte in the detection process cannot be followed by another detector. In case one detector does not alter the nature of the analyte, a second detector can be selected such as UV followed by MS or MS/MS. The use of three detectors in line is not frequently utilized, although three detectors can be connected in line in an HPLC setup (e.g., UV, fluorescence, and RI) and utilized one by one depending when a specific type of detection that is necessary.

Key points • • • •

In HPLC instruments, the detection module can be based on different physical or chemical properties of the analyte. The most common detectors are UV-Vis, refractive index, fluorescence, and mass spectrometric. Other detectors may have specific applications such as conductometric detectors used in ion chromatography. Many types of detection techniques were experimented for being used as detection for HPLC.

3.4

Mass spectrometers used as HPLC detectors

Mass spectrometric detection is commonly applied for HPLC, offering a diversity of capabilities depending on the type of instrumentation used. The mass spectrometers can range from simple MS instruments to tandem MS/MS instruments, and from unit resolution to high resolution and to very high-resolution instruments such as the Fourier transform ion cyclotron resonance (FT-ICR) instruments [71]. The mass spectrometers can also be connected to the HPLC column via an ion mobility unit that further improves analytes separation. Mass spectrometric detectors are characterized by their capability to be adjusted for the detection of a specific compound

Instrumentation in high-performance liquid chromatography

123

improving significantly the selectivity of analytical methods. In some applications, the mass spectrometric detectors can also be used for unknown compound identifications or even structural elucidations. The wide utilization of mass spectrometry hyphenated with chromatography is covered by a large volume of information presented in dedicated books, peer review papers, and on the web. In a mass spectrometer, gaseous ions are first generated from the analytes present in the eluate flowing from the chromatographic column. These ions are separated based on their mass to charge ratio (M=z units sometimes indicated as Th ¼ Da/ e z 1.036426 108 kg/C), and then detected. For each of these three functions in a mass spectrometer exists a basic component: the ion source, the mass analyzer, and the ions detector. Some information regarding those components is further provided.

Ion source for mass spectrometers hyphenated with HPLC The ion source has the role of generating gaseous ions from the sample components eluted from the separation module (column) of the HPLC instrument. Ionization of analytes eluted from the HPLC column cannot be performed like in GC-MS where effective ionization can be achieved using electron impact or chemical ionization. In HPLC, the analytes and mobile phase are in liquid form, and the ratio of analyte molecules versus mobile phase molecules is very low. The most successful and frequently utilized procedure for generating ions from the analytes in an HPLC eluate is electrospray ionization (ESI) (e.g., Ref. [72]). ESI is a soft ionization procedure in which mainly molecular ions are formed, although some fragmentation can also be obtained, depending on the stability of the incoming molecules. Gaseous ions are generated even from large molecules by this procedure. In positive mode, from a compound X, it is common that ions of the type X‑H þ are formed in the ion source, and in negative mode, ions of the type X  are formed. The ions formed in ESI are in most cases even-electron ions (EE ions) which is different from the ions generated by electron impact used in GC-MS where mainly radical ions are formed (odd-electron OEþ ions). In addition to single charged ions, the formation of multicharged ions is also possible in LC-MS. Because the ions formed in LC-MS have a close electron shell structure, they are more stable than the radical ions (open shell structure) formed by electron bombardment in GC-MS. For positive ion formation, the Hþ ions which are abundant in a mobile phase containing water lead to the production of positive molecular ions of the type XHþ where X are the molecules of the analyte by the following reaction: X þ H þ /XHþ

(3.4.1)

The displacement of reaction 3.4.1 toward the formation of more XHþ ions is determined by the proton affinity of molecule X which is characterized the gas-phase basicity (GPB) (see Section 5.2). The GPB is given by the free energy (enthalpy) released by reaction 3.4.1 where DG0GPB is given by the expression:     DG0GPB ¼ DG0 XHþ  DG0 ðXÞ  DG0 Hþ

(3.4.2)

124

Method Development in Analytical HPLC

This type of reactions is always exergonic in the gas phase. However, proton affinities are conventionally quoted with the opposite sign from other thermodynamic parameters (exothermic reactions having assigned a negative free enthalpy), and a positive value of DG0GPB indicates a release of energy by the system. The higher is the proton affinity, the stronger is the base, and the weaker is the conjugate acid in the gas phase. Usually, DG0GPB for most organic compounds lies between 500 and 1000 kJ/mol. Besides the formation of single charged ions, the formation of multiple charged ions is also possible in ESI [73,74]. The formation of multiple charged ions takes place mainly for molecules such peptides and proteins that have in their structure atoms with high electron density such as nitrogen atoms. Formation of multiple charged ions has considerable importance by extending the possibility to analyze larger molecules. These multiple charged molecules are separated in the mass analyzer by their M=z value and not by their mass M. As a result, if the maximum single charged mass that can be analyzed is M, the maximum double charged mass is 2 M, the maximum triple charged mass is 3 M, etc., and in this way, the mass range that can be measured by the instrument can be significantly extended. Similar to the formation of positive ions, molecules of the type XH can be ionized to generate negative ions in a reaction as follows: XH / X  þ H þ

(3.4.3)

Reaction 3.4.3 is related to the tendency of losing a proton, a property also known as gas phase acidity (GPA), and it is described by the variation of free enthalpy DG0GPA :   DG0GPA ¼ DG0 ðX  Þ þ DG0 Hþ  DG0 ðXHÞ

(3.4.4)

The values for DG0GPA for most organic compounds lie between 1300 and 1650 kJ/ mol. The larger values for DG0GPA than for DG0GPB are responsible in part for the usual lower sensitivity of negative ionization than of positive ionization in LC-MS. The gas phase basicity and acidity are energetically significantly different from the same ionization process in solutions, due to the energy contribution of hydration of ions in aqueous phase. However, directionally they tend to have similarities in the sense that more basic compounds in solution have more tendency to form positive gas ions and more acidic compounds in solution have more tendency to form negative gas ions. The values for DG0GPA in positive ionization and for DG0GPB in negative ionization are important values for the ionization efficiency in ESI and various attempts were made to evaluate this efficiency (e.g., Ref. [75]). The formation of ions in the source of a mass spectrometer is a more complex process than the simple reaction described by Eq. 3.4.1. Ionization can also take place by a process indicated as charge transfer, and this process occurs between the solvent molecules and the molecules of the analyte. The difference in the polarity between the solvent and the analyte molecules favors the formation of positive ions from the analyte

Instrumentation in high-performance liquid chromatography

125

(or negative ions when working in negative mode), and much less from the solvent. However, due to the high concentration of the solvent, (although it is significantly less ionized than the analyte) it is common that some of its molecules are also ionized to form ions SolvHþ (or in case of a SolvH molecules to form by negative ionization Solv ). Further ionization of the analyte can be described as a proton transfer from the ionized solvent to the molecule of the analyte. X þ SolvHþ % XHþ þ Solv

(3.4.5)

or in case of negative ionization: XH þ Solv % X  þ SolvH

(3.4.6)

The proton transfer reactions show that the solvent has an important role in the ionization process of the analytes in LC-MS. Similar to the case of analyte molecules, the gas-phase basicity (or acidity) of the solvent molecules DG0GPB ðSolvÞ (or DG0GPA ðSolvHÞ) are important values regarding the formation of ionized solvent and the proton transfer reactions. The proton transfer process as a source of analyte ions can have an important contribution to the ionization efficiency of the analyte and therefore is related to the sensitivity of mass spectral detection. For this reason, the sensitivity of the MS detectors is strongly affected by the nature of the mobile phase. A volatile acid such as HCOOH is typically added in the mobile phase to favor the process of positive ion formation. For negative ion formation, NH4OH or salts such as HCOONH4 or CH3COONH4 are usually added to favor the ionization of the analyte, although HCOOH can sometimes be used as additive even in negative mode ionization. The schematic description of an ESI source is indicated in Fig. 3.4.1 for positive ionization mode. A similar process takes place in negative ionization when the polarity of the power supply is reversed.

Capillary with effluent from HPLC

N2

Charged droplets

N2

- + + + - +- - - - +

N2

Radiative heater Charged progeny Charged droplets at Rayleigh limit droplets Solvent, neutral Coulomb Solvent molecules evaporation fission

+

+

+

+ +

+ +

e-

+

+

+

+ + +

+

+

+

+

+ +

+

Chamber at atmospheric pressure

Taylor cone

+V

Power supply 2.5 – 5.5 kV

+ +

+ + + +

+ +

+ +

-V

Curtain plate

+

Charged naked analyte molecules

+ +

Curtain gas

Mass analyzer

Figure 3.4.1 The schematic description of an electrospray (ESI) source for positive ionization mode.

126

Method Development in Analytical HPLC

The ESI source consists of a chamber kept at atmospheric pressure in which the eluent from the HPLC flows usually at values lower than 1.0 mL/min (typically 0.1e0.8 mL/min). Two main stages are involved in the mechanism of ion generation: the formation of the charged droplets from the sprayed solution in the presence of a high electric field and the production of gas-phase ions from the dispersed charged droplets [76,77]. The eluent carrying the analytes is introduced in the ESI source through a metal capillary (0.1e0.2 mm i.d.) which is kept at a high voltage (2.5e5.5 kV) relative to the inlet of the mass analyzer. The inlet is an opening in the curtain plate of the mass analyzer which is typically located at 2e3 cm from the tip of the capillary, and positioned using several geometries depending on instrument manufacturer. In Fig. 3.4.1, the schematic indicates an orthogonal position common in some instruments. Other geometries are also utilized such as a dual orthogonal design (e.g., ZSpray [78]). A gas flow (usually N2) coaxial with the capillary assists the dispersion of the HPLC eluent into an aerosol of charged droplets. The charged droplets are diminished in size by solvent evaporation assisted by a radiative heater and in some sources by additional N2 flow. The desolvation of droplets continues to different stages, reaching the point where surface tension of the droplets becomes equal with the Coulomb repulsion of charges (Rayleigh limit) followed by the disintegration of droplets due to electrostatic charges repulsion (Coulomb fission), resulting in charged progeny droplets [79,80]. The analyte molecules generate gas phase ions that are directed toward the curtain plate of the mass analyzer and are further transmitted to mass analyzer. This process involves very low energies that usually allows the ions to remain intact (with no fragmentation) even for large molecules [81]. In the ionization process, most solvent molecules and some of the analytes molecules do not form ions. As a result, they are not directed toward the mass analyzer entrance by the difference in electric potential. A limited number of neutral molecules can still enter the mass analyzer. To avoid this process, a flow of a curtain gas is used, such that the neutral molecules are not pulled by the vacuum in the mass analyzer, and only the ions of the analyte and some molecules of the curtain gas (usually N2) are introduced there. Additional pumping further diminishes the content of curtain gas in the mass analyzer. The mass analyzer is kept at a very low pressure (2e3 106 Torr or even lower, depending on the type of instrument). For the electrospray, the ionization efficiency depends on analyte properties, instrumental settings, and also on the mobile phase composition and flow rate [82]. This efficiency is one of the main factors affecting the overall detector sensitivity. In the process of formation of ions in an ESI source, water and, in many applications, low levels of HCOOH are used in the mobile phase carrying the analytes from the chromatographic column. The result of water presence is that Hþ ions are abundant and lead to the production of positive molecular ions of the type XHþ where X are the molecules of the analyte. This process is valid for positive ion detection, where the presence of water is also important in the negative ion formation (when negative ionization is used). In some applications, besides water generating OH ions, volatile salts and/or NH4OH can be added to enhance negative ion formation. In some applications, where

Instrumentation in high-performance liquid chromatography

127

water cannot be used in the mobile phase for specific restrictions imposed by the separation (e.g., water insoluble analytes, or chiral separation that are negatively affected by the presence of water), protic solvents such as methanol or ethanol may also assist the ionization. The ionization process is also influenced by other factors related to the mobile phase and analyte properties, among them being the surface tension and surface charges interactions in the sprayed droplets [83]. The instrumental parameters affecting ionization efficiency are different on different instrument models, some such parameters are fixed in some instruments and can be set by the user in others. In some cases, parameters are indicated by different names even if they have the same function. Ion spray voltage, nebulizing gases flow rate, drying (desolvation) temperature, etc., are usually common adjustable parameters. Other parameters such as curtain gas flow, declustering potential, cone voltage, etc., are parameters to be set only in some instruments. These parameters affect in different ways the ionization process. For example, the nebulizing gas flow influences the growth of the mobile phase droplets and the drying temperature affects the analyte stability and efficiency of ions formation. A higher flow rate for the mobile phase requires higher flow of the nebulizing gas and higher desolvation temperature to increase ionization [84]. A number of modifications of the typical ESI source were reported, having the purpose to enhance the nebulization process, such as heated ESI for handling flow rates higher than 1.0 mL/min, pneumatically assisted electrospray [85], ultrasonic nebulizer electrospray [86], etc. Also, for very low flow rates of the mobile phase (20e50 nL/ min), a nanoelectrospray version has been developed [87]. A rather similar process with ESI type of ion formation takes place in the atmospheric pressure chemical ionization (APCI) process. The schematic description of an APCI source is indicated in Fig. 3.4.2 for positive ionization mode. The main difference in the ionization process between ESI and APCI is that the ionization in APCI takes place for partially (or mainly) desolvated droplets which are closer to a gas phase. In this way, a potential advantage of APCI is that it is possible

Capillary with effluent from HPLC

Radiative heater Droplets

N2

- + + - +- - - - +

+

+ + +

Further desolvation

Solvent evaporation

-+

Charged partially desolvated droplets Solvent, neutral molecules + + + + +

-

+ + + + + +

- --

+ + + + ++ + +

N2

Corona needle Chamber at atmospheric pressure

e-

+V

Power supply 2.5 – 5.5 kV

Curtain plate

+ + +

Charged naked analyte molecules

+

-V

+

Curtain gas

Mass analyzer

Figure 3.4.2 The schematic description of an APCI source for positive ionization mode.

128

Method Development in Analytical HPLC

to use a nonpolar solvent as a mobile phase solution, instead of a polar solvent, since the solvent and molecules of interest started to be converted to a gaseous state before reaching the corona discharge needle. Because APCI involves basically a gas-phase ionization, the role of water or of other protic solvents in the ionization is less important, and a wider range of solvents can be used with APCI including nonaqueous type. APCI appears to be a more versatile MS ion source than ESI, although the ionization efficiency in the presence of protic solvents is usually significantly higher in ESI which explains the use of ESI in most applications. Modifications of typical APCI ionizations were also described in the literature. One such procedure is based on a chemical ionization process whose energy is generated through collisions between analyte molecules with reagent ions generated with a buffer gas [88]. The buffer gas can be N2 or He, which is ionized by a beam of electrons accelerated in a high electric field [89]. Another APCI type ionization is based on low temperature plasmas resulted from a “dielectric barrier discharge source” (DBD) [90]. Other procedures besides ESI and APCI are used to form ions from the analyte molecule from the HPLC effluent. One of these is the use of an intense beam of UV light for the analyte ionization instead of a corona discharge. This technique is known as atmospheric pressure photoionization (APPI). Also, in an effort to generate for the analytes in the LC eluent mass spectra that are similar to EI þ mass spectra (searchable in standard mass spectral libraries such as NIST8), special instrumentation was being developed such as LC-MS with supersonic molecular beam and high energy electron ionization capability [91,92]. Older techniques to interface an LC with a mass spectrometer were experimented such as particle beam (PB), continuous flow fast atomic bombardment (FAB), thermospray, direct analysis in real time (DART) [93], etc. One limitation of ion sources in LC-MS (and LS-MS/MS) is related to the requirement that the evaporation of the solvents from the mobile phase should not produce solid residues. Very low concentration of a wide variety of molecules, even not volatile, do not pose a problem with the ion generating sources. For example, non-volatile additives in the mobile phase at levels not higher than 40e50 m M/L can be used, and buffers with salts such as KH2PO4 or K2HPO4 at higher than those levels cannot be used. The formation of fine solid particles in the ionization chamber of the MS leads to a decrease in ionization efficiency and also generates a very instable signal in the MS instrument (high background noise). This imposes that the buffers that can be used must contain acids salts or bases that will volatilize at the high temperature of the radiative heaters used in the ion source. Acids such as HCOOH, CH3COOH, CF3COOH, bases such as NH4OH, and salts such as HCOONH4, CH3COONH4, or NH4HCO3 are used for buffers or additives [94]. Another problem with LC-MS using ESI or APCI sources is related to the problem of ion suppression when the number of ions generated in the source is too high. These high levels of ions can be caused by a high level of analyte but also by high level of matrix that was not well separated by the chromatographic column. In addition to the formation of the ions from the analyte, in the MS source, a possible effect is the formation of ions from various adducts of the analyte and other ions present in the mobile phase. Such adducts are formed by ion-dipole, ion-induced dipole, hydrogen bonds, and even by van der Waals interactions (see Section 2.2). In

Instrumentation in high-performance liquid chromatography

129

some instances, the adducts are stable enough to be seen in the mass spectrum instead of the analyte molecular ion. Besides adducts with solvent molecules, other adducts can be formed between the analyte X and other species in the eluted material. For example, in positive ionization mode, ions such as [X-Na]þ, [X -K]þ, [X-NH4]þ, [X eH2OeNa]þ, [X-Solv-Na]þ, [X-2 Solv-Na]þ can be seen. Negative ionization is less favorable to adduct formation, but still possible with ions of the following types: [X-Cl]e, [X-Br]e, [X-HCOO]e, [X-CH3COO]e, [X-CF3COO]e, etc. The formed ions from the adducts can also have multiple charges. A list of masses for some possible adducts seen in mass spectrometry is given in Table 3.4.1 (Note: In mass spectrometry, the mass of the measured ion is the mass corresponding to the ion formed from the most abundant isotopes of the constituent atoms and it is indicated as the exact mass, see Section 5.1). For the detection of ions in MS and MS/MS, the most common procedure is to use in positive ionization the ion [X þ H]þ and in negative ionization, the ion [X - H]e (see Section 5.2). However, in some cases, different ions can be utilized as parent ion. For example, for the analysis of a ginsenoside (Rg1), the ion used as parent for detection was [Rg1 þ Na]þ. The mass spectrum of Rg1 ginsenoside is shown in Fig. 3.4.3. In the mass spectrum, the ion intensity corresponding to the Rg1 þ H þ ¼ 801.5 is very low. The spectrum also shows various fragments with loss of glucose moiety (glc) and of water and can be useful for qualitative identification of the ginsenoside.

Table 3.4.1 The change in the mass of ions generated as a result of adduct formation. Ionization mode Positive

Negative

Adduct ion [X [X [X [X [X [X [X [X [X [X [X [X [X [X [X [X [X [X [X [X

þ

þ H] þ 2H]2þ þ NH4]þ þ H þ NH4]2þ þ Na]þ þ H2Oþ Na]þ þ 2Na]2þ þ K]þ þ H þ K]2þ þ Cs]þ þ CH3OH þ H]þ þ CH3CN þ H]þ þ DMSO þ H]þ þ CH3CH(OH)CH3 þ H]þ - H]e þ Cl]e þ Br]e þ HCOO]e þ CH3COO]e þ CF3COO]e

Nominal mass

Exact mass

Xþ1 X/2 þ 1 X þ 18 X/2 þ 9.5 X þ 23 X þ 41 X/2 þ 23 X þ 39 X/2 þ 20 X þ 133 X þ 33 X þ 42 X þ 79 X þ 61 X-1 X þ 35 X þ 79 X þ 45 X þ 59 X þ 113

X þ 1.007825 X/2 þ 1.007825 X þ 18.034374 X/2 þ 9.521100 X þ 22.989769 X þ 41.000334 X/2 þ 22.989769 X þ 38.963706 X/2 þ 19.985766 X þ 132.905452 X þ 33.034040 X þ 42.0343741 X þ 79.0217610 X þ 61.065340 X - 1.007825 X þ 34.968853 X þ 78.918338 X þ 44.997654 X þ 59.013304 X þ 112.985039

130

Method Development in Analytical HPLC 423.4

7.9e6

[Rg1 - 2 glc - H2O + H]+

7.5e6 7.0e6 6.5e6 6.0e6 5.5e6

Intensity, cps

5.0e6

[Rg1 - glc - H2O + H]+

4.5e6 4.0e6

603.4

[Rg1 - 2 glc + H]+

3.5e6 3.0e6

[Rg1 - 2 glc - 2 H2O + H]+

441.4

2.0e6

823.5

0.0

[Rg1 + Na + HCOOH]+

405.4 869.3

1.5e6

5.0e5

Rg1 Mw = 800.5

621.4

2.5e6

1.0e6

[Rg1 + Na]+

[Rg1 - glc + H]+

217.2 175.2 203.2

235.1 367.3 123.2 199.0 385.2 459.3 239.1 281.2 299.1 541.4 585.4 150

200

250

300

350

400

450

500

550

707.4 643.4 661.3 689.4 665.4 600

650

801.5839.0 854.4

700 750 m/z, Da

800

850

921.3 913.1953.2 1009.31057.5 1083.2 900

950

1213.2 1242.6

1000 1050 1100 1150 1200 1250 1300

Figure 3.4.3 Spectrum of Rg1 in the range 120e1300 Da, obtained in single quadrupole Q1 MS conditions in positive mode with Naþ ions added to the mobile phase.

The spectrum from Fig. 3.4.3 was generated on a Sciex 6500þ triple quadrupole mass spectrometer (Danaher Corp, Washington DC, USA) controlled by Analyst 1.7.2 software [95]. Although in most MS instruments the ESI source and the APCI source are different, new instruments having a single nebulizer to produce aerosols which can switch between ESI and APCI to generate ions have been commercialized [96] and utilized in various practical applications (e.g., Ref. [97]).

Mass analyzers The ions generated in the ion source of an MS are separated, in a mass analyzer (mass filter) module, in time or space based on their M=z value. This separation is performed by a number of procedures and with various separation “distances.” The separation power of the mass analyzer is characterized by its resolution expressed by the formula: R¼

M DM

(3.4.7)

In Eq. 3.4.7, DM is the closest spacing of two masses that can be separated (in mass units). The separated masses are assumed to be of equal intensity with the valley between them less than a specified fraction of the peak height (e.g., 50%). The value of M is taken as the mass of the second peak. Sometimes DM is replaced with FWHM (full width at half maximum) characterizing a width of a single peak for a specific mass M of an ion. Based on their resolution, mass spectrometers are usually classified as “low

Instrumentation in high-performance liquid chromatography

131

resolution” and “high resolution” (sometimes “medium resolution” and “ultra-high resolution” are recognized). The low resolution typically discriminates ions with a difference in mass of 0.2 m.u. (R z 1000 for M ¼ 200), while a high-resolution instrument can discriminate ions with a difference in mass of 0.0001 m.u. (R z 2,000,000 for M ¼ 200). The values for the masses of the separated ions as produced by the mass analyzers can be obtained following calibrations with compounds with known mass values. Based on their construction, mass analyzers can be of different types, the more common ones being the quadrupole, ion trap, and time-of-flight (TOF). The quadrupole type mass analyzer separates the ions by passing them along the central axis of four parallel equidistant rods that have a fixed voltage (DC) and an alternating (RF) voltage applied to them. The field strengths (voltage) can be set such that only ions of one selected mass can pass through the quadrupole, while all other ions are deflected to strike the rods. By varying (with a precise rate) the strength and frequencies of the electric fields, different masses can be filtered through the quadrupole. The theory of ion motion in a quadrupole field is reported in the literature [98]. A schematic diagram of a quadrupole mass analyzer is shown in Fig. 3.4.4 and the following are typical values for the parameters of the electrical field: u is about 1e2 MHz, V1 ¼ 1000 V, and V2 ¼ 6000 V. The resolution R obtained with a quadrupole analyzer is usually around 1000 and the mass range can go as high as 2000 although is usually limited to 1100. By varying the voltages of the quadrupole with the frequency u, only ions of a specific M=z reach the detector at a given time t. For recording the whole range of masses (e.g., from 2 to 1100 amu), the instrument is scanning the mass range with a specific speed. For a common quadrupole, the scan speed is, for example, 5200 amu/sec with 0:1 amu scan step size. The instrument can be used not only in scan mode but also in SIM mode (single ion monitoring). In SIM mode, the dwell time can be between 9 ms and 10 s per mass and can be selected by the user (the dwell time is the amount of time that the mass spectrometer spends on each ion). The quadrupole mass analyzer has the disadvantage of measuring individual M=z values in a sequence and cannot be used for generating high resolution spectra. The system is typically kept at vacuum in the range of 5 105e5 106 Torr. The mass separation in a quadrupole must be calibrated for generating accurate mass values. This is usually performed for ESI ion sources using mixtures of special calibration compounds with known mass values. Common calibrants are mixtures of polyethylene glycols (PEGs) of specific Mw values, of polypropylene glycols (PPGs), compounds such as CsI which forms clusters of the form (CsI)nCs þ or (CsI)nIe, or special mixtures of perfluorinated organic acids, etc.

m2

B

Detector

A

A

B

-V1- V2 cos (ω t)

B

A

A

+V1+ V2 cos (ω t) Ion

beam

m1

B

Figure 3.4.4 Schematic diagram of a quadrupole mass analyzer.

132

Method Development in Analytical HPLC

Entrance endcap electrode

Ring electrode at RF potential

Exit endcap electrode

Ion beam Detector

Figure 3.4.5 Schematic diagram of an ion trap mass analyzer.

The ion trap mass analyzer (IT) consists of a hyperbolic cross-section center ring electrode (with a doughnut shape) and two hyperbolic cross-section endcap electrodes. The ion trap works in cycles. A cycle starts with the application of a low RF amplitude and fixed frequency to the ring electrode (no DC), while the endcaps are grounded. A pulse of ions is injected into the ion trap. The ring electrode at low RF amplitude traps all the ions. The RF amplitude is then increased, and ions of increasing mass (in fact M=z) are sequentially ejected from the trap and detected. The schematic diagram of an ion trap mass analyzer is shown in Fig. 3.4.5. A key parameter to the operation is the gas pressure inside the trap (usually He), which must be maintained at 10e3 Torr. The gas forces the ions toward the middle of the trap and provides a better sensitivity [99]. A small AC voltage of fixed frequency and amplitude is also applied to the endcaps during the analysis part of the cycle for ejecting the ion with a specific M=z value. Then the process is repeated. The ion traps generate low-resolution mass spectra but can have very good sensitivity. A special type of ion trap, used for generating high-resolution separations and mass detection is the orbitrap mass analyzer (under the name “Orbitrap” are also commercialized whole mass spectrometers such as various series of Orbitrap from Thermo Scientific). Orbitrap mass analyzers are a special type of ion traps consisting of an outer barrel-like electrode and a coaxial inner spindle-like electrode that traps ions in an orbital motion around the spindle. The resulting frequency of the signal generated by the moving ions is converted to a mass spectrum using the Fourier transform. The ions injected in an orbitrap to be analyzed for their M=z must have very stable trajectories and precise kinetic energies. This is achieved using a special devise indicated as a C-trap which is a special type of ion guide. The C-trap þ orbitrap mass analyzer achieves very high resolution (e.g., M=DM z 60,000 for M=z ¼ 400 or even up to 100,000) (e.g., Ref. [100]). Some Orbitrap instruments (BioPharma option) can cover a mass range from 40 to 8000 M=z (Th). Differently for typical ion trap mass analyzers, the orbitrap mass analyzers require a vacuum around 2 1010 Torr. The time-of-flight (TOF) analyzer is another common type of mass analyzer. The principle of TOF instruments is based on the separation of ions of different M= z ratio based on their different kinetic energies after being accelerated in an electric field. In an electric field with the potential difference V, the energy of the particles having the charge z is E ¼ zV and their kinetic energy is E ¼ 1=2 Mv2 . Therefore, the following expression can be written for the particle:

Instrumentation in high-performance liquid chromatography

1 zV ¼ Mv2 2

133

(3.4.8)

In a field-free TOF tube of length L, the t ¼ L=v “time of flight” for a particle can be therefore obtained from the expression: L pffiffiffiffiffiffiffiffiffi t ¼ pffiffiffiffiffiffi M=z 2V

(3.4.9)

Eq. 3.4.9 indicates that in a TOF instrument, the ions will reach the detector at a time proportional to the square root of their M=z ratio. Various construction types of TOF mass analyzers have been developed. A common type indicated as reflectron TOF uses a constant electrostatic field with a potential gradient to reflect the ion beam toward the detector. The more energetic ions penetrate deeper into the path within the reflection lenses and take a longer time to reach the detector. Less energetic ions of the same M=z penetrate a shorter distance into the reflectron and take a shorter time to reach the detector. A schematic diagram of a reflectron TOF is indicated in Fig. 3.4.6. TOF spectrometers work on scans using a repeated pulse of ions. This can be achieved, for example, using an orthogonal acceleration where the ions resulting continuously from the collision cell are directed orthogonally to the direction of the ion beam by a pulsed electric field applied to a repeller plate. The number of acquired spectra per second for a TOF instrument can be as high as 1000. Depending on their construction, some TOF instruments are made to generate low resolution spectra and other instrument are made for high resolution (R z 60,000). TOF instruments can have wider mass range than quadrupole instruments and some instruments can reach 10,000 M=z (Th). For molecules with multiple charge such as proteins, this allows separation of molecules as large as having up to 40 kDa. Also, the TOF instruments measure all M=z simultaneously and not a mass unit at a time as the quadruple mass analyzer. The TOF instruments must be kept at very low vacuum, usually 10e7 to 10e8 Torr to avoid collision of the ions with neutral molecules present in the flight tube. Also, mass calibrations are necessary. The mass calibration can be performed using an external calibration when the calibrant solution is used previous to the analysis Accelerating lenses

Figure 3.4.6 Schematic diagram of a reflectron TOF.

Flight tube

Ion trajectory

Ions M1/z

M2/z

Detector Reflection lenses

134

Method Development in Analytical HPLC

of samples. In internal calibrations, the calibrant is infused together with the sample, providing higher precision and accuracy of the calibration, but at the same time having the disadvantage of potential interference with the analysis [101]. Several other types of mass analyzers have been developed and are utilized. Magnetic sector instruments, for example, were used for the ion separation. In a magnetic sector instrument, the ions from the ion source are sent through a nonhomogeneous magnetic field where they are deflected on different arc trajectories based on their M=z ratio and focused using an electric field followed by detection. Magnetic sectors mass detectors can separate ions within a wide range of masses and can achieve high resolution with excellent sensitivity. The cost of magnetic sector instruments is however higher than for other types of mass detectors. A number of other techniques for the separation of ions are known, such as Fourier transform ion cyclotron resonance (FTICR-MS) capable of 1,000,000 resolution, Wein filter, etc.

Collision-induced dissociation cell and ion guide The sources of ions in the mass spectrometers used as detectors in HPLC involve in most cases a soft ionization. This is the case for ESI, APCI, APPI, etc., type sources. This soft ionization generates for most small molecules only molecular ions, and no or very little fragmentation. The signal based only on molecular weight can be utilized for quantitation, but provides only the value of M as a unique parameter related to a molecule chemical composition. To improve this level of information, and also for obtaining other advantages such as a better signal to noise (S=N), tandem mass spectrometers are utilized. Such instruments will have after the first mass filter a collision-induced dissociation (CID) cell. In the collision cell, the ions from the first mass filter (e.g., a quadrupole Q1) are accelerated by applying an electrical potential to increase the ion kinetic energy and then allowed to collide with the molecules of a gas such as nitrogen, argon, helium, or even a mixture such as nitrogen þ helium. In the collision, some of the kinetic energy is converted into internal energy leading to bond breakage and fragmentation of the molecular ion into fragments. These fragment ions are analyzed by a second mass analyzer. Depending on the energy involved in the collision cell, these can be classified in low energy CID (with energy around 1 keV) and high energy CID (with energy between 1 and 20 keV). The two types of CID generate different fragmentation, with stronger fragmentation as the energy is higher and as the pressure of the collision gas is higher. The collision cell is an important component of tandem mass spectrometer [102]. Various constructions are known for a collision cell, in many instruments the collision cell being a quadrupole where the incoming ions are accelerated and the collision gas is present at a specific pressure, followed by the expulsion of fragment ions toward the second mass analyzer. Other types of collision cells are also known, such as hexapole or octopole. Depending on the instrument model, several instrument parameters affecting fragmentation can be selected. Among those are collision energy, collision cell entrance and exit potential, collision gas pressure, nature of collision gas, etc. For some instruments such as for orbitrap mass analyzers or for FT-ICR mass analyzers, special types of collision cell are utilized. The role of collision cells of breaking the molecular ion (parent ion) in fragments

Instrumentation in high-performance liquid chromatography

135

can be substituted with other forms of dissociation such as electron transfer dissociation (ETD), electron capture dissociation (ECD), or electron activated dissociation (EAD). These types of dissociation are used, for example, in proteomics to generate different types of dissociation compared to those obtained in standard CID low energy. The process in ETD consists in generating separately singly charged anions (e.g., from fluoranthene) that are transferred in an ion trap where the analyte such as a multiply charged peptide cations are present. The transferring of the electrons to the analyte that become one less positive charged also induces fragmentation of the peptide backbones in a specific manner that further help the protein structure elucidation. In EAD, a low energy beam of electrons is used for fragmentation in a reaction cell that replaces the CID [103]. In mass analyzers, besides the main components including ion sources, mass analyzers, detectors, other necessary components are the ion guides that are part of the ion optics. Various types of ion guides are utilized in mass spectrometers, such as multipole ion guides, C-trap used for orbitrap, etc. However, the ion optics of a mass spectrometer is not subject of modification by the user.

Ion detection in MS After the separation by M=z, the detection of ions can be done using different procedures. Two classes of detectors are known: point ion detectors, which detect the arrival of all ions sequentially at one point, and array detectors, which detect all ions simultaneously along a surface. These detectors record the number (abundance) of individual ions at each M=z. A few different types of point ion detectors are known, such as Faraday cup, electron multiplier, and scintillator. The array detectors are commonly made of separate point detectors (of miniature dimensions) called dynodes, clustered together in the area exposed to the incoming ions. The energy of ions impact is dissipated by ejection of electrons from the dynode material, creating an electrical charge. Additional electrons are ejected by a cascade process through subsequent dynodes. The array detectors have the advantage of collecting simultaneously the signal for a series of ions, but their dynamic range and sensitivity are usually lower than that of point detectors. The signal from the detector is further processed (amplified and analyzed) by an electronic data system. Problems such as mass/time calibration, linearity of the detector response, etc., are important parameter to consider related to a detector. The typical response S for the mass spectrometer detector is related to the instantaneous mass q of the analyte entering the detector by the expression: S ¼ a$q þ b

(3.4.10)

In Eq. 3.4.10, parameters a and b are dependent on detector type and detector settings. The mass of the analyte depends on the concentration of the analyte in the injected sample in the HPLC, and of the injected volume Vinj . As a result, chromatographic peaks are generated with peak areas that depend on total amount of analyte injected in the mobile phase. By setting a specific injected volume Vinj , both the sample

136

Method Development in Analytical HPLC

concentration and the amount of analyte can be measured. In certain applications and for concentration within a wider range, the response S may not depend linear on instantaneous concentration and quadratic dependencies are common.

Types of mass spectrometric instruments A variety of constructions of mass spectrometers designed for the detection and measurement of components of the eluate from the separation module of an HPLC are available. The simpler models are LC-MS instruments that include an ion source, a mass analyzer (mass filter), and a detector. The LC-MS instruments can work in either full scan, or in SIM mode. The full scan can detect a range of ions generated in the ion source and this can provide some structural information on the analyte. SIM mode is used mainly for quantitation purposes. From LC-MS level, more complex instruments are available such as the tandem MS/MS systems. A common tandem mass spectrometer type is the triple quadrupole. In a triple quadrupole, the first quadrupole (Q1) has the role of separating the (precursor or parent) ions generated in the ion source. These molecular ions are selected for further interactions in the collision cell (that is also a quadrupole Q2) where they can be fragmented. The third quadrupole Q3 is used for the separation of the resulting product ions following the fragmentation from the collision cell. After Q3, the ions are detected by procedures similar to those used in LC-MS. An MS/MS analyzer can work in various modes such as: (1) product ion scan, when the whole range of ions generated in Q3 by an ion selected by Q1 is analyzed, (2) precursor ion scan, when only one ion is selected for the detector by Q3, while Q1 is scanning the whole range of ions produced by the source, (3) neutral loss scan, when the instrument scans for a specific mass difference between the ions from Q1 and Q3, and (4) multiple reaction monitoring (or MRM) where a specific precursor ion is selected by Q1 and a specific fragment (product ion) is detected by Q3 (more than one pair of ions can be analyzed by MRM at the same time). This mode of operation is very common for quantitative analysis. The instruments have also the capabilities to scan all the ions generated in Q1, or work in SIM mode for a number of selected ions in Q1, etc. Qualitative information from LC-MS/MS can be generated based on fragmentation of the parent ions. Quantitative information using the MRM mode is characterized by exceptional sensitivity (as low as fmol/mL concentration in the analyzed solution in some cases). The fragmentation in LC-MS/MS is characteristic for each molecule but depends on the collision cell conditions (gas nature and pressure, collision energy, collision cell construction). For this reason, the nature of fragment ions in LC-MS/MS is typically used for confirmation purposes, and provides less help regarding compound identification as it is done in GC-MS. Both CID typical fragmentation as well as special types of fragmentation (ETD, ECD, EAD) are used in proteomics and with such fragmentation systems more qualitative information can be obtained. Another common type of tandem mass spectrometer is the quadrupole time-offlight (QTOF) type instrument. In a QTOF instrument, the ions from the source are guided into the quadrupole system for isolation (MS) and further into a collision cell (typically another quadrupole). The molecular fragments generated from the

Instrumentation in high-performance liquid chromatography

137

collision cell as a permanent stream of ions are than divided in discrete packages of ions and emitted into the flight tube of a TOF. QTOF-MS instruments can utilize the capability of high resolution. Instruments in which the ions generated in the source are entering directly a trap collision cell type followed by a TOF ion separation component are also used in particular for the analysis of very large molecules of biological importance [104]. This type of system takes advantage of the capability of TOF instruments to separate large molecules, and also can modulate fragmentation based on the conditions in the collision cell. Some mass spectrometers containing an orbitrap mass analyzer are commercialized under the name Orbitrap. For example, the Orbitrap Tribrid type instruments contain besides the ionization source and collision cell, a quadrupole, a C-trap, an orbitrap mass analyzer, and a linear ion trap. Such instruments acting as tandem MS/MS instruments are capable of excellent sensitivity and generate very high-resolution mass spectra. More details about the ion sources, ion optics, detectors, or data processing systems of different mass spectrometers can be found in the literature dedicated to mass spectrometry (e.g., Ref. [105]).

Ion mobility hyphenated with mass spectrometry Additional separation capability for an HPLC analysis can be obtained by using an electro separation technique named ion mobility (IM) between the chromatographic column of an HPLC and the MS detection [106,107]. IM can also be used as an independent separation technique, but more frequently is used in an HPLC-IM-MS or HPLC-IM-MS/MS system. In IM, the ions generated from an ion source are sent into a drift tube against the flow of a gas (buffer gas). The buffer gas can be He, N2, or other common gases and can have ambient pressure or reduced pressure (a few Torr). The time taken by the ions to travel the length of the drift tube (drift time) will be different depending on the ion dimensions. Due to the collisions with the buffer gas, molecules with larger dimensions will move slower than the smaller molecules leading to their separation. The separation in IM occurs at a time scale of milliseconds (in the range 10e1 to 10e3 s). Because in TOF mass spectrometers, the detection occurs on a microseconds scale making, the MS-TOF a suitable detection procedure for the ions separated by IM. Although the ions are continuously generated in an ion source, a gating mechanism is used in IM such that an ion pulse of precise width is repeatedly admitted into the drift chamber. This pulsed flux of ions fit the detection in a TOF mass analyzer that also works with ion pulses. A schematic diagram of an IM unit is indicated in Fig. 3.4.7. There are several procedures to move the ions along the drift chamber that differentiates the types of IM such as drift time (DT-IMS), traveling wave (TW-IMS), field asymmetric (FAIMS), differential (DMS), etc. [108,109]. In DT-IM, the ions are moved in one direction in a linear electric field gradient through the buffer gas. The electric field gradient is generated using a number of electrodes. In TW-IM, the ions are moved in one direction by a transport potential wave. This transport potential is in the shape of a well that travels along the drift chamber. The ions generated by the ion gate are trapped in the potential well and move along the drift tube in the

138

Method Development in Analytical HPLC

Figure 3.4.7 Schematic diagram of an ion mobility unit.

Buffer gas inlet Electric field

Ions

Ion gate +

+

+

+

Ions

Gas

Detector

Drift chamber Differential pumping

opposite direction of the flow of the stream of neutral molecules (buffer gas). By properly adjusting the “potential depth” of the well, some molecules are transported less efficiently than others by “skipping” from one well into the previous one following collision with the buffer gas molecules. In this way, molecular species can be separated based on their mass, charge, size, and shape, depending on the interactions with the buffer gas (e.g., collisions that have higher probability for larger molecules). Higher pressures of the buffer gas led to a higher separation selectivity due to a higher rate of ionemolecule interactions (collisions). A more complicated separation system is used in FAIMS, or DMS. Both positive and negative working mode for the separation can be selected, depending on the charge of the ions of interest. After the IM unit, the ions are typically injected in a collision cell followed by a TOF-MS analyzer.

Utilization of mass spectrometric detection in HPLC Mass spectrometric detection is very common in HPLC. A common utilization is in quantitative analysis. The expression 3.4.10 is frequently used for the measurement of concentration of the analytes, and based on the relation: q ¼ ½XVinj

(3.4.11)

where Vinj is the injected volume in the HPLC system, Eq. 3.4.10 can be written in the form: 0

S ¼ a $½X þ b

(3.4.12)

In some instances, the linearity of the signal as a function of concentration is affected by different processes taking place in the mass analyzer and a better description of the dependence is provided by a quadratic equation. The use of MS is also very useful when an isotopically labeled internal standard is utilized for quantitation. An isotopically labeled internal standard offers the possibility to have as internal standard a compound with very close properties as the analyte, but differing by the Mw. Only a

Instrumentation in high-performance liquid chromatography

139

mass detector can differentiate the analyte and the standard, while other detectors such as UV, FL, RI, and electrochemical do not have this capability. Sensitivity of mass spectrometric detection is usually very high (see also Table 4.4.1). This sensitivity can be applied for the measurement of very low levels of analytes. As an example, the analysis of vitamin C can be performed on a sample containing as low as 0.1 ng/mL analyte. Such analysis was performed using LCMS/MS with ESI type ionization and separation on an InertSustain AQ-C18 column 150  3 mm, 3 m m particles and with detection using MRM in negative mode for the transition 175.1 / 115.0. The chromatogram showing the peak of ascorbic acid is given in Fig. 3.4.8 [110]. Various procedures are used in practice to enhance as much as possible sensitivity for MS and MS/MS detection. Among these are the use of different additives (as indicated, for example, by the addition of HCOOH in the mobile phase), or by having the analytes already ionized in the mobile phase. In cases of the separations that require a mobile phase that does not contain water, APCI type ionization must be utilized. However, when ESI ionization can be utilized, it typically provides better sensitivity than APCI. As previously indicated, the flow rate utilized for an HPLC system influences the sensitivity of the MS, by affecting ionization efficiency and the residence time of the analytes in the detector. Since the flow rate is related to the type and dimensions of the chromatographic column, in the HPLC practice, the column selection is related to the resulting sensitivity. Smaller columns with small particles (e.g., dp ¼ 1.7 m m) being typically used with lower flow rates may be useful for increasing sensitivity [111]. Mass spectrometry can also be used for qualitative analysis, starting with verification of the presence of a specific molecule and in some cases even with the

20000

Simulated matrix

18000

Internal standard (fumaric acid)

16000

Intensity, cps

14000

Ascorbic acid

12000

3.094

10000

8000 6000 4000 2000 0

0.4

0.8

1.2

1.6

2.0

2.4

2.8

3.2

3.6

4.0 4.4 Time, min

4.8

5.2

5.6

6.0

6.4

6.8

7.2

7.6

8.0

Figure 3.4.8 Chromatogram generated for the extracted ion 115 by an LC-MS/MS negative ionization procedure from a standard containing 0.1 ng/mL (100 pg/mL) ascorbic acid and 5 mL injection volume.

140

Method Development in Analytical HPLC

identification of unknown compounds and structural elucidation. A wide variety of procedures are used in MS to facilitate qualitative analysis, starting with simple value of the Mw of the analyte that is typically only helping but not being diagnostic in qualitative analysis, continuing in the case of MS/MS to the utilization of various ions in a fragmentation or of the entire spectrum characteristics for compound identification, and extending to other various techniques such as changing fragmentation depending on collision cell parameters, use of different collision gas molecules for collisioninduced dissociation (e.g., Ref. [112]), or using D2O infusion in the mobile phase of the HPLC and detecting the molecules showing isotopic exchange [113]. The mass range M=z that can be detected by the MS and MS/MS instruments depends on the instrument model. For example, Sciex Triple Quad 7500 system has a range between 50 and 2000 Th. Other instruments have a different mass range, and the system AB 5000 (from Sciex) has the range between 5 and 1250 Th, the Acquity QDa MS system from Waters has the mass range between 30 and 1250 Th, and the Agilent 6475 Triple Quadruple instruments have a mass range between 5 and 3000 Th. LC-MS/MS systems with larger upper limit of M=z are also available. In general, at lower values of the M=z, the detection shows more noise due to the ionization of the mobile phase components, and typically molecules with M=z higher than 60e70 Th are recommended to be analyzed by LC-MS systems (small molecules may also be volatile and easily analyzed by GC-MS). Regarding the upper limit of M= z, the instruments can measure ions with higher value for M when z ¼ 2, 3, etc., such in the case of multicharged ions generated from peptides.

Key points • • • • • •

Mass spectrometry is a common type of detection in HPLC. Mass spectrometry instrumentation covers a range of instruments from simple to very complex instruments. The mass accuracy achieved by MS instruments starts from unit mass resolution to high and very high mass resolution (R from 1000 to 2,000,000). Mass spectrometry main use is for quantitative purposes, but compound identification and structure elucidation can also be provided in many applications. The mass range of detection for MS instruments depends on the instrument model, but larger molecular ions than the instrument limit for M can be measured when the generated ions have multiple charges. When water is present in the mobile phase, ESI type ionization is preferred to APCI ionization.

References [1] E.M. Sheldon, Development of a LC-LC-MS complete heart-cut approach for the characterization of pharmaceutical compounds using standard instrumentation, J. Pharm. Biomed. Anal. 31 (2003) 1153e1166.

Instrumentation in high-performance liquid chromatography

141

[2] D.R. Stoll, X. Li, X. Wang, P.W. Carr, S.E.G. Porter, S.C. Rutan, Fast, comprehensive two-dimensional liquid chromatography, J Chromatogr. A 1168 (2007) 3e43. [3] J. Pol, M. Kivilompolo, T. Hyötyl€ainen, Comprehensive two-dimensional liquid chromatography (LCLC): a review, LC-GC Eur. 24 (2011) 232e237. [4] W.-Y. Lee, T.A. Nieman, Determination of dansyl amino acids using tris(2,20 -bipyridyl) ruthenium (II) chemiluminescence for post-column reaction detection in highperformance liquid chromatography, J. Chromatogr. A 659 (1994) 111e118. [5] A. Tsiasioti, C.K. Zacharis, P.D. Tzanavaras, Development and evaluation of pulsed post column derivatization in liquid chromatography as a concept to minimize reagent consumption, J. Chromatogr. A 1690 (2023) 463791. [6] C.K. Zacharis, P.D. Tzanavaras, Liquid chromatography coupled to on-line post column derivatization for the determination of organic compounds: a review on instrumentation and chemistries, Anal. Chim. Acta 798 (2013) 1e24. [7] K.B. Lynch, A. Chen, S. Liu, Miniaturized high-performance liquid chromatography instrumentation, Talanta 177 (2018) 94e103. [8] M. Szumski, B. Buszewski, Miniaturization in separation techniques, in: B. Buszewski, I. Baranowska (Eds.), Handbook of Bioanalytics, Springer, Cham. Edinburgh, UK, 2022, https://doi.org/10.1007/978-3-030-63957-0_32-1. [9] A. Medvedovici, V. David, Pump/solvent delivery system design for HPLC, in: Encyclopedia of Chromatography, third ed., Taylor & Francis, New-York, 2009, pp. 1946e1962. [10] Dionex Eluent Generator Cartridges, Product Manual, Thermo Scientific, P/N 065018-05, 2014. [11] J. De Vos, K. Broeckhoven, S. Eeltink, Advances in ultrahigh-pressure liquid chromatography technology and system design, Anal. Chem. 88 (2016) 262e278. [12] https://www.sigmaaldrich.com/deepweb/assets/sigmaaldrich/marketing/global/documen ts/421/096/4643.pdf. [13] I. Ueta, Y. Saito, Needle extraction device, in: C.F. Poole (Ed.), Solid-Phase Extraction, Elsevier, Amsterdam, 2020, pp. 429e442. [14] S.C. Moldoveanu, V. David, Modern Sample Preparation for Chromatography, second ed., Elsevier, Amsterdam, 2021. [15] C.J. Welch, X. Gong, W. Schafer, E.C. Pratt, T. Birkovic, Z. Pirzada, J.F. Cuff, B. Kosjek, MISER chromatography (multiple injections in a single experimental run): the chromatogram is the graph, Tetrahedron: Assim. 21 (2010) 1674e1681. [16] C.V. McNeff, B. Yan, D.R. Stoll, R.A. Henry, Practice and theory of high temperature liquid chromatography, J. Sep. Sci. 30 (2007) 1672e1685. [17] G. Vanhoenacker, P. Sandra, High temperature and temperature programmed HPLC: possibilities and limitations, Anal. Bioanal. Chem. 390 (2008) 245e248. [18] T. Teutenberg, High temperature liquid chromatography e a brief review about an emerging technique, Chromatography Today (August/September 2010). [19] https://www.waters.com/waters/en_US/ionKey-MS—microflow-UPLC-Seperation-with -iKey/nav.htm?cid¼134782630&locale¼en_US. [20] B. Bao, Z. Wang, D. Thushara, A. Liyanage, S. Gunawardena, Z. Yang, S. Zhao, Recent advances in microfluidics-based chromatography - a mini review, Separations 8 (2021) 3. [21] D.A. Vargas Medina, N.G. Pereira dos Santos, J.S. da Silva Burato, J.V. Basolli Borsatto, F.M. Lanças, An overview of open tubular liquid chromatography with a focus on the coupling with mass spectrometry for the analysis of small molecules, J. Chromatogr. A 1641 (2021) 461989.

142

Method Development in Analytical HPLC

[22] D. Neue, HPLC Columns: Theory, Technology, and Practice, John Wiley & Sons, Hoboken, New Jersey, 1997. [23] A. Medvedovici, V. David, Spectrophotometry: diode array, in: P. Worsfold, C. Poole, A. Townshend, M. Miro (Eds.), Encyclopedia of Analytical Science, third ed. Vol 9, Elsevier, Amsterdam, 2019, pp. 221e235. [24] A.K. Gaigalas, L. Li, O. Henderson, R. Vogt, J. Barr, G. Marti, J. Weaver, A. Schwartz, The development of fluorescence intensity standards, J. Res. Nat. Inst. Standards Technol. 106 (2001) 381e389. [25] J. Dolan, How does it work? Part V: fluorescence detectors, LCGC North Am. 34 (9) (2016) 710e719. [26] D.C. Shelly, I.M. Warner, Chapter 3: fluorescence detectors in high-performance liquid chromatography, in: T.M. Vickrey (Ed.), Liquid Chromatography Detectors, CRC Press, Boca Raton, 1983, pp. 87e125. [27] Q. Xu, Y. Yin, Y. Liu, Q. Ma, X. Chen, J. Zhao, Simultaneous determination of six a-dicarbonyl compounds in traditional Chinese medicines using high-performance liquid chromatography-fluorescence detector with pre-column derivatization, J. Sep. Sci. 46 (2023) 2300435. [28] S. Moldoveanu, V. David, Essentials in Modern HPLC Separations, second ed., Elsevier, Amsterdam, 2022. [29] M. Tsunoda, Chemiluminiscence detection with separation techniques for bioanalytical applications, Bioanal. Rev. 1 (2009) 25e34. [30] J.L. Adcock, J.M. Terry, C.J. Barrow, N.W. Barnett, D.C. Olson, P.S. Francis, Chemiluminescence detectors for liquid chromatography, Drug Test. Anal. 3 (2011) 139e144. [31] F. Zhou, H. Zhao, K. Chen, S. Cao, Z. Shi, M. Lan, Flexible electrochemical sensor with Fe/Co bimetallic oxides for sensitive analysis of glucose in human tears, Anal. Chim. Acta 1243 (2023) 340781. [32] B.E. Erickson, Product review: electrochemical detectors for liquid chromatography, Anal. Chem. 72 (2000) 353Ae357A. [33] C. Wang, J. Xu, G. Zhou, Q. Qu, G. Yang, X. Hu, Electrochemical detection coupled with high-performance liquid chromatography in pharmaceutical and biomedical analysis: a minireview, Comb. Chem. High Throughput Screen. 10 (2007) 547e554. [34] J.-M. Kauffmann, N.K. Bakirhan, B. Bozal-Palabiyik, B. Uslu, R. Rodríguez-G omez, M. Vandeput, S.A. Ozkan, Electrochemical detectors in liquid chromatography: recent trends in pharmaceutical and biomedical analysis, Curr. Med. Chem. 25 (2018) 4050e4065. [35] https://chemistry.stackexchange.com/questions/144360/molar-conductivities-of-aqueous -electrolytes-at-1-m. [36] https://www.thermofisher.com/us/en/home/industrial/chromatography/ion-chromatograp hy-ic/ic-consumables/ic-rfic-suppressors.html. [37] J.S. Fritz, D.T. Gjerde, Ion Chromatography, Wiley-VCH, Weinheim, 2009. [38] C.A. Pohl, J.R. Stillian, P.E. Jackson, Factors controlling ion-exchange selectivity in suppressed ion chromatography, J. Chromatogr. A 789 (1997) 29e41. [39] P.K. Dasgupta, F. Maleki, On exchange membranes in ion chromatography and related applications, Talanta 204 (2019) 89e137. [40] J. Xu, Y. Wang, J. Jiang, X. Li, Y. Xu, W. Song, Quantification of underivatized amino acids in solid beverages using high-performance liquid chromatography and a potentiometric detector, J. Chromatogr. A 1696 (2023) 463986.

Instrumentation in high-performance liquid chromatography

143

[41] H. Matsuda, Zur theorie der station€aren strom-spannunos-kurven von redoxelektrodenreaktionen in hydrodynamischer voltammetrie: II. laminare rohr-und kanalstkömungen, J. Electroanal. Chem. Interf. Elect. 15 (1967) 325e336. [42] S. Moldoveanu, J.L. Anderson, Amperometric response of a rectangular channel electrode, J. Electroanal. Chem. Interf. Elect. 175 (1984) 67e77. [43] S. Moldoveanu, G.S. Handler, J.L. Anderson, On convective mass transfer in laminar flow between two parallel electrodes in a rectangular channel, J. Electroanal. Chem. Interf. Elect. 179 (1984) 119e130. [44] J. Yamada, H. Matsuda, Limiting diffusion currents in hydrodynamic voltammetry: III. Wall jet electrodes, J. Electroanal. Chem. 44 (1973) 189e198. [45] S. Moldoveanu, J.L. Anderson, Numerical simulation of convective diffusion at a microarray channel electrode, J. Electroanal. Chem. 185 (1985) 239e252. [46] A. Natale, D. Nardiello, C. Palermo, M. Muscarella, M. Quinto, D. Centonze, Development of an analytical method for the determination of polyphenolic compounds in vegetable origin samples by liquid chromatography and pulsed amperometric detection at a glassy carbon electrode, J. Chromatogr. A 1420 (2015) 66e73. [47] C.F.B. Coutinho, L.F.M. Coutinho, L.H. Mazo, S.L. Nixdorf, C.A.P. Camara, Rapid and direct determination of glyphosate and aminomethyl-phosphonic acid in water using anion-exchange chromatography with coulometric detection, J. Chromatogr. A 1208 (2008) 246e249. [48] T.H. Mourey, L.E. Oppenheimer, Principles of operation of an evaporative lightscattering detector for liquid chromatography, Anal. Chem. 56 (1984) 2427e2434. [49] S. Heron, M.-G. Maloumbi, M. Dreux, E. Verette, A. Tchapla, Method development for a quantitative analysis performed without any standard using an evaporative lightscattering detector, J. Chromatogr. A 1161 (2007) 152e156. [50] B. Mathews, P. Higginson, R. Lyons, J.C. Mitchel, N.W. Sach, M.J. Snowden, M.R. Taylor, A.G. Wright, Improving quantitative measurements for the evaporative light scattering detector, Chromatographia 60 (2004) 625e633. [51] J. Wang, S. Tang, K. Zhang, Quantitation of polyethylene glycol by size exclusion chromatography with charged aerosol, differential refractive index, and multi-angle light scattering detectors, J. Pharm. Biomed. Anal. 238 (2024) 115854. [52] F. Weber, L. Rahnfeld, P. Luciani, Analytical profiling and stability evaluation of liposomal drug delivery systems: a rapid UHPLC-CAD-based approach for phospholipids in research and quality control, Talanta 220 (2020) 121320. [53] T. Gorecki, F. Lynen, R. Szucs, P. Sandra, Universal response in liquid chromatography using charged aerosol detection, Anal. Chem. 78 (2006) 3186e3192. [54] Y. Jiang, D. Hascall, D. Li, J.H. Pease, A charged aerosol detector/chemiluminescent nitrogen detector/liquid chromatography/mass spectrometry system for regular and fragment compound analysis in drug discovery, J. Chromatogr. A 1411 (2015) 63e68. [55] E.M. Fujinari, L.O. Courthaudon, Nitrogen-specific liquid chromatography detector based on chemiluminescence: application to the analysis of ammonium nitrogen in waste water, J. Chromatogr. A 592 (1992) 209e214. [56] L.K. Sutton, J.A. Caruso, Liquid chromatography-inductively coupled plasma mass spectrometry, J. Chromatogr. 856 (1999) 243e258. [57] T. Van Acker, E. Bolea-Fernandez, F. Vanhaecke, G. Koellensperger, Inductively coupled plasma mass spectrometry, Nat. Rev. Met. Prim. 3 (2023) 52. [58] G.W. Somsen, C. Gooijer, N.H. Velthorst, U.A.T. Brinkman, Coupling of column liquid chromatography and Fourier-transform infrared spectrometry, J. Chromatogr. A 811 (1998) 1e34.

144

Method Development in Analytical HPLC

[59] M.A. Mottaleb, Coupling the high-performance liquid chromatography to Fouriertransform infrared spectrometry using a heated gas flow modified thermospray, Anal. Sci. 15 (1999) 57e62. [60] T.D.N. Hong, M. Jouan, N.Q. Dao, M. Bouraly, F. Mantisi, Coupling of highperformance liquid chromatography with Raman spectrometry, J. Chromatogr. A 743 (1996) 323e327. [61] Y.-H. Lo, H. Hiramatsu, Online liquid chromatographyeRaman spectroscopy using the vertical flow method, Anal. Chem. 92 (2020) 14601e14607. [62] T. Gebretsadik, W. Linert, M. Thomas, T. Berhanu, R. Frew, LCeNMR for natural product analysis: a journey from an academic curiosity to a robust analytical tool, Science 3 (2021) 6. [63] A.B. Schefer, U. Braumann, L.-H. Tseng, M. Spraul, M.G. Soares, J.B. Fernandes, M.F.G.F. da Silva, P.C. Vieira, A.G. Ferreira, Application of high-performance liquid chromatographyenuclear magnetic resonance coupling to the identification of limonoids from mahogany tree (Switenia macrophylla, Meliaceae) by stopped-flow 1D and 2D NMR spectroscopy, J. Chromatogr. A 1128 (2006) 152e163. [64] Y. Luo, L. Wu, B. Yang, Y. Jin, K. Zheng, Z. He, A novel potential primary method for quantification of enantiomers by high performance liquid chromatography-circular dichroism, Sci. Rep. 8 (2018) 7390. [65] Y. Zhang, S. Shi, J. Guo, Q. You, D. Feng, On-line surface plasmon resonance-high performance liquid chromatographyetandem mass spectrometry for analysis of human serum albumin binders from Radix Astragali, J. Chromatogr. A 1293 (2013) 92e99. [66] G.E. Fredheim, S.M. Braaten, B. E Christensen, Molecular weight determination of lignosulfonates by size-exclusion chromatography and multi-angle laser light scattering, J. Chromatogr. A 942 (2002) 191e199. [67] K.-L. Cheong, D.-T. Wu, J. Zhao, S.-P. Li, A rapid and accurate method for the quantitative estimation of natural polysaccharides and their fractions using high performance size exclusion chromatography coupled with multi-angle laser light scattering and refractive index detector, J. Chromatogr. A 1400 (2015) 98e105. [68] S. Kreimer, I.S. Krull, A. Rathore, The use of light-scattering detection with SEC and HPLC for protein and antibody studies, Part I: background, theory, and potential uses, LCGC North Am. 30 (9) (2012) 842e849. [69] L. Gentiluomo, V. Schneider, D. Roessner, W. Friess, Coupling multi-angle light scattering to reverse-phase ultra-high-pressure chromatography (RP-UPLC-MALS) for the characterization monoclonal antibodies, Sci. Rep. 9 (2019) 14965. [70] E.M. Fujinari, J. Damon Manes, Nitrogen-specific liquid chromatography detection of nucleotides and nucleosides by HPLC-CLND, Develop. Food Sci. 37 (1995) 379e396. [71] W. Schrader, H.-W. Klein, Liquid chromatography/Fourier transform ion cyclotron resonance mass spectrometry (LC-FTICR MS): an early overview, Anal. Bioanal. Chem. 379 (2004) 1013e1024. [72] S. Banerjee, S. Mazumdar, Electrospray ionization mass spectrometry: a technique to access the information beyond the molecular weight of the analyte, Intern. J. Anal. Chem. 2012 (2012) 282574. [73] T.R. Covey, R.F. Bonner, B.I. Shushan, J. Henion, The determination of protein, oligonucleotide and peptide molecular weights by ion-spray mass spectrometry, Rapid Commun. Mass Spec. 2 (1988) 249e256. [74] M. Mann, C.K. Meng, J.B. Fenn, Interpreting mass spectra of multiply charged ions, Anal. Chem. 61 (1989) 1702e1708.

Instrumentation in high-performance liquid chromatography

145

[75] M. Oss, A. Kruve, K. Herodes, I. Leito, Electrospray ionization efficiency scale of organic compounds, Anal. Chem. 82 (2010) 2865e2872. [76] G. Taylor, Disintegration of water drops in an electric field, Proc. Roy. Soc. London A 280 (1964) 383e397. [77] L. Tang, P. Kebarle, Effect of the conductivity of the electrosprayed solution on the electrospray current. Factors determining analyte sensitivity in electrospray mass spectrometry, Anal. Chem. 63 (1991) 2709e2715. [78] https://www.waters.com/webassets/cms/library/docs/720000584en.pdf. [79] C.E.C.A. Hop, Generation of high molecular weight cluster ions by electrospray ionization implications for mass calibration, J. Mass Spectrom. 31 (1996) 1314e1316. [80] A.P. Bruins, T.R. Covey, J.D. Henion, Ion spray interface for combined liquid chromatography/atmospheric pressure ionization mass spectrometry, Anal. Chem. 59 (1987) 2642e2646. [81] M.G. Ikonomou, A.T. Blades, P. Kebarle, Electrospray ion spray: a comparison of mechanisms and performance, Anal. Chem. 63 (1991) 1989e1998. [82] J. Hermans, S. Ongay, V. Markov, R. Bischof, Physicochemical parameters affecting the electrospray ionization efficiency of amino acids after acylation, Anal. Chem. 89 (2017) 9159e9166. [83] X. Han, K. Yang, J. Yang, K.N. Fikes, H. Cheng, R.W. Gross, Factors influencing the electrospray intrasource separation and selective ionization of glycerophospholipids, J. Am. Soc. Mass Spectrom. 17 (2006) 264e274. [84] S. Lupo, LC-MS sensitivity: practical strategies to boost your signal and lower your noise, LCGC North Am. 36 (9) (2018) 652e660. [85] T.R. Covey, A.P. Bruins, J.D. Henion, Comparison of thermospray and ion spray mass spectrometry in an atmospheric pressure ion source, Org, Mass Spectrom 23 (1988) 178e186. [86] J.F. Banks, J.P. Quinn, C.M. Whilehouse, LC/ESI-MS determination of proteins using conventional liquid chromatography and ultrasonically assisted electrospray, Anal. Chem. 66 (1994) 3688e3695. [87] M. Wilm, M. Mann, Analytical properties of the nanoelectrospray ion source, Anal. Chem. 68 (1996) 1e8. [88] E.C. Horning, M.G. Horning, D.I. Carroll, I. Dzidic, R.N. Stillwell, New picogram detection system based on a mass spectrometer with an external ionization source at atmospheric pressure, Anal. Chem. 45 (1973) 936e943. [89] F.J. Andrade, J.T. Shelley, W.C. Wetzel, M.R. Webb, G. Gamez, S.J. Ray, G.M. Hieftje, Atmospheric pressure chemical ionization source. 1. Ionization of compounds in the gas phase, Anal. Chem. 80 (2008) 2646e2653. [90] A. Pape, O.J. Schmitz, Dielectric barrier discharge in mass spectrometry e an overview over plasma investigations and ion sources applications, TrAC - Trends Anal. Chem. 170 (2023) 117420. [91] B. Seemann, T. Alon, S. Tsizin, A.B. Fialkov, A. Amirav, Electron ionization LC-MS with supersonic molecular beams-the new concept, benefits and applications, J. Mass Spectrom. 50 (2015) 1252e1263. [92] A. Cappiello, G. Famiglini, P. Palma, A. Siviero, Liquid chromatographyeelectron ionization mass spectrometry: fields of application and evaluation of the performance of a direct-EI interface, Mass Spectrom. Rev. 24 (2005) 978e989. [93] J.H. Gross, Direct analysis in real time - a critical review on DART-MS, Anal. Bioanal. Chem. 406 (2014) 63e80. [94] A. Brookhart, M. Arora, M. McCullagh, I.D. Wilson, R.S. Plumb, J.P.C. Vissers, N. Tanna, Understanding mobile phase buffer composition and chemical structure effects

146

[95] [96] [97]

[98] [99] [100] [101] [102]

[103]

[104] [105] [106] [107]

[108] [109] [110] [111]

[112]

[113]

Method Development in Analytical HPLC

on electrospray ionization mass spectrometry response, J. Chromatogr. A 1696 (2023) 463966. S.C. Moldoveanu, H. Gan, Comparison of two methods for ginsenosides quantitation, J. Sep. Sci. 46 (2023) e2201063. W.P. Duncan, P.D. Perkins, LC-MS with simultaneous electrospray and atmospheric pressure chemical ionization, Am. Lab. 37 (6) (2005) 28e33. T. Galaon, C. Vacaresteanu, D.F. Anghel, V. David, Simultaneous ESI-APCI(þ) ionization and fragmentation pathways for nine benzodiazepines and zolpidem using single quadrupole LC-MS, Drug Test. Anal. 6 (2014) 439e450. R.E. March, An introduction to quadrupole ion trap mass spectrometry, J. Mass Spectrom. 32 (1997) 351e369. X. Zhou, Z. Ouyang, Flowing gas in mass spectrometer: method for characterization and impact on ion processing, Analyst 139 (2014) 5215e5222. R.A. Zubarev, A. Makarov, Orbitrap mass spectrometry, Anal. Chem. 85 (2013) 5288e5296. J. Romson, A. Emmer, Mass calibration options for accurate electrospray ionization mass spectrometry, Intern. J. Mass Spectrom. 467 (2021) 116619. A.R. Johnson, E.E. Carlson, Collision-induced dissociation mass spectrometry: a powerful tool for natural product structure elucidation, Anal. Chem. 87 (2015) 10668e10678. T. Baba, P. Ryumin, E. Duchoslav, K. Chen, A. Chelur, B. Loyd, I. Chernushevich, Dissociation of biomolecules by an intense low-energy electron beam in a high sensitivity time-of-flight mass spectrometer, J. Am. Soc. Mass Spectrom. 32 (2021) 1964e1975. D.R. Allen, B.C. McWhinney, Quadrupole time-of-flight mass spectrometry: a paradigm shift in toxicology screening applications, Clin. Biochem. Rev. 40 (2019) 135e146. E. de Hoffmann, V. Stroobant, Mass Spectrometry: Principles and Applications, third ed., John Wiley, Hoboken, 2007. A.B. Kanu, P. Dwivedi, M. Tam, L. Matz, H.H. Hill, Ion mobility-mass spectrometry, J. Mass Spectrom. 43 (2008) 1e22. R.O. Bussey III, Advantages of ion mobility couples with HPLC/UPLC, in: N. Perspectives, S.C. Moldoveanu, V. David (Eds.), Analytical Liquid Chromatography, IntechOpen, London, 2022. J.C. May, J.A. McLean, Ion mobility-mass spectrometry: time-dispersive instrumentation, Anal. Chem. 87 (3) (2015) 1422e1436. A.A. Shvartsburg, R.D. Smith, Fundamentals of traveling wave ion mobility spectrometry, Anal. Chem. 80 (2008) 9689e9699. S.C. Moldoveanu, Comparison of several HPLC methods for the analysis of vitamin C, Biomed. Chromatogr. 38 (2024) e5753. E.S. Grumbach, D.M. Diehl, U.D. Neue, The application of novel 1.7m ethylene bridged hybrid particles for hydrophilic interaction chromatography, J. Sep. Sci. 31 (2008) 1511e1518. R.B. Cole, J.-C. Tabet, Stereospecific ionemolecule reactions of nucleophilic gas-phase reagents with protonated bifunctional tetracyclic terpene epimers in the triple quadrupole collision cell, J. Mass Spectrom. 32 (1997) 413e419. H. Thakkar, S. Jain, S. Kumar, V.S. Bhalekar, S. Gangakhedkar, R.P. Shah, Hyphenated liquid chromatography - diode array detection - charged aerosol detection - high resolution - multistage mass spectrometry with online hydrogen/deuterium exchange: one stop solution for pharmaceutical impurity profiling, J. Chromatogr. A 1689 (2023) 463725.

4

Performance and utilization of high-performance liquid chromatography separation and detection 4.1

Parameters related to the characterization of the chromatographic peak

The quality of a high-performance liquid chromatography (HPLC) separation can be evaluated based on several parameters, some of these being related to the properties of the chromatographic peaks from the chromatogram. Present section describes some of those characteristics, and settings or selections performed on the HPLC instrument that affect the chromatographic peaks. These parameters can be used to describe the quality of a peak of interest and therefore are related to the quality of the results of a chemical analysis.

Flow rate of the mobile phase The pumps of the HPLC instrument deliver the mobile phase at a volumetric flow rate U specified by the operator (see Section 3.2). The volumetric flow rate U is the volume of fluid that flows per unit time (usually expressed in mL/min) through the chromatographic column. Besides the volumetric flow rate U, the mobile phase has also a linear flow rate u. The linear flow rate u is the velocity of a point in the fluid passing through the column (expressed as length per time), and can be considered the velocity of an unretained molecule in the chromatographic column. The linear flow rate depends on the volumetric flow rate and on the area Ac of the channel (in the chromatographic column) in which the flow takes place. This relation is expressed by the formula: u ¼ U=AC

(4.1.1)

The area Ac depends on both the column physical dimensions (surface area of the empty column) and also on the porosity ε of the column packing. For a cylindrical column with the inner diameter d, (usually expressed in mm) the surface area of the  empty column is Ac ¼ pd2 4 and as a result, the value of Ac is given by the expression: Ac ¼ ε

pd2 4

(4.1.2)

Method Development in Analytical HPLC. https://doi.org/10.1016/B978-0-443-29849-3.00006-3 Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

148

Method Development in Analytical HPLC

For a chromatographic column containing particles (spherical) with about 5 mm diameter, the column packing porosity ε can be taken with the approximate value ε z 0.7, although this value can be different from 0.7 and varies depending on the stationary phase particle size and structure such that for columns having particles with smaller size, the value of ε is smaller than 0.7 [1]. The relation between volumetric flow rate U (mL/min) and linear flow rate u (cm/min) is given by the formula (for d in cm): U¼

ε pd2 u 4

(4.1.3)

As further indicated, the flow rate is an important parameter which is set by the user of the HPLC instrument. The flow rate is related to retention time tR as well as with other parameters characterizing the HPLC separation and even with the detector response. Optimization of flow rate for an HPLC analysis is of interest in the development of a method, as it affects both the separation and the detection.

Retention time The peak retention time tR ðXÞ is the time (usually measured in min) from the sample injection into the chromatographic system to the time of elution of the compound X. The time is taken at the maximum (the apex) of the chromatographic peak. The value of retention time depends on the nature of the molecular species X, but also on the mobile phase composition, the type of chromatographic column (stationary phase, column dimensions, etc.), as well as on the conditions the separation is performed (flow rate U of the mobile phase, temperature). The retention time tR ðXÞ is an important chromatographic characteristic for a molecular species X, since it has a constant value as long as the other parameters of the chromatographic separation are kept constant. After establishing the retention time for a specific compound, for example, using a standard, the retention time can be used for the compound identification, if no changes in the chromatographic conditions are made, and no interference from other compounds occur in the separation. The retention time tR ðXÞ can be separated into two components: the time analyte X spends in the mobile phase moving through the column known as dead time or void time t0 , and the time the analyte is retained on the stationary phase t 0 R ðXÞ known as reduced retention time. The retention time is given by the expression: tR ðXÞ ¼ t 0 R ðXÞ þ t0

(4.1.4)

The retention time tR ðXÞ is determined by several factors including the nature of compound X, the mobile phase, and the chromatographic column used for the separation. This parameter also depends on the flow rate U in the chromatographic system, and the accuracy and precision of measurements of tR are influenced by the changes in flow rate provided by the pumping system. The value of tR increases with the decrease

Performance and utilization of HPLC

149

of the flow rate, and therefore the fluctuating values of U due to pump malfunction, such as problems with check valves, pump seals, leaks, or bubbles, may influence the accuracy and precision of this parameter. For example, common specifications for flow rate accuracy are 1% and precision given by the relative standard deviation (RSD%) of 0.07%, with variations between 0.02 min for binary high-pressure mixing of solvents systems. For quaternary low-pressure mixing systems, the specifications for flow rate are typically between 0.02 and 0.04 min because of proportioning errors between the utilized solvents [2]. The retention time in HPLC can be in a rather large range of values, for common chromatographic separations being between 1.0 and 25.0 min, but it can be even shorter (e.g., as low as 0.1 min) or longer (e.g., 40e50 min) in more special separations. The dead time t0 is not dependent on the compound X. It depends on column construction and on the flow rate of the mobile phase. The dependence of t0 on the linear flow rate u and the length L of the chromatographic column is given by the formula: t0 ¼ L=u

(4.1.5)

A small fraction from the value of t0 is the time taken for the sample to flow through the tubing from the injector to the column and from the column to the detector. However, for a length of 20 cm for tubing with 0.12 mm i.d. (0.005 in), at a flow rate of 1 mL/min, the tubing volume contributes with about 0.16 s delay which is usually neglected. The dependence of t0 on volumetric flow rate U is given by the formula: t0 ¼

ε pd 2 L 4U

(4.1.6)

The value for t0 can be obtained experimentally and a number of studies were reported in the literature regarding this subject [3e7]. An approximation of t0 can be obtained by measuring the time for the elution of a compound that is virtually not retained (very slightly retained, since it can be difficult to find a compound that is not retained at all on a chromatographic column). In reversed phase HPLC (RPHPLC), common compounds used for measuring t0 are uracil, thiourea, or some inorganic salts such as NaNO2 or NaNO3. The solvent used for injecting the sample (when it is different from the mobile phase) also can be such a compound, and the retention time of this solvent peak can be taken as dead time. Similarly, the retention time of a deuterated solvent analogous of a mobile phase component can be used for measuring t0 . One procedure uses the minor disturbances in the background signal created by the sample injection. Another more elaborate procedure involves the use of a plot of retention times for a homologous series of compounds that are retained less and less as the number of carbon atoms is decreasing and the plot is extrapolated to zero. A similar approach based on linear free energy relationships and the measurement of the elution volume of a series of alkylbenzene standards from toluene to heptadecylbenzene was used for estimating the void volume in HILIC-type HPLC for different mobile phase compositions [8,9]. The dead time t0

150

Method Development in Analytical HPLC

can also be obtained when the dead volume V0 is known (see further). Depending on column and on the flow rate U, the dead time can vary from 0.1 min to several min.

Run time The time utilized for a chromatographic separation is indicated as the run time trun (see Section 3.2). The run time can be selected by the operator and should be slightly longer than the retention time of the last peak in the chromatogram. It is common during the run time to have both the flow in the HPLC and the detection active for the same length of time. However, for intervals of time in a chromatographic separation where the eluted compounds are not of interest, the flow from the column can be diverted to waste, or the detector can be deactivated. In such situation, the length of the chromatogram will appear shorted than the run time. Also, in some chromatographic separations, it is necessary to keep the mobile phase flowing through the column for the purpose of column equilibration and not to perform a separation. The column equilibration time is not usually included as run time, but it contributes to the length necessary for running a sample. It would be recommended to indicate the sum of equilibration time and the run time as total run time (although this convention is not always utilized). When multiple samples are analyzed by an HPLC method, the run time plus the equilibration time are important parameters since they determine the length of time required to run a given number of samples. Shorter run time and equilibration time are leading to a faster analysis. Common run times vary depending on analytes, column, and mobile phase composition and flow rate U having values as short as 1 min up to 25e30 min for common separations, but can be longer in special cases.

Retention volume For a specific molecular species X, the retention volume VR ðXÞ is defined as the volume of mobile phase flowing from the time of injection until the corresponding retention time tR ðXÞ of a molecular species (VR is sometimes indicated as elution volume). The values for VR ðXÞ and tR ðXÞ are related by the simple formula: VR ðXÞ ¼ U tR ðXÞ

(4.1.7)

Similar to the retention time, the retention volume depends on the nature of the molecular species X, the nature of mobile phase, the type of chromatographic column (stationary phase, column dimensions, etc.), as well as on the conditions the chromatogram is performed (e.g., the mobile phase temperature). Also, similar to the retention time, VR ðXÞ can be separated into two components: the volume corresponding to the dead time t0 known as dead volume V0 , or void volume, and the mobile phase volume corresponding to the reduced retention time t 0 R indicated as reduced retention volume V 0 R defined by the formula: V 0 R ¼ VR  V0

(4.1.8)

Performance and utilization of HPLC

151

Neglecting the small volume of the tubing (including the dwell volume VD ), the dead volume V0 can be considered the volume of the liquid in the column when this is filled with stationary phase. The dead volume V0 of a chromatographic column can be calculated from the dead time based on the relation V0 ¼ U t0 but can also be found by direct measurements. For this purpose, a column is sequentially filled with two solvents of different densities (e.g., dichloromethane and tetrahydrofuran) and weighed (pycnometric method) [1,10]). The volume V0 is calculated based on the expression: V0 ¼

msolv1  msolv2 rsolv1  rsolv2

(4.1.9)

where msolv1 is the mass of column þ solvent1 (with density rsolv1 ), and msolv2 is the mass of column þ solvent2 (with density rsolv2 ). V0 such measured can be used to obtain t0 . The value of V0 accounts for both external porosity (interstitial voids in the column) and internal porosity (the volume of particle pores). As expected, V0 is proportional with the volume of the empty column Ve , the proportionality constant ε being the column packing porosity that also appeared in Eq. 4.1.2 and is usually taken as ε ¼ 0.7. For a column of length L and inner diameter d, the empty column volume is Ve ¼ ðp =4Þd2 L and when filled with the stationary phase, the relation between V0 and the volume of empty column Ve is given by the formula: V0 ¼ ε Ve ¼ ε ðp = 4Þd2 L

(4.1.10)

In Eq. 4.1.10, d is the column i.d. and L is the column length, which are typically expressed in mm, while the volume V0 is usually expressed in mL (cm3), such that a factor of 103 must be included in the calculation of V0 when these units are used. Some common values for the empty volume Ve and the void volume V0 of HPLC columns (taking ε ¼ 0.7) are given in Table 4.1.1. However, accurate void volume of a column must be experimentally measured with an unretained compound. At U ¼ 1 mL/min, the void time t0 is numerically equal to the void volume V0 . The estimation of the void volume of the column is important for the calculation of several parameters in chromatography (see further retention factor). Also, during gradient operation, the void volume must be considered for the evaluation of the time when a certain mobile phase concentration is reaching the end of the column.

Retention factor An important parameter characterizing the chromatographic peak for a compound X is the retention factor k0 ðXÞ (in some texts indicated as capacity factor) defined by the formula: k0 ðXÞ ¼

tR ðXÞ  t0 t 0 R ðXÞ ¼ t0 t0

(4.1.11)

152

Method Development in Analytical HPLC

Table 4.1.1 Typical values for the void volume of packed HPLC columns for ε ¼ 0.7. Dimensions (i.d x length in mm)

Empty volume Ve mL

Void volume V0 mL

2.1  100 2.1  150 2.1  250 2.1  300 4.6  100 4.6  150 4.6  250 4.6  300 10.0  100 10.0  150 10.0  250 10.0  300

0.35 0.52 0.87 1.04 1.66 2.49 4.15 4.99 7.85 11.78 19.63 23.56

0.24 0.37 0.61 0.73 1.16 1.75 2.90 3.49 5.50 8.25 13.75 16.49

In Eq. 4.1.11, the value of t0 must be obtained (using an unretained compound) in identical experimental conditions (mobile phase composition, temperature, flow-rate) with those used for measuring tR ðXÞ. The values for log k 0 can be in the range between 0.3 and 1.5, although even larger values are possible for some compounds. When log k0 exceeds 1.3e1.5 for a compound, the corresponding retention times are usually too long for a practical separation. (The recommended notation for retention factor is k and not k0 as used in this book. However, for keeping a distinctive notation for the retention factor, the notation k0 was preferred). Retention factor does not depend directly on the flow rate of the mobile phase, and while t 0 R ðXÞ is more useful for analytical purposes, k0 ðXÞ is a better parameter for the characterization of the analyte, the column, and the mobile phase composition. Retention factor k0 ðXÞ is related to the equilibrium constant KðXÞ for the equilibrium: Xmo % Xst

(4.1.12)

For this equilibrium, the constant KðXÞ is given by the expression: KðXÞ ¼

½Xst ½Xmo

(4.1.13)

The value of KðXÞ depends on the nature of X, the nature of stationary phase and that of mobile phase. Its values for common HPLC separations vary in a large range and log KðXÞ can have values in common HPLC separations between 0.2 and 2.5, although it can be lower for not retained compounds or higher for very strongly retained compounds.

Performance and utilization of HPLC

153

In HPLC, the nature of mobile phase can be kept constant during a separation (isocratic mode), or can be modified (gradient mode). In the condition of isocratic separations, KðXÞ is constant, but for gradient separations, a parameter KðX; tÞ should replace the constant KðXÞ. The relation between k0 ðXÞ and KðXÞ is the following: k0 ðXÞ ¼ KðXÞ

Vst Vst h KðXÞ Vmo V0

(4.1.14)

As indicated in Eq. 4.1.14, retention factor is directly proportional with the equilibrium constant KðXÞ and each analyte for a specific stationary phase and mobile phase composition has its own retention factor k0 . In HPLC, the separation of a sample components is based on the difference between the values of their retention factors k0 that determines by Eq. 4.1.11 the retention time of each compound. In Eq. 4.1.14, Vst is the volume of stationary phase, and Vmo ¼ V0 is the volume of the mobile phase in the chromatographic column. The ratio Vst =V0 is indicated as phase ratio J and its value depends on column construction and also on other factors, the range of log J for RP-HPLC column varying between 0.3 and 0.9. Eq. 4.1.14 can be written in the form: k0 ðXÞ ¼ KðXÞJ

(4.1.15)

(In case the compound X is present in more than one form, as in the case of an acid that can be present in solution as XH and X  , the equilibrium constant KðXÞ should be replaced in Eq. 4.1.15 with the distribution constant DðXÞ given by Eq. 2.1.13). The proof of dependences described by Eq. 4.1.15 can be obtained by considering the migration rate uR ðXÞ in the chromatographic column of a retained molecule. If during the separation all the molecules of compound X would be all the time in the mobile phase, then uR ðXÞ is equal to u (the linear flow rate of the mobile phase). However, some of the molecules are intermittently retained such that only a fraction of molecules of compound X are present in mobile phase and moving, while the other fraction is immobilized. Assuming that during the separation process the number of molecules of compound X that are in the mobile phase is nmo ðXÞ and those present in the stationary phase is nst ðXÞ, then the fraction of molecules of X that are present at a given time in the mobile phase is given by the expression: nðXÞ ¼

nmo ðXÞ nmo ðXÞ þ nst ðXÞ

(4.1.16)

Using Eq. 4.1.16, the value of uR ðXÞ can be written as follows: uR ðXÞ ¼ nðXÞ u

(4.1.17)

154

Method Development in Analytical HPLC

Knowing the fraction of molecules nðXÞ in the mobile phase, and the fraction of molecules X remaining in the stationary phase being (1 - nðXÞ), it is possible to calculate equilibrium constant KðXÞ by the formula: KðXÞ ¼

ð1  nðXÞÞ V0 nðXÞ Vst

(4.1.18)

On the other hand, in a chromatographic column of length L, it can be seen that the following expression is valid: L ¼ uR ðXÞ tR ðXÞ ¼ u t0

(4.1.19)

Based on Eqs. 4.1.17 and 4.1.19, the ratio (1nðXÞ)/nðXÞ can be written as follows: ð1  nðXÞÞ tR ðXÞ  t0 ¼ nðXÞ t0

(4.1.20)

As a result of Eq. 4.1.20, the expression of KðXÞ becomes: KðXÞ ¼

tR ðXÞ  t0 V0 t0 Vst

(4.1.21)

and the proof of Eq. 4.1.14 is complete. Eq. 4.1.14 plays an important role in connecting the retention time of a compound X with the retention factor as expressed by the formula: tR ðXÞ ¼ t0 ½1 þ k0 ðXÞ ¼ t0 ½1 þ KðXÞJ

(4.1.22)

The value of k0 ðXÞ is a good indicator regarding how strong a compound is retained on a chromatographic column. However, the value of k0 ðXÞ is a function of time in case of gradient separations which are presented in Section 4.3. For value of k 0 ðXÞ < 1.0, the retention is considered very weak, but values k0 ðXÞ > 20.0 (log k0 ðXÞ z 1:3) although indicate good retention, may lead to retention times that are too long. Retention factor k0 being dependent on an equilibrium constant (see Eq. 4.1.15) also depends on temperature. It depends on the nature of the analyte, the nature of stationary and mobile phase, but in principle, the retention factor should be independent of the dimensions of the column that is used for its measurement. However, in practice, it has been demonstrated, for example, that k 0 ðXÞ measured for columns with small volumes (i.e., 5  2.1 mm i.d.) compared to conventional columns (e.g., 100  2.1 mm i.d.) even if the stationary phase is the same, are different by about 20% [11]. Also, because KðXÞ depends on the mobile phase composition, retention factor as previously described varies when mobile phase composition changes as in

Performance and utilization of HPLC

155

gradient separations. For this reason, in gradient separations, k 0 ðXÞ will become k 0 ðX; tÞ, where t indicates the time during the gradient. The value of phase ratio J depends on column construction [10,12] but also on the mobile phase that is adsorbed as a fixed layer on the stationary phase [13e15] and even on temperature [16]. In addition, the changes in the penetration of the analytes in the stationary phase pores may be affected by the pressure in the chromatographic column which may affect J and consequently the value of k0 ðXÞ. An estimation of the value of J can be obtained from the value of V0 based on Eq. 4.1.10 and various approximations for Vst . For RP-HPLC for columns made using packed particles of silica having a bonded alkyl silane moiety, Vst can be estimated based on the following expression [10]: Vst ¼

%C Msil Wpac 1201:1 nC rbond

(4.1.23)

In Eq. 4.1.23, %C is the number of g of carbon per 100 g of bonded silica, Msil is the molecular weight of silane groups, Wpac is the weight (in g) of the packing material (amount of stationary phase), rbond is the density of the bonded alkyl silane groups (with values around 0.86), and nC the number of carbon atoms in the alkyl silane.

Characteristics of an ideal peak shape The sample is injected in the HPLC system as an extremely narrow band (plug). The width of this band is usually neglected. During the chromatographic process, the band is broadened due to a number of effects. One of these effects is the ordinary diffusion which leads to a Gaussian bell curve for the shape of the chromatographic peaks. In addition to diffusion, other processes contribute to the peak broadening in HPLC. Peak broadening due to diffusion can be studied based on Fick’s laws. Fick’s second law (e.g., Ref. [17]) describes the variation of instant concentration of compound X during the diffusion process in one direction (longitudinal diffusion in the direction of x). For a chromatographic process, the instant concentration of X denoted as y will be described by Fick’s equation: dy v2 y ¼D 2 dt vx

(4.1.24)

In Eq. 4.1.24, t is time, the peak height y is expressed in units of mass per units of length, and D is the diffusion coefficient of the diffusing species in a specific solvent, and at a specific temperature. The values of D are typically expressed in cm2s1 (for a specific compound X, the diffusion coefficient is indicated as DX;A where A indicates the solvent). Fick’s law can be directly applicable to the diffusion in a tube or a rectangular channel where the peak height y varies only along the channel length and is the same across the channel. On the basis of the assumption that for t ¼ 0 (initial

156

Method Development in Analytical HPLC

condition) the concentration is described by a variable yðx; 0Þ, the solution of Eq. 4.1.24 can be written as follows (e.g., Ref. [18]): 1 yðx; tÞ ¼ pffiffiffiffiffiffiffiffi pDt



ZþN yðh; 0Þ exp N

 ðh  xÞ2 dh 4Dt

(4.1.25)

With the assumption that D is constant, and the whole amount qinj of material was initially injected (at t ¼ 0) and is contained in one point at x ¼ 0, upon integration, Eq. 4.1.25 leads to the following expression (e.g., Ref. [19]):  2 qinj x yðx; tÞ ¼ pffiffiffiffiffiffiffiffi exp 4Dt 2 pDt

(4.1.26)

In Eq. 4.1.26, t is the time of diffusion, x is the distance from the initial point of application of the amount qinj , and y is a mass per length (since D is expressed in cm2/s, y is expressed in mass/cm). Variable yðx; tÞ can be considered equivalent with an instantaneous concentration. A common notation is introduced in Eq. 4.1.15: s2L ¼ 2Dt

(4.1.27)

(sL is expressed in length since D is expressed in cm2/s). With this notation, the formula for yðx; tÞ (where t ¼ const. and y depends on t through sL ) can be written as follows: qinj yðx; tÞ ¼ qffiffiffiffiffiffiffiffiffiffiffi exp 2ps2L



x2 2s2L

 (4.1.28)

Eq. 4.1.28 characterizes a typical Gaussian bell curve (as a function of x while t is fixed but part of the value of sL ). This curve is called a normal probability density function, and is used to describe a random process. The parameter s2L is called the variance, and sL is called standard deviation. The parameter sL describes the width of the Gaussian curve, larger sL leading to wider bell shapes. This can also be seen by obtaining from Eq. 4.1.28, the value of x as a function of y which can be done using the expression: rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi h pffiffiffiffiffiffiffi  2 x ¼  2sL ln y=qinj sL 2p

(4.1.29)

From Eq. 4.1.29, the bell width W ¼ 2jxj is obtained as an increasing function of sL . Of particular importance in chromatography are also the width at half height Wh

Performance and utilization of HPLC

157

and the width at the inflection point Wi of the Gaussian curve. Taking in Eq. 4.1.29 the 1=2

 and the following result is obtained for values y qinj ¼ 0.5, ymax ¼ 0:5 2ps2L the Gaussian width: Wh ¼ 2ð2 ln 2Þ1=2 sL

(4.1.30)

From the second derivative (as a function of x) of Eq. 4.1.28 (that gives the inflection of Gaussian curve), the value for Wi is obtained as: Wi ¼ 2sL

(4.1.31)

Up to this point, the diffusion process was considered for a “static material” that diffuses in one dimension and formula 4.1.28 gives the peak height y as a function of distance x from the point of application. However, in a chromatographic process, the compound of amount qinj not only diffuses (along the length x) but also advances (is eluted) along the chromatographic column due to the movement of the mobile phase. At the same time as the analyte is diffusing, it is also moved with the distance x. The expression for the peak height y, as a function of the distance x from the point of application, after the diffusion zone was eluted with the distance x, is the following: qinj yðxÞ ¼ qffiffiffiffiffiffiffiffiffiffiffi exp 2ps2L



ðx  xÞ2 2s2L

 (4.1.32)

Eq. 4.1.32 can be used for the understanding of peak broadening in a chromatographic process. For a separation where the mobile phase has a linear flow rate u, the distance from the origin to the center of the moving zone is x ¼ u t R . Therefore, from Eq. 4.1.27 for a given x, the resulting s2L is given by the expression: x s2L ¼ 2D ¼ 2DtR u

(4.1.33)

Formula 4.1.33 shows that at larger values for the distance x (equivalent to a larger tR ), the value of s2L is larger and as a result the peaks are broader and the value of y is lower. This process is illustrated in Fig. 4.1.1. Assuming that the chromatogram is registered using a detector with a specific response coefficient b, the height of a peak in the chromatogram will be h ðxÞ ¼ b$yðxÞ, and as a result, Eq. 4.1.32 can be written in the form: bqinj hðxÞ ¼ qffiffiffiffiffiffiffiffiffiffiffi exp 2ps2L



ðx  xÞ2 2s2L

 (4.1.34)

158

Method Development in Analytical HPLC

Figure 4.1.1 Peak broadening (in units of length) of an eluting analyte along a chromatographic column with sL ¼ 0.2, sL ¼ 0.4, and sL ¼ 0.8. These values for sL can be obtained, for example, for D ¼ 0.01 and x=u ¼ 2, 4, and 32, respectively.

The broadening in units of length of the chromatographic peak previously discussed should now be converted into a broadening in units of time. In the chromatographic processes, the measured parameter is the retention time and not the length of the path in the column, and for this reason, sL (function of distance) should be replaced with a st a function of time (time broadening). This time broadening obtained from Eq. 4.1.34 is given by the formula: st ¼

sL tR sL ¼ tR u t0 L

(4.1.35)

With this replacement, Eq. 4.1.34 will give the variation of hðtÞ (as peak height per time) by the formula: bqinj hðtÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffi exp 2pst 2



ðt  tR Þ2 2st 2

 (4.1.36)

Taking the time as the x-axes of a chromatogram, peak broadening can be described in time units. For example, half height width Wh and the width at the inflection point Wi of the Gaussian curve are needed in units of time. A width at the baseline Wb ¼ 2 Wh is also measured in time units. The peak width Wb is measured between the points of intersection of baseline with the tangents to the curve at the inflexion points to the Gaussian curve representing the chromatographic peak. Fig. 4.1.2 shows the measurements of tR , Wi ; W h , and Wb on a (model) chromatographic peak having the x-axis expressed as time. The calculation of st from Eqs. 4.1.30 and 4.1.31 generates the following formulas (where st , Wi ; W h , and Wb are in time units): st ¼ ð8 ln 2Þ1=2 Wh ¼ 0:5 Wi

(4.1.37)

Performance and utilization of HPLC

159

Figure 4.1.2 Measurements of retention time tR and peak broadening Wi ; W h , and Wb on a (model) chromatographic peak with time as x-axes.

and also: st ¼ 0:25 Wb

(4.1.38)

Broadening of the chromatographic peaks is not caused only by diffusion. Other processes with random character also contribute to peak broadening.

Efficiency of a chromatographic column The peak characteristics previously described are used for the evaluation of the efficiency of a chromatographic column used for the analysis. The value of sL (space broadening) is related to another parameter used to characterize zone spreading, namely the height equivalent to a theoretical plate H ðHETPÞ, which is defined as: H¼

s2L L

(4.1.39)

This parameter is very useful in chromatography for the characterization of peak broadening per unit length of the column. Expressed as a function of time (see Eq. 4.1.35), Eq. 4.1.39 can be written in the form: H¼

s2t L tR2

(4.1.40)

In addition to H, the peak broadening characterization in a column can be obtained using the theoretical plate number N. For a column of length L, N is defined as: N¼

L H

(4.1.41)

160

Method Development in Analytical HPLC

The theoretical plate number N can be expressed as a function of length and s2L by a simple substitution of Eq. 4.1.39 in Eq. 4.1.41 to give the expression: N¼

L2 L2 ¼ 2 sL 2DtR

(4.1.42)

The value of s2L in Eq. 4.1.42 is expressed in length and can be further related to peak broadening in time Wb . For this purpose, sL will be replaced with st using Eq. 4.1.35, and L will be expressed as function of time as L ¼ t0 u. With these two substitutions in Eq. 4.1.42, the following formula for N is obtained: N¼

tR2 tR2 ¼ 16 s2t Wb2

(4.1.43)

Similar relations can be established between N and Wi or Wh (time units) with the exchange of coefficient 16 with 4 or 5.5452, respectively. In addition to the theoretical plate number N, an effective plate number n is defined by using t 0 R in Eq. 4.1.43 instead of tR . The formula for n will be: n ¼ 16

t 0 2R Wb2

(4.1.44)

From Eqs. 4.1.43 and 4.1.44, the relation between n and N can be written as follows: n¼N

t 0 2R tR2

(4.1.45)

The value for n is smaller than that for N since t 0 R < tR . Since tR (and t 0 R ) as well as Wb are chromatographic parameters dependent on the eluting compound (index X was omitted in previous formulas), the value for N and n are also compound dependent. Both Eq. 4.1.43 or Eq. 4.1.44 can be used to measure the theoretical plate number or effective plate number based on experimental data obtained with a given column. This measurement is useful in practice to select columns (higher N gives lower peak broadening) and also to assess the loss in performance of a column after a certain period of usage when the N values start to decrease. Because N is related to the important characteristic of peak broadening, it is common to indicate it as a parameter to characterize the efficiency of a column. The values for N for HPLC columns can be given either for a specific column or reported as efficiency per meter (N=m). Both N and n are used for the characterization of column efficiency, and n is sometimes named simply efficiency. For modern HPLC for columns with stationary phase made from porous particles, N can vary (per m) between 40,000 and 120,000 (or even higher), and for core-shell columns can be as high as 300,000 (common column length L is between 50 and

Performance and utilization of HPLC

161

250 mm). The values for N (per unit length of the column) are influenced by physical properties of the stationary phase such as dimension of stationary phase particles, homogeneity of the particles’ dimensions, and structure of particles. For a chromatographic column packed with particles made from a solid support coated with an active stationary phase (as most chromatographic columns are), a number of other effects contribute to the peak broadening, besides simple diffusion. All of these effects have their own contribution, but they are also random processes and the Gaussian shape of the chromatographic peaks is preserved. However, the variance s2 (either s2L or s2L ) must be replaced with a combination of these contributions as indicated by Eq. 1.7.25, leading to a value for s2 as given by the following expression: s2 ¼

X

s2n

(4.1.46)

n

where s2n are the variances generated by each of these random independent processes. Based on Eq. 4.1.46, the expression for Wb can be written as follows: Wb ¼ 4st ¼ 4

rffiffiffiffiffiffiffiffiffiffiffiffi X s2n

(4.1.47)

n

The variances included in Eq. 4.1.46 include: (1) longitudinal diffusion with variance sL contributing to the height equivalent to a theoretical plate H ðHETPÞ with value HL , (2) eddy diffusion (multipath of the same type of molecules in the porous structure of the column particles) with variance sE and contributing to H with value HE , (3) lateral movement of molecules due to convection with variance sC and contributing to H with value HC , (4) the differences between individual molecules in the mass transfer rate in and out the stationary phase with variance sT and contributing to H with value HT , and (5) presence of random spots of stagnant mobile phase in the porous material of the column (mass transfer in and out mobile phase) with variance sS and contributing to H with value HS . The inclusion of the contribution of all these factors to the value of plate height H ðHEPTÞ requires the replacement of Eq. 4.1.46 with a new formula of the form: H¼

s2L þ s2E þ s2C þ s2T þ s2S ¼ HL þ HE þ HC þ HT þ HS L

(4.1.48)

Each term in Eq. 4.1.48 can be evaluated and also experimental procedures were developed for their measurement [20]. The detailed evaluation of each term in Eq. 4.1.48 leads to the following expression for H: " !# df2 dp2 dp2 2D k0 þ G þQ þ H ¼ Ldp þ g u 2 u D ð1 þ k0 Þ Ds D

(4.1.49)

162

Method Development in Analytical HPLC

In Eq. 4.1.49, L is a coefficient dependent on how irregular are the particles in the column, g is another coefficient depending on column packing and is typically around 0.625, D is the diffusion coefficient of a separated species X in a specific mobile phase, G is a parameter depending on column packing, dp is the average diameter of particles in the column, Q is a proportionality constant (estimated as Q ¼ 1/30), df is the depth of the stationary phase on the solid support, Ds is the diffusion coefficient of the analyte in the stationary phase, and k 0 is the retention factor for the solute X. Eq. 4.1.49 indicates the dependence of the value of H on linear flow rate, this dependence including parameters related to the column construction as well as on the nature of the analyte (through k0 and D) and even mobile phase (through D). A simplified formula for Eq. 4.1.49 can be written in the form: dp2 D H ¼ A dp þ B þ C  u u D 

A ¼ L;



B ¼ 2g;



C ¼G þ Q

(4.1.50) k0 ð1 þ k0 Þ

df2 D 2

Ds dp2

! þ1

In a further simplified form Eq. 4.1.49 can be written as follows [21]: H ¼A þ

B þ Cu u

(4.1.51)

Eq. 4.1.51 is known as van Deemter equation and it shows how the value of H depends on the linear flow rate in a chromatographic column made from coated particles. An example of the graph of van Deemter equation for A ¼ 3.0 mm, B ¼ 1000 mm2 = s, and C ¼ 0.0008 s is given in Fig. 4.1.3.

Figure  4.1.3 The plot of van Deemter equation and of its components for A ¼ 3 mm, B ¼ 1000 mm2 s, and C ¼ 0.0008 s.

Performance and utilization of HPLC

163

The dependence of H on u given by Eq. 4.1.50 for van Deemter equation indicates that a minimum value for H is obtained at a specific u value that can be obtained from the condition dH=du ¼ 0, leading to the expression: D uopt ¼ dp

rffiffiffiffiffiffi B C

(4.1.52)

For the optimum volumetric flow rate, Eq. 4.1.52 leads to the following expression: pε d 2 D Uopt ¼ 4dp

rffiffiffiffiffiffi B C

(4.1.53)

Eq. 4.1.53 indicates that Uopt for a chromatographic column is larger when the column diameter d is larger, the diffusion coefficient D is larger, and when the particle diameter dp is smaller. Variation of the plate height with H with the flow rate U for several types of particles in the column and for a monolith are indicated in Fig. 4.1.4 [22]. The column construction characteristics are included in parameters ε , B , and C  . From Eq. 4.1.52 for uopt placed in Eq. 4.1.50, the expression for Hmin for a chromatographic column is obtained as follows:  pffiffiffiffiffiffiffiffiffiffiffi Hmin ¼ dp L þ 2 B C

(4.1.54)

pffiffiffiffiffiffiffiffiffiffiffi It can be roughly estimated that L þ 2 B C  z 2 to 3, depending how irregular are the particles in the column as described by parameter L, and the values of B

Figure 4.1.4 Variation of the plate height with H with the flow rate U for several types of particles in the column and for a monolith.

164

Method Development in Analytical HPLC

and C  . As a result, Hmin ¼ ð2 to 3Þdp . A too slow flow or an increase in flow rate beyond uopt leads to a decrease of column performance due to the increase in H. Faster flows than uopt (and Uopt ) may be sometimes needed for generating shorter run times, but they should not exceed too much Uopt in order to not decrease too much the theoretical plate number N. An estimated value for N can be obtained from Eqs. 4.1.41 and 4.1.54 with formula: N¼

1000 L pffiffiffiffiffiffiffiffiffiffiffi dp L þ 2 B  C 

(4.1.55)

(the coefficient 1000 compared to Eq. 4.1.41 is necessary to keep similar units). Eq. 4.1.55 indicates that length of the column and the diameter of particles in the chromatographic column play an important role in the value of column efficiency, with smaller particles leading to higher efficiency. Also values for L that are lower for particles that are spherical and with more homogeneous size distribution contribute to the increase of N. Eq. 4.1.55 can be approximated with the following expression: N¼

1000 L Ct dp

(4.1.56)

The value of the constant Ct varies between 2 and 3 depending on particle shape and structure, as well as other stationary phase characteristics. An irregular particle shape leads to a higher Ct. Particles dimension also contributes to the increase of backpressure of the chromatographic column. This backpressure Dp is the difference between the pressure at the column inlet and that at the outlet of the column and is given by the following expression (Darcy equation): Dp ¼

hufr L hfr L2 ¼ 2 dp2 dp t0

(4.1.57)

where h is the mobile phase viscosity, L is column length, dp is the diameter of the particles in the bed, and fr is the column flow resistance factor (and u is linear flow rate).

Peak asymmetry Besides the effects previously indicated to affect the peak shape but maintain their Gaussian shape because they are random processes, other effects can take place in the chromatographic column leading to peak asymmetry. For example, the presence of the analyte in more than one form and interacting differently with the stationary and mobile phase may lead to peak asymmetry. Also, the adsorption of analyte on certain sites from stationary phase slower than desorption may be another cause of peak asymmetry. One additional common cause for peak asymmetry is the coelution of two compounds in the same peak with the detector unable to differentiate them (e.g.,

Performance and utilization of HPLC

165

Figure 4.1.5 Front f and rear r for an asymmetric chromatographic peak.

absorbing at the same wavelength in case of UV detection). When the two peaks have slightly different retention times, the resulting peak is not Gaussian as illustrated in Fig. 4.1.5. Peak asymmetry can be characterized by two parameters. One is the asymmetry parameter AsðXÞ, and the other is tailing factor TFðXÞ. The asymmetry AsðXÞ is defined as the ratio of the rear r versus front f segments cut on the chromatographic peak by a parallel to the baseline at 10% peak height (some other choices are possible), and separated by the perpendicular from the apex of the peak. AsðXÞ is given by the formula: AsðXÞ ¼

r f

(4.1.58)

The rear r and front f obtained at 10% of the height are shown on an asymmetrical peak in Fig. 4.1.5. When r ¼ f , the value of AsðXÞ becomes 1 (ideal Gauss peak), a reason why this parameter is also indicated as symmetry factor. A value of 0.9 for AsðXÞ is acceptable as a slight fronting of the peak. The peak tailing is defined by the formula: TFðXÞ ¼

f 0 þ r0 r0 ¼ 0:5 þ 0 0 2f 2f

(4.1.59)

where f 0 and r 0 are measured in the same way as f and r, but at 5% of the peak height. The increase in TFðXÞ from 1.0 indicates peak tailing, while the decrease from 1.0 indicated fronting (peak with f 0 > r 0 ). In practice, peak tailing and peak asymmetry are increased along the usage of a chromatographic column. In time, after a number of separations, the quality of column is diminished and more peak fronting or tailing can be observed. A new column is considered acceptable when the values of As for a probe compound are between 0.9 and 1.2.

166

Method Development in Analytical HPLC

Key points • •

A chromatographic peak is characterized by a number of parameters describing its position in the chromatogram and its shape. Parameters related to HPLC system and utilization such as flow rate U of the mobile phase, chromatographic column construction, and mobile phase composition contribute to peak shape and position in the chromatogram.

4.2

Parameters related to peak separation in HPLC

The separation of the peaks in a chromatogram is characterized by several useful parameters, including selectivity a and resolution Rs. In addition, an overall parameter Pc indicating the total number of peaks that can be separated in a chromatogram is also useful for describing how good is an HPLC separation. Some aspects regarding these parameters are further presented.

Selectivity (separation factor) Selectivity (or separation factor) is a parameter a used for the characterization of the separation of two compounds X and Y. Selectivity is defined by the ratio: aðX; YÞ ¼

t 0 R ðYÞ t 0 R ðXÞ

(4.2.1)

where t 0 R ðYÞ > t 0 R ðXÞ. Eq. 4.2.1 shows that a can also be expressed by the formula: aðX; YÞ ¼

k 0 ðYÞ KðYÞ ¼ k0 ðXÞ KðXÞ

(4.2.2)

As expected, selectivity depends on the corresponding equilibrium constant for each analyte. In a chromatographic separation, larger a values are desirable for a better separation of compounds X and Y. However, even when a is large, peak broadening of the two chromatographic peaks for X and Y can be large as well, such that the peaks still can have some overlapping. This broadening is accounted for in another parameter indicated as resolution.

Resolution Resolution Rs, or the resolving power for a pair of analytes [23], is a parameter that considers both the separation in time of two chromatographic peaks, as well as their broadness. This parameter is defined by the formula: RsðX; YÞ ¼

2½tR ðYÞ  tR ðXÞ Wb ðXÞ þ Wb ðYÞ

(4.2.3)

Performance and utilization of HPLC

167

The difference in the retention times tR ðYÞ  tR ðXÞ ¼ DtR can be replaced with t 0 R ðYÞ  t 0 R ðXÞ. The values for Eq. 4.2.3 can be obtained from the chromatogram as illustrated in Fig. 4.2.1. A peak separation is typically considered acceptable when Rs > 1:0 and good when Rs > 1:5. In the case when 2DtR > Wb ðXÞ þ Wb ðYÞ, the two peaks are separated at the baseline. The widths at the baseline of the two peaks can be different (as also shown in Fig. 4.2.1), but as an approximation it is possible to take Wb ðXÞ z Wb ðYÞ ¼ Wb . With these assumptions, the formula for Rs can be written in the form: Rs ¼

DtR Wb

(4.2.4)

Expression 4.2.4 clearly indicates that resolution is higher when the peaks are more separated and when they are narrower. The generation of narrower peaks in the chromatogram is very much dependent on chromatographic column construction, and significant effort continues in the direction of making columns that generate narrow peaks [24]. The difference in retention time between two analytes DtR can be written as a function of selectivity a in the following form: DtR ¼ ða  1Þt 0 R ðYÞ

(4.2.5)

The expression for Rs with this substitution becomes: Rs ¼ ða  1Þt 0 R ðYÞ=Wb

(4.2.6)

Figure 4.2.1 An idealized chromatogram showing the measurable parameters used for the calculation of resolution Rs.

168

Method Development in Analytical HPLC

The value for t 0 R ðYÞ expressed in terms of tR ðYÞ using Eq. 4.1.11 generates the following formula: Rs ¼ ða  1Þ

k0 ðYÞ tR ðYÞ 1 þ k0 ðYÞ Wb

(4.2.7)

  pffiffiffiffi From Eq. 4.1.43, using the value tR ðYÞ Wb ¼ 1 4 N , the value for Rs will become: 1 k 0 ðYÞ Rs ¼ ða  1Þ N 1=2 4 1 þ k 0 ðYÞ

(4.2.8)

Eq. 4.2.7 indicates the resolution relative to the “previous peak.” Similar formula for Rs can be obtained relative to the “next peak” as follows: Rs ¼

1 ða  1Þ k0 ðXÞ N 1=2 4 a 1 þ k0 ðXÞ

(4.2.9)

Both relations 4.2.8 and 4.2.9 are approximations and the two values should be close to each other. Both expressions are obtained by taking the peak width at the baseline as equal for the two peaks, and the theoretical plate number N as measured for one compound (Y in Eq. 4.2.8) or for the other (X in Eq. 4.2.9). The value for Rs is most sensitive to the parameter a, which is critical for obtaining a good separation. Larger k0 values are also useful, but as k 0 increases, its importance for the increase in Rs is diminished. This is exemplified in the diagram from Fig. 4.2.2, for a chromatographic column with N ¼ 18,000.

Figure 4.2.2 Graph showing the variation of Rs as a function of a and k 0 assuming a chromatographic column with N ¼ 18,000 [22].

Performance and utilization of HPLC

169

The requirement for the value of resolution Rs to be higher than 1.0 in order to have a good separation is translated for the theoretical plate number N in the requirement to satisfy the relation: ð1 þ k 0 Þ a2 2 k0 ða  1Þ2 2

N > 16

(4.2.10)

In many practical applications, the separation factor a between an analyte and other components of a specific matrix may be too close to unity. The increase in the number of theoretical plates N of the column can be helpful for enhancing separation in these cases. The increase in value of k 0 can be achieved by increasing KðXÞ, and also J. The value of KðXÞ depends on the nature of X, but also on the nature of stationary phase and mobile phase. The goal of developing a successful separation, the selection of stationary phase and mobile phase are very important and specifics about such selections are discussed in present book. Another path for the increase of k 0 is through the increase of the phase ratio J. Based on the estimation indicated in Eq. 4.1.23, the increase in J can be obtained for RP-HPLC by having more stationary phase. The N dependence on column length (see Eq. 4.1.56) and J dependence on the amount of stationary phase indicate that columns that are longer and with wider diameter tend to show better resolution. The structure of particles in the column also can increase the value of N by using more homogeneous particles.

Peak capacity The efficiency of an HPLC separation can also be characterized by a parameter known as peak capacity Pc. This parameter gives the number of peaks in a chromatogram that can be separated from one another with a resolution Rs ¼ 1 within a specified windows of time defined by the longest retention time tR;max . The theory of a maximum peak capacity in isocratic elution is based on the fact that an ideal peak (Gaussian shape) has the peak width Wb ¼ 4st (see Eq. 4.1.38) and the definition of Pc can be interpreted as the length (in time) of a chromatogram divided by one peak width pffiffiffiffi (Wb ¼ 4st ) and the value of st can be related using Eq. 4.1.43 (st ¼ tR N ). This leads to the following expression for Pc: pffiffiffiffi   N tR;max Pc ¼ 1 þ ln t0 4

(4.2.11)

It should be specified that Eq. 4.2.11 giving the number of peaks in a chromatogram that can be separated from one another is valid only for isocratic separations. Eq. 4.2.11 indicates that in a longer chromatogram for a column with a higher number of theoretical plates, a larger number of compounds can be in theory separated [25]. Eq. 4.2.11 is usually applied in reversed-phase HPLC, but for other separations, mechanisms such as size exclusion or ion exchange, special formulas for Pc were also

170

Method Development in Analytical HPLC

developed [26]. An alternative value for peak capacity that considers the resolution between the separated peaks is given by the expression: pffiffiffiffi N Pc ¼ 1 þ ln ðk 0 max þ 1Þ 4Rs

(4.2.12)

In Eq. 4.2.12, k 0 max is the largest retention factor of the analytes in the separation. Modern HPLC columns having small particles (e.g., core-shell) can reach Pc values between 100 and 200 for a run time trun of about 15 min.

Key points • •

Separation in a chromatogram is characterized by selectivity, resolution, and peak capacity, the most relevant parameter being the resolution. Resolution depends on the nature of separated compounds, chromatographic column, and mobile phase used for the separation.

4.3

Parameters for characterization of gradient separation in HPLC

The gradient separation in HPLC is very common, and designing an appropriate gradient is an important part of a method development. The most typical gradient is performed with the change in mobile phase composition either only regarding the proportion of the different mobile phase components or even by changing the nature of those components. During gradient, the mobile phase composition is usually modified such that the retention factors k0 of the analytes are decreased (therefore the retention times tR are decreased) and the mobile phase becomes a “stronger” eluent. Reversed gradient is also possible but not common. The change can start at the beginning of the chromatographic run and continue with the same gradient slope during the whole run (see Section 3.2). However, the run may start with an isocratic hold for a short period of time and then the gradient starts. Also, multiple gradient slopes can be used in a run, can be performed as a step gradient, and it is common toward the end of the run to use a resetting of the mobile phase composition to the initial values to prepare the system for the analysis of a next sample. For RP-HPLC, which is the most common type of HPLC separation, reduction of retention factors k0 is achieved by increasing the organic component in the mobile phase. For HILIC, the gradient can be obtained by increasing the content in water during the run. In other cases, the changing the pH leads to lower retention times. Other types of gradients are also possible such as gradient in the temperature or in the flow rate. The gradient is used for several purposes. One main purpose is to reduce the total run time without deteriorating the separation. For this purpose, it is common to keep an isocratic separation at the beginning of the run and as the gradient starts, to gradually reduce the retention times only for the compounds that would elute late with the initial

Performance and utilization of HPLC

171

mobile phase. In this manner, the compounds with low k 0 values that are early eluting with the initial mobile phase are not affected maintaining good Rs values (see Eq. 4.2.9). The compounds with large k0 values at the initial mobile phase composition have their k 0 reduced during the gradient, but these values still remain high enough to allow good separation. As an example, the separation of benzo[a]pyrene (BaP) and indeno[1,2,3-cd]pyrene separated from a moist snuff sample is shown in Fig. 4.3.1. The separation was performed on a Zorbax Eclipse XDB-C18 4.6  250 mm 5 m m column, using gradient starting with 25% water and 75% acetonitrile for 0.5 min, then to 100% acetonitrile at 12.0 min (linear), holding at 100% acetonitrile for 6 min. At 18 min, the eluent was returned to initial composition within 0.5 min and held for column equilibration for another 1.5 min (total run time 20 min). The flow rate was 1 mL/min, and the column temperature was 25 C. The detection of BaP was done using fluorescence with excitation at 378 nm and emission at 405 nm and after 15 min the excitation was changed to 370 nm and emission to 460 nm (for the detection of indeno[1,2,3-cd]pyrene). The injection volume was 20 m L. The level of BaP corresponded to about 7.6 ng/mL [27]. As indicated in Fig. 4.3.1, both benzo[a]pyrene (BaP) and indeno[1,2,3-cd]pyrene are well separated, while a large component of the matrix is eluted early in the chromatogram not affecting the analytes. Another purpose of gradient is just to modify during the run the k 0 values of sample components with the goal of improving the separation. Different compounds X respond differently to the change in the mobile phase composition, and their k0 (and retention times) are modified differently during gradient. This may allow in some cases a better separation. This result is frequently obtained by trial and error, since the change in gradient may also decrease the quality of a separation. The combination of various gradient alternatives is commonly used in HPLC for achieving the best result on separation.

Figure 4.3.1 Chromatogram of an extract of a moist snuff sample showing the peaks of benzo [a]pyrene (corresponding to about 7.6 ng/mL BaP) and of indeno[1,2,3-cd]pyrene.

172

Method Development in Analytical HPLC

One more common purpose of gradient is to clean the column by purging it toward the end of a run with a “strong” mobile phase that would elute all the remaining components from a sample. As samples may contain compounds that are very strongly retained on the stationary phase, such compounds may practically never elute from the column when the mobile phase is not sufficiently efficient for this purpose. By using at the end of a gradient a mobile phase capable to “purge” the column, for example, a pure organic solvent in case of RP-HPLC, the column is maintained clean from one run to the next.

Retention factor in RP-HPLC gradient separation Because the most common type of HPLC is RP-HPLC, further discussion will focus on the gradient in RP-HPLC. For simplifying the presentation, the type of gradient further discussed is a continuous gradient starting at the beginning of the chromatogram and continuing with the same gradient slope during the whole chromatographic run and it is referred only to the change in mobile phase composition which is considered as made from two components (water and organic). The mobile phase composition for a binary system water/organic solvent is typically described by the volume fraction f given by the ratio: f¼

Vorg Vorg þ Vw

(4.3.1)

In Eq. 4.3.1, Vorg indicates the volume of the organic component and Vw the

volume of water (the values of f varies between 0 and 1). Indicating by ffinal    f0 tgrad ¼ Df tgrad , the gradient slope given by expression 3.2.3 (f0 is the initial mobile phase composition ffinal the final one and tgrad is length of time of the gradient). In considering the starting of the gradient and its ending time based on the formula tgrad ¼ tinit  tfinal where tfinal is the end of the gradient and tinit is the beginning time for the gradient, it must be noticed that a delay of time equal with VD =U exists between the time of starting the gradient at the head of chromatographic column and the time the gradient start as decided in the gradient program. The change in the mobile phase composition in time will be: fðtÞ ¼ f0 þ

Df t tgrad

(4.3.2)

In RP-HPLC, it was basically decided that the equilibrium of the analyte between the mobile phase and stationary phase is a partition process [28]. It should be assumed as a result that the dependence of equilibrium constant KðXÞ on molar fraction x of the organic component follows Eq. 2.1.12. The volume fraction f being proportional with the molar fraction x, and considering that phase ratio J is not changing with the mobile

Performance and utilization of HPLC

173

phase composition, the dependence of the retention factor on f can be expressed by the following relation (e.g., Ref. [29]): log k 0 ðXÞ ¼ log k 0 w ðXÞ  SðXÞf

(4.3.3)

The values of log k0 w(X) and of S(X) are usually obtained from experimental data using the least square deviation method. In Eq. 4.3.3, the value of SðXÞ is known as elution strength parameters and it is assumed constant for a chromatographic column and a mobile phase containing water and a specific organic modifier, and k0 w ðXÞ refers to retention of compound X in pure water as mobile phase. The value for k 0 w ðXÞ is not usually known and it is obtained from extrapolation to 100% from different values of k0 ðXÞ experimentally obtained as the content of water increases (without reaching 100%). The direct determination of k 0 w ðXÞ is not commonly performed by using just water as a mobile phase. For most columns and compounds, such procedure will generate unreasonably long retention times. Formula 4.3.3 is only an approximation and different values for SðXÞ are reported in the literature. These values vary and some reports indicate number as low as 3 or as high as 12 (e.g., Ref. [22]). Also, the value of k0 w ðXÞ is only an approximation of the true value that would be obtained in pure water as mobile phase. In addition, Eq. 4.3.3 may not be valid for the whole range of values for f since the assumption that phase ratio J is not changing with the composition of mobile phase is not valid for very low concentrations of the organic constituent. Several studies are presenting the results regarding k0 w ðXÞ showing its variability [30,31]. In RP-HPLC, SðXÞ is positive such that log k0 ðXÞ decreases when the organic content of mobile phase increases. The parameter SðXÞ is specific for a solute X, a solvent mixture, and a column, but does not depend on f. With fðtÞ given by Eq. 4.3.2 and based on Eq. 4.3.3, the expression for the instantaneous retention factor (at time t) k0 ðX; tÞ becomes:   Df t log k 0 ðX; tÞ ¼ log k 0 w ðXÞ  SðXÞ f0 þ tgrad

(4.3.4)

Expression 4.3.4 shows that in linear gradient conditions, the retention factor k 0 ðX; tÞ depends across the chromatogram not only on the nature of the compound X but also on the instantaneous mobile phase composition. The variation of k0 ðX; tÞ during a gradient time tgrad for two compounds having different k0 w ðXÞ values and the same S is illustrated in  Fig. 4.3.2. The gradient slope Df tgrad is typically included in another parameter indicated as gradient steepness defined by the formula: b¼

SðXÞDf tinit tgrad

(4.3.5)

With the use of gradient steepness, expression 4.3.4 is written in the form: log k 0 ðX; tÞ ¼ log k 0 w ðXÞ  SðXÞf0  b

t tinit

(4.3.6)

174

Method Development in Analytical HPLC

Figure 4.3.2 Variation of k’(X,t) during a linear gradient in RP-HPLC for two compounds with k 0 w ðXÞ ¼ 20 and k 0 w ðYÞ ¼ 200, S ¼ 4, f0 ¼ 0.1, Df ¼ 0.7, tinit ¼ 0.2 min, and tgrad ¼ 10 min.

Because in gradient a compound has different k0 values during the chromatographic run, the instantaneous value for k0 ðtÞ does neither provide a possibility to calculate the true tR as in the case of isocratic separations, nor a possibility to calculate other parameters such as a, or Rs that are necessary for the characterization of the chromatographic process. For a gradient separation, a real retention time considering the overall change in k 0 ðtÞ across the change in f during the gradient is given by the formula (for generating this expression, see e.g., Ref. [22]): tR ðXÞ ¼ tinit þ

tinit logð2:303k0 0 ðXÞb þ 1Þ þ tD b

(4.3.7)

In Eq. 4.3.7, the value of k0 0 ðXÞ is the value of k0 at f0 and it is given by the formula: log k0 0 ðXÞ ¼ log k 0 w ðXÞ  SðXÞf0

(4.3.8)

Also, in Eq. 4.3.7 appears the dwell time tD . This time is caused by the fact that the change in gradient as it is directed by the computer gradient program is reaching the head of the chromatographic column with a certain delay. This is illustrated in Fig. 4.3.3 where a profile of a gradient is shown as programmed and as it is in fact reaching the chromatographic column (real gradient).

Performance and utilization of HPLC

175

Figure 4.3.3 Difference between programmed gradient and real gradient reaching the chromatographic column.

From Eq. 4.3.7, an effective retention factor k can be obtained and should be used in gradient separations instead of k0 . The value for k is usually taken as an average value of k 0 values and an usually accepted value of k  is given by the following expression: 2 k ðX; bÞ ¼ log½2:303k0 0 ðXÞb þ 1 b

(4.3.9)

An approximation for Eq. 4.3.9 is offered by the expression: k ðXÞ ¼

U tgrad 0:5V0 DfSðXÞ

(4.3.10)

In Eq. 4.3.10, V0 is the dead volume of the column and SðXÞ for small molecules has a value around 12. Based on the value of the effective retention factor k , effective parameters describing separation such as a , Rs , and Ps can be defined by replacing k 0 with k  in the corresponding formulas. Even parameters such as height equivalent to a theoretical plate H (HETP) and theoretical plate number N being dependent on peak width are affected by the gradient. One additional parameter that changes during the gradient is the column backpressure. The viscosity of the mobile phase changes as its composition changes. Eq. 4.1.57 can be written in this case in the form: Dp ¼

4hðfÞUfr L pε d2 dp2

(4.3.11)

176

Method Development in Analytical HPLC

In Eq. 4.3.11, the viscosity hðfÞ changes during the separation, while the other parameters remain the same (fr is an empirical constant). This change in viscosity leads to changes in the column backpressure during the gradient and may lead to significantly higher backpressure.

Peak compression in gradient elution As indicated in Section 4.1, the peak width is characterized in an isocratic separation by the value Wb which is related to the peak retention time by Eq. 4.1.22 and Eq. 4.1.43 that can be written in the form: 4tR 4 Wb ¼ pffiffiffiffi ¼ pffiffiffiffi ð1 þ k0 Þt0 N N

(4.3.12)

In a gradient separation, the retention times tR are lower compared to the corresponding ones in an isocratic separation, and as a result, the widths of peaks should also be reduced as indicated by Eq. 4.3.12 if lower tR are utilized to calculate Wb . Besides this reduction caused by shorter retention times, another consequence of using a gradient separation is the effect indicated as peak compression. The gradient in mobile phase composition across the width of a peak generates for each compound a higher k0 value at the beginning of the peak compared to that at the end of the peak when k0 is diminished because the mobile phase becomes “stronger.” In a separation performed isocratically with the composition equal to the one at the tip of the peak, the resulting peak width (Wb ) can be indicated as Wpeak while the width of the peak including peak  compression can be indicated as Wcompr . The ratio Wcompr Wpeak can be used to evaluate peak compression [32e34]. The replacement of k0 in Eq. 4.3.12 with the effective retention factor k generates for the Wpeak the expression: 4 Wpeak ¼ pffiffiffiffi ð1 þ k Þt0 N

(4.3.13)

In reality, because of the change in mobile phase composition across the peak, a compression factor G < 1 must be introduced to describe Wcompr , as follows: 4G Wcompr ¼ pffiffiffiffi ð1 þ k Þt0 N

(4.3.14)

 As a result, Wcompr Wpeak ¼ G and several expressions were developed for the evaluation of G [35,36]. The value of G increases when the gradient steepness b increases, the resulting value also depending on the retention time of the solute in the initial mobile phase. Some values for G are closer to 1.0, but others are in the range 0.75e0.9 with a possible minimum value G z 0.6 [37]. The gradient peak

Performance and utilization of HPLC

177

compression is also associated with a change in the Gaussian peak shape, reduction of peak tailing, and increase in peak capacity.

Key points • • •

Designing an appropriate gradient is an important part of a method development. Gradient composition of mobile phase in HPLC is used for both reducing the retention time of late eluting peaks and for modifying separation with the goal of improving it. In gradient separations, retention factor is not a constant parameter as in isocratic HPLC, its value typically decreasing during the gradient separations.

4.4

Characterization of detector performance in HPLC

One of the advantages of HPLC as an analytical method is that it can use a variety of detector types which are based on certain physical or physicochemical characteristics of the analytes. The measurements are based on specific properties of the analyte molecules that are different from those of the mobile phase. The signal from the detector is recorded in the form of a chromatographic peak. The signal generated by detectors depends on the nature of the analyte, but for some detectors, this signal does not provide a way to identify the analyte. These detectors with signal independent on analyte nature are indicated as having a universal response. Other detectors in addition to quantitative information also provide information regarding the nature of the analytes, but the extent to which this information is diagnostic for identification of the analyte depends on many factors. The peak area in a chromatogram is dependent on either the instantaneous concentration of the detected species or on the instantaneous mass (amount) of analyte present in the detector. For this reason, the quantitation in HPLC is based on the area of the chromatographic peak (see Eqs. 4.1.36 and 4.5.3). Peak height is sometimes utilized for quantitation when the peak width is equal for all concentrations used for calibration. Besides the concentration or amount of the analyte in the detector, the electrical signal creating the chromatogram depends on the detector setting. Several characteristics can be considered for evaluating the quality of a detector and these are further presented. Before describing the detectors characteristics, it should be emphasized that the detector is part of a system that includes a separating module (chromatographic column). For this reason, the detector characteristics should be considered in relation to the separation procedure. Since the HPLC system is used in a specific analytical method, most characteristics of the detector are included in the description of the analytical method. For example, the detector selectivity is part of the selectivity of the analytical method, and the same is applicable for other characteristics such as minimum concentration (or amount) of an analyte that can be detected (limit of detection or LOD), response precision, linearity, etc. The main properties characterizing a detector include: (1) capability to provide only quantitative information or both quantitative and qualitative, (2) selectivity, (3) sensitivity and the stability of sensitivity, (4) limit of detection, (5) dynamic range and

178

Method Development in Analytical HPLC

linearity, (6) precision and reproducibility, (7) baseline noise and drift, (8) suitability to a given mobile phase and to the flow rate of the mobile phase, (9) volume of measuring cell, (10) low contribution to peak broadening, (11) dependence of response to the changes in flow rate of mobile phase, (12) frequency of data collection, (13) backpressure accepted and backpressure generated by the detector, (14) the capability of not alter the analytes or to destroy them, etc. The evaluations of some of those characteristics (as well as the characteristics of an entire analytical method) are based on individual measurements. For this reason, those specific characteristics are evaluated using a statistical approach, and the information is provided based on average of measurements and the associated errors (see Chapter 11).

Capability of a detector to provide quantitative and qualitative information Qualitative information in HPLC can be obtained from both the separation since the retention time tR ðXÞ of a compound X depends on the nature of X, as well as from the additional information from the detector. As previously indicated, the signal from universal detectors contains only quantitative information. Their signal is dependent on the concentration of an analyte in the analyzed solution, and although the nature of the analyte may influence the signal intensity, this signal cannot be used for qualitative analysis. Examples of universal detectors include refractive index detector, conductivity electrochemical detector, and evaporative light scattering (ELS) type detectors. When using this type of detectors, only the retention time can be used as a qualitative indicator. For other detectors, the signal may contain information that can be used for the qualitative recognition of a known compound and in some cases even for the identification of unknowns. These capabilities are common to a number of detectors including spectroscopic, mass spectrometric, and electrochemical detectors. For example, some UV-Vis detectors (diode array detectors (DADs)) are capable of recording the UV-Vis spectrum of a detected compound. The comparison of the recorded spectrum of an analyte expected to be compound X with a known standard spectrum of the compound X can identify the compound and even more can be used for assessing the purity of a chromatographic peak that is assumed to belong to compound X. The purity is assessed from the differences of the peak spectrum to that of the standard (see Section 11.2). The usefulness of specific signals generated by the detectors for a compound in order to be recognized in an eluted chromatographic peak varies. For example, recording UV absorption at a specific wavelength l characteristic for a specific compound, combined with the specific retention time, can be good qualitative indicators. In the same category can be included the excitation lex and emission lem wavelengths for a fluorescence detector. However, because the UV-Vis spectra as well as fluorescence spectra are usually presented as relatively broad bands, those parameters are not exact values but ranges. For this reason, it is still difficult to use such information for the discovery of unknown structures. A similar case is presented by the specific redox potential for an electrochemical detector. A more precise information about a specific compound can be obtained from a single-MS detector. This detector provides a specific value for the mass (in fact for Mw=z) of the molecular ion used for detection.

Performance and utilization of HPLC

179

A single number when also associated with a specific retention time are good indicators for a compound recognition, although interferences are still possible in particular from isomers. Identifications of unknowns, however, cannot be usually performed even from two precise numbers (mass and retention time). High-resolution MS, however, improves the chances for such identifications, mainly when combined with additional information that can be available about the sample. More qualitative information is generated when using an MS/MS detector. The use of multiple reaction monitoring (MRM) type detection offers two qualitative parameters (the mass of the molecular ion and that of the product ion used in the MRM transition). This technique is very common in LC-MS/MS (see Section 3.5) since it provides one of the most sensitive ways for a quantitative chemical analysis. When combined with the retention time of an analyte, the three parameters are providing excellent capability for the recognition of a specific compound, but it is still not necessary a procedure to identify unknown compounds. The use of confirmation MRM transitions (e.g., from the molecular ion to a different product ion) is sometimes used for further proof of a compound identity in an HPLC separation. Only when combined with a specific fragmentation pattern (the nature of fragment ion), the technique has significant capability even for unknown identification. The most qualitative information can be usually obtained using MS/MS with the full spectrum from the second MS module. Significant effort is being made to use such spectra for compound identification with dedicated computer programs (e.g., Smile MS [38]) and dedicated library for specific group of compounds [39,40]. Also, the use of high-resolution mass spectra is of further utility for compounds identifications. This capability allows the use of high-resolution LC-MS/MS as a tool for the analysis of untargeted compounds [41,42].

Selectivity The selectivity of a detector is related to the response to only a specific type of analyte. This characteristic is not applicable to universal detectors, those being designed to respond to any analyte. For a good HPLC separation and for the analysis of sample with a limited number of components, the universal detectors are useful since they account for all the components in the sample. Selective detectors respond to only a class of compounds or they can be set to detect only specific compounds. The detector selectivity adds one more capability for differentiating certain analytes during detection, and this can be a significant advantage when the HPLC separation is not possible (or it is incomplete). However, the use of detectors with high selectivity is applied only for the analysis of targeted compounds. Such detectors are sometimes indicated as specific. In case of certain detectors such as of mass spectrometric type, specificity for a specific compound can be set by the operator such as the response to come from a single compound. This procedure can significantly increase selectivity/specificity of an analytical method.

Sensitivity and the stability of sensitivity In general terms, sensitivity is defined as the change in signal output of a detector as reported to the concentration change. The detector generates a signal y for the amount

180

Method Development in Analytical HPLC

qinj of an analyte and for different detectors specific responses y ¼ F qinj are obtained. However, in the chromatographic practice, a specific volume Vinj of sample is injected in the HPLC system and the amount qinj ¼ Vinj $ x where x is the sample concentration (see Eq. 4.5.4). As a result, it is more convenient to consider the dependence of y directly on x (although in fact y depends directly on qinj ) and this dependence is described by a function FðxÞ in the form: y ¼ FðxÞ

(4.4.1)

The response y of the detector is a signal affected by fluctuations, and each measurement y1 ; y2 ; .yn can be slightly different from each other. As a result, the function FðxÞ should be considered the “average” response of the detector to the concentration x of an analyte (for a selected injection volume Vinj ). Based on Eq. 4.4.1, the sensitivity Se of a detector relative to the analyte concentration is defined by Eq. 4.4.2: Se ¼

dFðxÞ dx

(4.4.2)

For a linear dependence of the detector signal on the concentration x, Eq. 4.4.1 has the form: FðxÞ ¼ a $ x þ b ðor y ¼ a $ x þ bÞ

(4.4.3)

For Eq. 4.4.3 describing the dependence, the sensitivity Se ¼ a where a is the slope (a ¼ tan a) of the line showing the dependence of the response versus concentration as indicated in Fig. 4.4.1.

Figure 4.4.1 A modeled shape for a dependence with exaggerated nonlinear response at lower and higher concentrations.

Performance and utilization of HPLC

181

For a wider range of concentrations x for the analyte, it is common that the function F describing the dependence can change, and remains linear only for a specific range of x values as also illustrated in Fig. 4.4.1 with exaggerated nonlinear response at lower and higher concentrations. For a dependence indicated in Fig. 4.4.1, the sensitivity Se ¼ a when the response is linear and has a different value (usually lower) for the low end and the high end of the concentration range. For some detectors, the response is not linear for the entire range of concentration x, and quadratic dependences are common for some detectors. In such cases, sensitivity Se is different in every point of the whole concentration range. Sensitivity also depends on several factors such as analyte properties, sample matrix, mobile phase properties, and also on the injection volume Vinj and on the detector settings (e.g., electronic amplification of the signal). An increase in Vinj increases the sensitivity of the detection by placing more analyte in the HPLC mobile phase, but this volume is limited by other effects as further discussed in Section 4.5. Because sensitivity (as defined by Eq. 4.4.2) can be increased, for example, by electronic amplification, S becomes highly dependent on the instrument settings. Also, the electronic amplification does not increase only the signal of the detector, but also enhances the fluctuations of individual measurements and also the background signal (signal when no analyte is present) and decreases the measurements utility. For these reasons, a specific statement for best settings regarding a detector sensitivity is difficult to make. Sometimes for describing detector sensitivity, the minimum injected mass that generates a useable signal is indicated. However, this parameter is not directly related to sensitivity and reflects another parameter known as limit of detection (LOD) which is further described. Besides good sensitivity, one important quality of a detector is the constancy in time of its sensitivity. For a given concentration x of the analyte, in order to obtain reproducible results, the response FðxÞ must remain constant in time. The stability of sensitivity in time has significant impact in the precision and reproducibility of the results obtained from measurements repeated within a short period. For quantitative measurements in HPLC calibrations are necessary. If the calibration is obtained for the detector with one sensitivity and the measurement of samples with a changed sensitivity of the detector, this can significantly affect the accuracy or the results (how close the results are to their true value). Some detectors have typically a good stability of sensitivity, such as UV-Vis detectors. However, for MS and MS/MS detectors, this stability can be a problem, depending on the instrument model and manufacturer. In part, the detector sensitivity variation can be compensated by using internal standards for analytes quantitation. In order to provide reliable data, the inclusion of control samples and calibration standards in sample sets is strongly recommended.

Limit of detection and limit of quantitation The limit of detection (LOD) is a parameter describing the smallest amount of analyte that can be detected. However, similar to the discussion regarding sensitivity, because

182

Method Development in Analytical HPLC

a fixed volume Vinj is injected in the chromatographic system in a given analytical method, LOD is frequently indicated as the smallest concentration of analyte that can be detected (implying at a fixed Vinj ). The measurements for the signal generated by the detector ys1 ; ys2 ; .ysn for a compound at very low concentration x must be differentiated with a certain “confidence” from the signal yb1 ; yb2 ; .ybn for a blank. Assuming that these results are affected only by random errors, they will follow a Gaussian distribution with the frequency of error occurrence described by Eq. 1.7.8 with the mean m s for the signal and the standard deviation ss . At the same time, the measurements for a blank (x ¼ 0) will generate the signal yb1 ; yb2 ; .ybn and those results will have the mean m b . The standard deviation for the blank can be assumed to be the same ss as for the signal. A specific value of the signal considered as generated by the sample and not being noise (random oscillation of the detector electric output) should be decided for differentiating the signal from the noise. This value indicated as the decision limit Llim can be taken higher than the noise (which has the mean signal m b Þ with the value kss where the parameter k must be determined. The procedure how to take the value of k is further described and the value of Llim can be written as follows: Llim ¼ m b þ kss

(4.4.4)

The probability to obtain signals from the blank higher than the decision limit Llim is given by the expression (see Eq. 1.7.9): ZN a¼

f ðyb Þdyb

(4.4.5)

Llim

The probability to consider a noise as being signal with a value higher than Llim is given by P ¼ a. A value for the one-sided normal distribution that gives a ¼ 0.01 (or 1% if expressed in percent) is obtained for k ¼ 2.33. Therefore, if the signal is higher than m b þ 2:33ss , the probability of false positives (signal from the background to be considered analyte) is 1%. A signal y with m b ¼ Llim has, however, the problem of possibly generating false negatives. The probability of false negatives is given by the expression: ZLlim f ðys Þdys



(4.4.6)

N

In expression 4.4.6, f ðys Þ is a Gaussian with m ¼ m s and the same standard deviation ss . This has the probability P ¼ 0.5 (or 50%) because the normal curve is symmetrical around the mean. Therefore, the possibility of false negatives is very high, and 50% of the true positives will be considered noise.

Performance and utilization of HPLC

183

A higher signal (higher m s ) and keeping Llim ¼ m s will continue to diminish the chances for false positives since a will become smaller as seen from Eq. 4.4.5 with the integration performed for an interval with higher lower limit. However, maintaining Llim ¼ m s does not modify the probability of false negatives. For a chosen probability P ¼ 0.01 of obtaining a false negative, the corresponding value of the signal can be calculated. This value is noted by D, should be smaller than m s , and has the expression: D ¼ m s  kss

(4.4.7)

For the probability b ¼ 0.01, the resulting value for the one-sided normal distribution is k ¼ 2.33. For L ¼ D, the subtraction of rel. 4.4.7 from rel. 4.4.4 leads to the expression: m s ¼ m b þ ðk þ k 00 Þss

(4.4.8)

The result of these considerations is that in order to have a probability of 1% for a false negative signal, and a probability of 1% for a false positive signal (99% confidence in the results), it requires k þ k00 ¼ 4.66, and in this case, it can be written the following formula: m s ¼ m b þ 4:66ss

(4.4.9)

If the signal is considered S ¼ m s  m b and the noise is taken as N ¼ ss , formula 4.4.9 is equivalent with the formula: S = N z 4:66

(4.4.10)

For the detection to be possible in analytical practice, a value of S=N z 3:33 that is slightly smaller than 4.66 was considered to be sufficient (with lower probability than 1% for false negatives, and higher than 1% to give false positives). The value S=N ¼ 3:33 has been adopted in analytical practice for the definition of detection limit or limit of detection, LOD [43,44]. The estimation of S=N ¼ 3:33 in a set of chromatographic measurements can be done directly on the plot of chromatograms (typically using the capabilities of the data processing software of the chromatographic instrument). The concentration (or amount) generating S=N ¼ 3:33 is taken as LOD (although S and N are not concentrations, their ratio can be considered as proportional with concentrations). As previously indicated, the theory supports the selection of the value S=N z 3:33 for LOD, although in chromatographic practice, this definition may generate scattered results. The noise is typically measured as the variation in detector response at the baseline (difference between the minimum and the maximum of the baseline signal), and the signal S for a peak is measured as the difference from the maximum of the peak

184

Method Development in Analytical HPLC

and the average value of the baseline. In HPLC, the mobile phase is continuously flowing in the detector during the chromatographic run and therefore the baseline noise N is measured as the variation of the detector response for a chromatogram without making an injection but having the mobile phase still flowing into the detector [45]. The measure of the variation of the detector response in any part of the chromatogram where no extraneous peaks are present is also common. The place where the noise is measured still remains arbitrary, in spite of some recommendations that the noise should cover a specific range of time, wider than the chromatographic peak of the analyte and on both sides of the peak. In addition to that, the noise can be different from one chromatogram to another for the same method of analysis. A common way to reduce baseline noise is to increase the detector time constant (time allowed for the detector to collect signal). An optimum value for the detection time constant is about 1/10 of the peak width of the narrowest peak of interest from the chromatogram [46]. A better procedure for LOD evaluation is based on repeated measurements of the signal for a very low concentration of a standard and the use of the approximations m s  m b z m and ss ¼ ss (as indicated in Section 1.7). The choice S= N ¼ 3:33 is equivalent with the expression: m ¼ 3:33ss

(4.4.11)

Expression 4.4.11 indicates that for detection to be possible, the signal of the detector at least equal to 3 ss for a set of measurements (at low analyte concentration) is necessary. In the analytical practice, it is important to know the limit of detection for the concentration (or the amount) of the analyte and not for the signal y. For this reason, based on a calibration instead of using the standard deviation for the signal ss , it is common to use the standard deviation s (proportional to ss based on Eq. 1.7.24) for the backcalculated levels of the concentration (in fact of the amount in a certain injected volume Vinj ) of the analyte in the injected sample, the concentration depending on the signal through an expression of the form: x¼F

1

ðyÞ

  1 b x¼ y a a

(4.4.12)

Based on expressions 4.4.11 and 4.4.12 and replacing the standard deviation of the signal ss with the standard deviation s of the calculated analyte concentration (in the injected volume), it results that for a set of measurements at low analyte concentration the definition of LOD should be based on the expression [47e49]: LOD ¼ 3:33s

(4.4.13)

The convention of taking LOD equal with 3.33 versus 4.66 increases the probability P of obtaining a false positive result and decreases the probability P of obtaining a false negative result refereeing to a probability of about 99%.

Performance and utilization of HPLC

185

As indicated by Eq. 4.4.13, the LOD for a detector (or HPLC instrument) is proportional with the measured standard deviation s. For being able to measure low quantity of analytes, as low LOD as possible is highly desirable, and instruments generating low s for samples with low amount of analyte (or low concentration in an injected volume) and with low background noise are highly desirable. Because the amount of analyte for a specific concentration of the injected solution depends on the injected volume (qinj ¼ ½X $ Vinj ), a practical procedure of increasing LOD is the increase of the injected volume of the sample (limitations for the Vinj are further discussed in Section 4.5). In chromatographic practice, it should be noted that the standard solution is not usually injected directly into the detector, but using the HPLC injector and passing the sample through the chromatographic column. For this reason, the LOD evaluated in this way is in fact the LOD of the chromatographic instrument and not of the detector alone. For an analytical method, an LOD is also an important parameter. However, LOD of a method must be reported as the minimum analyte in the sample that can be detected with confidence of about 99.0% and its value should be calculated from the LOD of the HPLC and considering the sample content in the injected solution. Values for the minimum injected mass that generates a usable signal from the detector are indicated in Table 4.4.1 for several common detectors. In analytical practice, an additional parameter is frequently utilized indicating the smallest amount (concentration in the injected sample) of an analyte that can be Table 4.4.1 Ranges of minimum mass detectable and dynamic range for various types of HPLC detectors.

Type of detector UV-Vis spectrometry Fluorescence spectrometry Refractive index Electrochemical amperometric Electrochemical conductometric Mass spectrometry Evaporative light scattering FT-IR spectrometry

Minimum mass injected (typical)

Minimum mass injected (extreme)

Dynamic range of concentrations

Linear range concentrations

0.1e1 ng

0.01 ng

10þ5

10þ4

1e10 pg

10 fg

10þ4

10þ3

100 nge1 mg 0.1e1 ng

10 ng 100 fg

10þ4 10þ4

10þ3 10þ3

0.5e1 ng

0.5 ng

10þ4

10þ3

1 pg - 1 ng

10 fg

10þ4e10þ5

10þ3

5 mg

0.5 mg

10þ3

10þ2

1 mg

0.5 mg

10þ3

10þ3

186

Method Development in Analytical HPLC

measured. This parameter is limit of quantitation (LOQ) and it is defined by the expression: LOQ ¼ 3 LOD ¼ 10 s

(4.4.14)

The choice of having LOQ three times larger than LOD was based on practical considerations. In practice, the evaluation of measured standard deviation s can be performed on one of the most diluted standards, but also in conditions close to the real sample when the matrix may play a role in interfering with the results. Also, in some cases, the concentration of the lowest calibration standard (for a fixed injection volume) can be used as LOQ value. In such cases, it is common to indicate the limit of quantitation with a parameter indicated as practical limit of quantitation (PLOQ). The value of LOQ being based on the standard deviation s of the measurements a low standard provides information regarding the precision of the measurement, which is considered good enough for the quantitation to be performed. However, the LOQ does not provide information about the accuracy of the measurement. Due to a potential deviation from linearity as show as possible in Fig. 4.4.1, a quantitation based on Eq. 4.4.3 established for the linear range of dependence is not applicable for the lower part of the dependence where the value of LOQ may be situated. For this reason, a measurement at LOQ when Eq. 4.4.3 does not indicate true analyte levels, may provide precise result but not accurate.

Dynamic range and linearity The dynamic range of a detector is the range of concentrations x of an analyte that generates a signal useable for measurements. The response FðxÞ of the detector may be linear for a specific range of x, and this range is indicated as linear dynamic range. However, the detector may not have a linear response but still a positive dependence between the signal versus the analyte concentration (or amount), and also an acceptably high S in order to define a dynamic range. A low value for S, approaching S ¼ 0 is sometimes indicated as signal saturation (see Fig. 4.4.1). The dynamic range of a detector can be independent on the nature of the analyte (for universal detectors), but also can be highly dependent on the analyte nature. The dynamic ranges of several types of detectors are given in Table 4.4.1 For the detectors, an estimated maximum mass of analyte can also be obtained based on the detector dynamic range. However, this level strongly depends on the nature of the measured compound. Some common such maximum levels of analyte are indicated in Table 4.4.2 for several detectors.

Precision and reproducibility The precision of the response of the detector to the same injected sample is an important characteristic and indicates how close the measurements are to each other. As indicated previously, for a number of measurements, the average of the response m s will

Performance and utilization of HPLC

187

Table 4.4.2 Estimation of maximum levels of analyte that can be loaded for various detectors to obtain a reliable response (the level may be very different from compound to compound).

Type of detector

Max. Estimated mass injected (typical) (mg)

UV-Vis spectrometry Fluorescence spectrometry Refractive index Electrochemical amperometric Electrochemical conductometric Mass spectrometry Evaporative light scattering FT-IR spectrometry

20.0 0.05 500.0 1.0 1.0 0.2 250.0 100.0

have a standard deviation ss . It is common for the value of ss to be reported to the average signal (usually multiplied by 100) and to generate a relative standard deviation of the signal, defined by the formula:   RSD% ¼ ss = m s 100

(4.4.15)

The low values for RSD% indicate good precision. Precision is affected by many factors, one such factor being the instability in time of detector sensitivity. Reproducibility and precision of the response of a detector are close related characteristics, but precision is referring to the variation within a set of measurements, while reproducibility is referring to the variation between different sets of measurement performed at some different points in time. For an analytical method, some differentiation is also made between reproducibility and repeatability. Various procedures were developed for accounting for systematic errors in detection [50].

Baseline noise, noise, and drift The baseline noise N is the variation in the response of a detector when a blank sample (or no sample) is injected, but still having the mobile phase is flowing into the detector. Although a specific threshold for the sensitivity of the detector can be set in some detectors, and the detectors from different manufacturers may have included in the electronic system of signal processing specific types of data filtering and smoothing, baseline noise still exists. Since the standard deviation for the signal at low concentrations is assumed to be the same as that for a blank sample, a low baseline noise is leading to a low standard deviation s value in Eq. 4.4.13 and consequently a low LOD which is highly desirable. However, not only the baseline is affected by noise. For some detectors, such as MS detectors, it is possible to have noise along the whole chromatogram. Noise depends on the quality of detector construction (e.g., its electronics)

188

Method Development in Analytical HPLC

but also on some detector parameters. One such parameter is the detector sensitivity and high sensitivity may generate higher noise N . Another such parameter is the time spend by the detector to accumulate signal (dwell time). In MS instruments, for example, a detector response (data point) is not produced continuously but at certain intervals of time. During the dwell time, the mass spectrometer accumulates signal for a specific ion. A longer dwell time reduces the noise since the signal is averaged for a longer period of time, but produces fewer data points of measurement. The noise in the data representing a chromatogram can also be reduced by data processing. Different algorithms, such as the Savitzky-Golay filter, Whittaker smoothing, Gaussian smoothing, have been proposed to smooth a chromatographic trace, by numerically eliminating the excessive noise. The Savitzky-Golay filter, for example, uses a moving window approach with a selected number of experimental data points in that window. From the data in the window, a polynomial curve (usually of low order) is generated using the least square approach to obtain the best fit for the data points. After this, the experimental central point of the window is replaced with the corresponding central point that fitted polynomial curve. The following window is using experimental data including all the previous ones except the first point in the previous window and adding the next experimental point from the chromatogram. The process is repeated along the whole chromatogram without distorting the signal tendency [51,52]. Other disturbances in the detector signal may include long-term noise and the drift. The long-term noise appears as a fluctuation of the signal with a wider frequency. The drift is a continuous variation (increase or decrease) of the baseline signal for a period of time comparable with the length of the chromatogram. Drift may be caused not by the detector itself, but by changes, for example, in the composition of the mobile phase during gradient elution.

Suitability to a given mobile phase, use of gradient, and flow rate of the mobile phase The mobile phase is a main contributor to the separation process, and many types of mobile phases are utilized in HPLC. As a consequence, the detectors for HPLC are exposed to a variety of mobile phases, this including a variety of solvents, of mobile phase compositions, and of the presence of potential additives. The mobile phase nature and composition may affect the detector response, but also may affect the detector functionality. Also, depending on the use of isocratic separations or gradient separations, the composition of the mobile phase is kept constant or is changed during a chromatographic run and this may affect the detector response. Some detectors such as refractive index ones cannot be used when the mobile phase composition is changing. Other detectors such as mass spectrometers and ELS must have all components in the mobile phase volatile at a specific temperature. The detector response is also in most cases dependent on mobile phase composition and show changes in response during gradient separations. Also, detector components must be resistant to the solvents used in the mobile phase, as well as to the acid or basic character of the mobile phase.

Performance and utilization of HPLC

189

Besides the nature of mobile phase, detectors may have limitations regarding the mobile phase flow rate U. For some detectors, there is a limitation of how much flow can be submitted to the detector. For example, the flow in an MS detector is usually kept at a maximum of 1.0 mL/min. For miniaturized systems, flow rate is typically very small, such as 0.1 mL/min or lower. For such slows, specialized equipment is usually necessary, or special diverting valves.

Volume of measuring cell The capability to make detection in a small volume of sample is usually related to the volume of a flow-through cell, or other parts in the instrument construction. For UVVis detectors, for example, instrument manufacturers may offer cells of different dimensions, with longer path of the beam of light in the flow-through cell for achieving higher sensitivity of the instrument, but with slightly larger cell volume, or cells with shorter path for the light beam, but also with smaller volumes. Microcells are also available for certain instruments that are designated for micro-HPLC systems. Special cells are also necessary when using capillary HPLC columns and low flow rates.

Low contribution to peak broadening The capability to not contribute to peak broadening is also related to the HPLC construction. The absence of dead volumes, small flow-through cells, laminar flow within the detector, and adjustment of the total volume of mobile phase in the detector in accordance with the range of injection volume for the sample are some of the steps toward achieving little or no contribution to peak broadening.

Dependence of response to changes in the flow rate of mobile phase For all detectors, the flow rate of the mobile phase influences the detector response, and in the chromatogram, a larger peak area is generated at lower flow rates. Because the quantitation in HPLC is based on the chromatographic peak area (see also Chapter 4), the increase of the peak area response of the detector strongly modifies the results of the quantitation. The peak area is larger because the analyte molecules spend more time within the detector (e.g., in a flow cell where a property is measured). This can be easily shown assuming for simplification that the analyte arrives at the detector as a well-defined square plug of length l (and not having a Gaussian distribution as it is in reality). Assuming that the detection of the analytes in the plug takes place for the residence time period tres of the analyte molecules in the detector, the detected peak area is A ¼ Itres where I is the signal intensity. On the other hand, the residence time tres in the cell is tres ¼ l =u where u is the linear flow rate of the mobile phase. With these approximations, the peak area will have the expression: A ¼ I l =u

(4.4.16)

190

Method Development in Analytical HPLC

Figure 4.4.2 Change in the peak area as linear flow rate u decreases for a chromatographic peak from an initial value Ct to Ct=2 and to Ct=4, with hypothetical square peak shape (the same is valid for the Gaussian peak shape).

Eq. 4.4.16 indicates that peak area is proportional with 1=u and therefore increases when u is smaller. An illustration of the enlargement of the peak area expressed by formula 4.4.16 is given in Fig. 4.4.2. In spite of the utilized approximation for the peak shape, the proportionality of the peak area with the inverse of the linear flow rate u is valid for a Gaussian shape of the peak and it was experimentally verified, for example, for UV type detection [53]. In addition to the increase in peak area solely because of the decrease of flow rate, other effects take place as the flow rate of the mobile phase decreases, such as changes in the Gaussian peak shape due to longer residence time of the analyte in the chromatographic column, etc. However, the increase in the response still takes place although is not rigorously proportional with the inverse of linear flow rate u. In addition, in some cases, higher flows beyond an optimal point may be detrimental to the sensitivity of the detector such as in case of MS detectors. In MS detection, the ionization efficiency is dependent on the flow rate and a decreased flow rate may significantly increase sensitivity (see Section 3.4). The difference in the flow rate may also affect the baseline stability such as for refractive index detectors. Because of the dependence of detector response on the flow rate, the quantitative analysis which are performed based on calibrations must have the calibration and the measurement performed at the same flow rate, and for the same reason, the flow in HPLC must be kept as constant as possible. A flow rate optimization may be recommended for some detectors.

Frequency of data collection Most modern detectors used in HPLC do not generate a continuous signal based on measured physicochemical property. The signal generated by the detector is typically obtained at short time intervals and a measurement frequency (sometimes indicated as sampling frequency) is characteristic for each detector. It is possible in some instruments that such frequency can be set by the operator, but in other detectors, this

Performance and utilization of HPLC

191

Figure 4.4.3 Two peaks recorded with two different frequencies but described by the same Gaussian function.

frequency is fixed. It is important for the measuring frequency of the signal to be high enough for obtaining an accurate representation of chromatographic peak. A sampling frequency of about five points per second could be sufficient for peaks with 4e5 s width (or larger). However, for narrow peaks, the sampling frequency must be higher such as 20 or 50 Hz. The curve representing the peak shape is generated by connecting each measurement point, and sparse points do not account properly for this shape and introduces errors, in particular regarding peak areas. This is exemplified in Fig. 4.4.3 for two peaks recorded with two different frequencies but described by the same Gaussian function. For some detectors, the frequency of generating a measurement point is set by the instrument manufacturer to generate a modulated signal. For other detectors, this frequency is not set arbitrarily, but determined by the measurement procedure. For example, in MS detectors, a data point (with a specific M=z value) is generated as the average of the signal accumulated during the dwell time. A longer dwell time reduces the noise, and generates a more accurate result. However, the longer dwell time also reduces the number of points that can be generated across the chromatogram. In such cases, a compromise must be chosen between the desired dwell time and the number of points necessary for correctly representing the peak shape [54].

Backpressure accepted and the one generated by the detector Some detectors, in particular those using flow-cells, have a limited capability regarding the maximum backpressure they can accept. This value must be well known and not exceeded. If the maximum backpressure is exceeded, the flow cell can be damaged. The backpressure accepted as well as the one generated by a detector is important in particular when more than one detector is present in the HPLC system. For example, UV-Vis and fluorescence detectors are frequently coupled in series, although not necessarily used simultaneously. Since the flow through a detector may pose some backpressure, when using detectors coupled in series, it should be verified that the flow-cell of the first detector can handle the backpressure generated by the second detector (and the connecting tubing). For the detector down-stream, it should be verified that undesirable peak broadening does not occur because of the up-stream detector. The coupling of detectors in parallel is also possible, but care must be taken to assure

192

Method Development in Analytical HPLC

that appropriate flow goes through all the detectors. Since different detectors may pose different backpressures to the flow, the risk exists that most (or even all) of the flow goes to a single detector. Several detectors used in HPLC are further described.

The capability to not alter the analyte In some detectors, the mobile phase containing the analytes are just flowing through a measuring flow-cell where a property such as the refractive index, the UV absorption, or the fluorescence is measured. Such detectors can be used followed by a second detector if necessary, considering only the backpressure generated by the second detector such that to not exceed the acceptable backpressure for the flow cell. Also, such detectors can be utilized being followed by a fraction collector in case the separated analytes should be collected and used for further investigation. Other detectors such as MS (and MS/MS), ELSC, and CAD are “consuming” the sample. This type of detector is usually an evaporative type, and after entering the detector, the analytes cannot be recovered.

The case of more than one detector to select In many laboratories, more than one type of detector is available and the selection must be made not only based on the detector properties but also based on the analysis requirements and the characteristics of the separatory component of the HPLC. In many applications, when the separation is good, the analytes have chromophores and the lowest concentration to be measured is above 0.5 mg/mL, and the detection of choice is usually UV. However, fluorescent analytes usually show better sensitivity when fluorescence detection (FLD) is utilized. The excellent sensitivity of MS detection, associated with better selectivity compared to other detector types, makes the MS and in particular MS/MS a highly preferred choice when such detectors are available. However, if a high sensitivity is not necessary, the separation is good, and UV detection is also applicable, it can be better to use UV detection which is typically more robust as compared to MS detection. The RI detection is useful when the analytes do not have chromophores and MS detection is not available. Selection of two detectors such as UV followed by MS provides valuable information combining in some cases the high reliability of UV detection with the selectivity and capability to provide qualitative information of the MS detector.

Key points • • • •

The detector is part of a system that also includes a separating module and the detector characteristics should be considered in relation to the whole HPLC system. A variety of detectors are used in HPLC and specific properties can be used for detector characterization. The capability for qualitative identification of detectors ranges from no such capability for universal detectors to advanced identification capability. The most useful parameter for the characterization of capability for quantitative measurement of detectors is the limit of quantitation (LOQ).

Performance and utilization of HPLC

4.5

193

Parameters related to quantitation in HPLC

HPLC analysis is used for both qualitative and quantitative analysis; however, the majority of applications of HPLC are related to quantitative measurements. The success for a quantitative analysis depends on both the quality of separation, as well as of quality of detection. Several parameters used in HPLC related to quantitation are further discussed.

Peak characteristics related to quantitation As previously indicated, the quantitation in HPLC is usually based on the measurement of the peak area Apeak of the compound to be quantitated. The use of Apeak for quantitation is based on Eq. 4.1.36 that gives the profile of the chromatographic peak as a function of time: bqinj hðtÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffi exp 2pst 2



ðt  tR Þ2 2st 2

 (4.5.1)

In Eq. 4.5.1, hðtÞ is the peak height at moment t, and describes the shape of the chromatographic peak, b is a parameter dependent on the detector response on which the peak is recorded, qinj is the amount of analyte injected in the HPLC, st is a parameter related to peak broadening (st ¼ 0:25 Wb where Wb is the peak width at the baseline), and tR is the retention time of the evaluated peak. The integration of the function hðtÞ for t between N and þN is equal with the area under the curve, and as a result, the following expression is valid: ZþN N

bqinj hðtÞdt ¼ Apeak ¼ pffiffiffiffiffiffiffiffiffiffiffi 2ps2t



ZþN exp N

 ðt  tR Þ2 dt 2s2t

Because the integral of the exponential in Eq. 4.5.2 is equal with can be written in the form: Apeak ¼ bqinj

(4.5.2) pffiffiffiffiffiffiffiffiffiffiffi 2ps2t , the result

(4.5.3)

The amount of material injected in the HPLC is typically contained in a volume of sample Vinj of a specific concentration ½X. The dependence of qinj on the injected volume Vinj is given by the known expression (see Eq. 3.4.5): qinj ¼ ½XVinj

(4.5.4)

Eqs. 4.5.3 and 4.5.4 show that the peak area is determined by the concentration of the analyte, the injected volume of sample, and the detector response b. As a

194

Method Development in Analytical HPLC

result, the concentration ½X can be obtained based on the value of Apeak ðXÞ from the expression: ½X ¼

Apeak b Vinj

(4.5.5)

 With the notation 1 bVinj ¼ a, Eq. 4.5.5 can be written in the form: ½X ¼ a Apeak

(4.5.6)

In the case of many dependences, due to the background signal of the detector, a constant b must be added to Eq. 4.5.6 leading to the following expression: ½X ¼ a Apeak þ b

(4.5.7)

The detector response is not always linear (or can be considered linear only for a small interval of concentrations for) and the dependence of ½X on Apeak is in some cases better described by a quadratic form (for some detectors even other dependences are possible): ½X ¼ a A2peak þ b Apeak þ c

(4.5.8)

Not only the peak area is important for quantitation in HPLC. Peak shape (peak width Wb ) also plays an important role, and this is illustrated in Fig. 4.5.1. In this

Figure 4.5.1 An idealized chromatographic profile showing two peaks with the same area but of different widths, as well as baseline noise.

Performance and utilization of HPLC

195

figure, two peaks of the same area Apeak are shown, with peak A having a small Wb (Wb ¼ 0.4 min in Fig. 4.5.1) and peak B with Wb ¼ 2.2 min. The narrow peak generates a much higher signal to noise ratio (S =N Þ, while for the same peak area, the broad peak generates a smaller (S =N Þ. In some cases, a broad peak, even having a considerable area, may not even fulfill the condition of Eq. 4.4.14 for the LOQ > 10 s, or even the condition of Eq. 4.4.13 for LOD > 3.33 s (s is the measured standard deviation). By approximating the peak shape with a triangle, Apeak ¼ 1=2 W b h and for a given value of Apeak , a lower Wb implies a higher h and therefore a higher S= N . One additional advantage of narrow peaks in a chromatogram is related to the procedure for peak detection and electronic area measurement in modern instrumentation. The peak detection is based in modern instruments on measuring the slope of the chromatographic profile, and for an established threshold for the slope, the start of a peak and the end of a peak are determined. The peak area is then obtained by numerical integration. A steeper profile allows a more precise detection of peaks and consequently a better evaluation of peak areas. Eq. 4.3.9 shows that a column with larger N generates a narrower peak, and continuous effort is made in the construction of modern columns to have higher and higher values for N [55,56].

Limitations for the sample volume and amount injected in the chromatographic column In HPLC practice, common volumes for the injected sample are between 1 mL to 25 mL, and in UPLC, common volumes are between 1 mL and 10 mL. For microHPLC, the injection volumes are usually in the range of 20e500 nL. Large volume injections up to 1 mL can also be used in special applications [57,58]. The amount qinj of material injected with those volumes depends on the sample solution concentration ½X and Vinj (Eq. 4.5.4). Some evaluations for the minimum and maximum qinj and Vinj in a chromatographic system are further given. The lower limit for qinj is typically determined by the detector LOQ (for quantitation) or LOD (for detection). Eq. 4.5.4 indicates that when a larger amount of analyte is necessary for obtaining a higher signal, either a more concentrated sample solution or a larger volume of sample must be injected. As the concentration of the sample solution may be limited by other factors, an increase in qinj can be obtained using larger injection volume. However, limitations to this volume also exist and this limitation will be further commented. The upper limit for qinj is determined either by the column properties or by detector capabilities (see Table 4.4.2), or by both. The overloading of the column with sample leads to distorted shape of the peaks, usually with tailing, and to changes of the retention times toward shorter values. For the detector, the overloading is manifested by the decrease in sensitivity and deviations from linearity of the response (see Fig. 4.4.1). A simple solution to avoid the overloading of the column or of the detector would be to dilute the sample injected in the HPLC, or to inject a smaller volume Vinj (or both). Although this dilution (or lower Vinj ) may resolve the problem of overloading for a specific analyte X, in the case of analyses performed for multiple components, the dilution and small volume injection will also diminish the injected amount for all the other components of the sample. If some of these components are at the low end of content in

196

Method Development in Analytical HPLC

the sample, the dilution or low injection volume is not anymore recommended, and as high amount of sample as possible should be injected for allowing the analysis of low concentration components. Therefore, it is important to know which is the highest amount of sample that can be injected in an HPLC and still obtain good results. The typical effect of column overloading is the decrease in the theoretical plate number of the column (due to peak broadening). Based on this observation, the sample loading capacity of a column (also indicated as the saturation capacity of the stationary phase) is indicated as the maximum amount of sample that leads to a theoretical plate number Ninj reduced by 10% from the maximum plate number N0 [59,60]. The evaluation of maximum acceptable volume that can be loaded in a chromatographic column starts with the evaluation of the volume of mobile phase passing through the column during an ideal peak for which the width of the injected “plug” is considered negligible (as it was done in the development of peak shape formulas in Section 4.1). For such injection, the peak volume is given by the expression: V0;peak ¼ Wb U

(4.5.9)

To the V0;peak given by Eq. 4.5.9 corresponds a theoretical plate number given by the formula: N0 ¼

16 ðtR UÞ2 2 V0;peak

(4.5.10)

For an injected sample with the volume Vinj , the peak volume increases and the new volume can be approximated using the expression [61]:  1=2 2 2 Vpeak ¼ 1:333 Vinj þ V0;peak

(4.5.11)

In expression 4.5.11, the coefficient 1.333 is an empirical value to account for additional “plug” broadening during the injection. From Eq. 4.5.11, a new Wb can be calculated based on Eq. 4.5.9, and a new Ninj can be calculated from Eq. 4.5.9. This leads to the formula: Ninj ¼

16 ðtR UÞ2 2 þ V2 1:333 Vinj 0;peak

(4.5.12)

Eqs. 4.5.12 and 4.5.10 allow the evaluation of the decrease in the column efficiency N when Vinj is added to the peak volume. The ratio is given by the expression: 2 V0;peak Ninj ¼ 2 þ V2 N0 1:333Vinj 0;peak

(4.5.13)

Performance and utilization of HPLC

197

 Based on Eq. 4.5.13, for a ratio Ninj N0 ¼ 0:9 (corresponding to a loss of efficiency of 10%), it can be calculated the following result: Vinj z 0:29 V0;peak

(4.5.14)

An estimation of V0;peak is now necessary for the estimation of an injection volume Vinj that would affect only by 10% the peak width. Based on Eq. 4.5.9 for Wb on Eq.

 4.1.22 for tR ðtR ¼ t0 ð1 þk0 ÞÞ and from Eq. 4.1.6 for t0 (t0 ¼ ε p d2 L 4U), the value for V0;peak is the following: V0;peak z ε pd2 Lð1 þ k 0 Þ = N 1=2

(4.5.15)

Taking ε ¼ 0.7, Eq. 4.5.15 leads to the following value for Vinj that reduced N0 by 10%: Vinj z 0:6 $ d 2 Lð1 þ k0 Þ = N 1=2

(4.5.16)

A few examples of column dimensions and properties with the maximum allowed injection volume required to achieve less than 10% change in the optimum column efficiency (as measured by N) and assuming k0 ¼ 1 are given in Table 4.5.1. Although the increased injection volume may produce a reduction of column efficiency, larger volumes than recommended are sometimes used when the resolution is still satisfactory and an increase in analysis sensitivity is necessary. Deviations from the maximums indicated in Table 4.5.1 are sometimes very significant and the results can still be acceptable. The amount of material injected in the chromatographic column is an additional parameter that affects the separation. Excessive amount of sample in the chromatographic column leads to stationary phase overload and peak shape distortion. The

Table 4.5.1 Examples of maximum injection volume of sample in a column with the condition of not decreasing efficiency with more than 10% (assuming k0 ¼ 1), calculated by Eq. 4.5.16. Length L (mm)

Diameter i.d. d (mm)

Efficiency N

Max V inj mL

150 150 100 100 50

4.6 3.0 4.6 2.1 2.1

9000 9000 10,000 10,000 8000

40.1 17.1 25.4 5.3 3.0

198

Method Development in Analytical HPLC

amount (in mg) that can be loaded in an analytical chromatographic column can be roughly estimated using the expression (e.g., Ref. [22]):

x ε p d 2 L ð1 þ k0 Þ qinj ¼ 4 N 1=2

(4.5.17)

In Eq. 4.5.17, x is a constant in mg depending on the nature of the stationary phase, k 0 is the retention factor, N is the number of theoretical plates of the column, d is column diameter, and L (in mm) is the column length. For conventional analytical columns, x can be between 0.02 and 0.2 mg, but the value may vary considerably depending on the nature of stationary phase. Larger columns may accommodate more mass of analyte. Eq. 4.5.17 seems to also indicate the preference for a lower number of theoretical plates N for achieving a larger qinj value. However, this is caused by the fact that larger N are more affected than lower ones by excessive loading, and it is not recommended to use columns with lower N to achieve larger qinj . As indicated in Section 4.4, the limitation on the amount of analyte that can be injected in the HPLC is also limited by the detector potential overloading (see Table 4.4.2).

Key points • • •

Peak area is used for quantitative measurements in HPLC. Peak shape is also important for quantitation accuracy and precision, and narrow peaks are better for obtaining better quantitation. A maximum accepted amount of analyte and a maximum accepted volume of sample can be estimated for an injection, although the calculated numbers are frequently under estimates.

4.6

Utilizations of HPLC in chemical analysis

HPLC is a very versatile analytical technique and has numerous applications for both qualitative and quantitative measurements, although the quantitative analysis is the main field of applications. Several analytical types of applications of HPLC are further commented.

Application of HPLC in qualitative analysis Qualitative analysis is based in HPLC on two different types of information, one type using the information obtained from the separation, namely the retention times tR , and the other from the information generated by the detector. The two types of information are frequently combined. The retention time tR in a chromatogram depends on the nature of the compound and on the separation conditions (chromatographic column dimensions, nature of stationary phase, of the mobile phase, flow rate of the mobile phase, etc.) and when keeping the separation conditions constant, the retention time

Performance and utilization of HPLC

199

of a specific compound is always the same. Using known standards, this specific retention time can be used for compound identification (in fact verification of presence). This procedure is very useful in routine applications, but cannot be used for unknown compounds identification, and because more than one compound may have the same retention time, it may lead to incorrect results. The lack of interferences from the sample matrix in the region where the expected analytes are eluting is an important requirement. The usefulness of retention time for providing qualitative information depends on the quality of the HPLC separation including stability of the values of retention times, and good resolution Rs of the separation. However, just the use of retention time as a parameter can provide in some cases valuable information even about a very detailed molecular structure. For example, some structural changes of peptides can be monitored by RP-HPLC [62]. The second type of information used for qualitative analysis is that provided by the detector used in HPLC. The universal detectors provide similar signal regardless of the nature of the compound, and their signal cannot be used for compounds identification. The signal from the selective detectors contains information related to the nature of the analyzed compound. This may be in the form of the UV spectrum corresponding to a peak in the chromatogram (in case of UV-Vis detectors), the excitation and emission wavelength (in case of fluorescence detectors), the molecular ion (in case of MS detector), the molecular ion and some fragments generated from it (in case of an MS/MS detector), etc. Most identifications which are using the detector signal are based on comparisons of the spectrum of an unknown with spectra from dedicated spectral libraries (MS spectra, UV spectra, etc.). The development of large libraries with standard spectra and of algorithms for automatic library searches (e.g., Ref. [63]) that are performed using computer programs made the use of these tools for spectra interpretation routine. The use of spectra generated by various on-line detectors in HPLC, although very useful for qualitative analysis, is not always sufficient for some unknown identifications. The UV-Vis spectra, for example, are more useful for checking the presence of a known analyte rather than for the identification of an unknown compound generating an HPLC peak. More informative are the MS spectra and even more, the tandem MS spectra (e.g., Ref. [64]), although even in this case the outcome can be influenced by the separation conditions [65]. In case of using high-resolution MS detection, the structural information can be enhanced by the possibility of determining the elemental composition from accurately measured M=z value and by using mass spectral databases [66e68]. Progress in using MS identification of unknown compounds has been also obtained by using very high accuracy in mass measurement for the molecular ion of the analyte and for its fragments (e.g., using orbitrap or cyclotron technologies). Also, specific computer programs (Mass Frontier, MassBank data, SmileMS) provide help for the identification of unknown compounds. Such programs were also developed for metabolite identification (MAGMa, CFM-ID, MetFrag, MIDAS) and have been applied on a large MS/MS dataset derived from METLINdMetabolite and Chemical Entity Database, which is a data management system aiming to assist in metabolite and chemical entity identification providing public access to its repository of comprehensive MS/MS Data [69]. One field where LC-MS/MS is an essential tool of investigation is bottom-up proteomics. The peptide fingerprinting of protein digests

200

Method Development in Analytical HPLC

uses LC-MS/MS as an essential tool for protein structure investigation. LC-MS/MS applications are continuous expanding such as for the identification of peptide biomarkers for various pathogens. In some cases, the MS detection is associated with other types of detection including UV and circular dichroism [70]. For some unknown identifications, the detector response such as MS following an HPLC separation is further combined with results generated by isolation and collection of the unknown compounds and further attempts for identification using techniques such as Fouriertransform infrared spectroscopy (FTIR) and nuclear magnetic resonance (NMR) [71]. In conclusion, the utility of HPLC in qualitative analysis covers a wide range, from not being capable of providing qualitative information to the identification of unknown compounds and even to the elucidation of specific structures such as the protein primary and secondary structure.

Application of HPLC in quantitative analysis The main use of HPLC in chemical analysis is the performing of quantitative measurements of the analytes. Depending on the type of detector used, the quantitation is possible for a specific range of concentrations and with a minimum amount of analyte, values described in Table 4.4.1. As shown in this table, for example, the mass spectrometric detector or fluorescence detector can measure a minimum amount of some analytes as low as 10 fg. Such low amounts are not always necessary to be measured, and HPLC is used in many practical applications for quantitation in wide ranges of concentrations. However, for concentration measurement in practice, it is necessary to perform a calibration. The calibrations consist in establishing an empirical relation, which permits the calculation of the values of the amount of a substance in a sample, from the measured values of the analytical signal. The subject of calibration is very important for obtaining the correct information regarding the levels of the analytes in a sample, and is presented in a number of dedicated studies [72,73]. As indicated by Eq. 4.5.6 ½X ¼ a Apeak for determining ½X when Apeak ðXÞ is known from measurements, the parameter a must be known. Similarly, for the dependence described by Eq. 4.5.7 ½X ¼ a Apeak þ b, parameters a and b must be known, and for the dependence described by Eq. 4.5.8 ½X ¼ a A2peak þ b Apeak þ c, parameters a, b, and c must be known. Values for those parameters can be obtained using several procedures. The most common such procedure is the use of a set of standard solutions of the analyte at known concentrations f½Xi i ¼ 1; 2; .:ng (calibration standards) that are injected at constant volume (the same as the injection volume for the sample), followed by the measurement of the corresponding chromatographic peak areas fAi g for each standard. The range of selected concentrations for the analytes ½X1 to ½Xn should be selected to cover the range of the expected values for the concentrations of the analytes in the samples. This range may not be known, and should be established by preliminary trials. It is also possible that the analytes concentrations in the injected samples must be adjusted by sample preparation or dilution to fit the calibration range. Also, the highest concentration level used for calibration should not exceed the maximum analyte concentration acceptable by the detector (see Table 4.4.2).

Performance and utilization of HPLC

201

It is common that the calibration solutions are made starting with a stock solution with a concentration higher or equal with the first standard. From this solution, a Standard 1 having the highest concentration is made. Sequential dilutions to generate 1/2, (1/2)2, (1/2)3, (1/2)4, ..(1/2)10, dilutes solutions are made. In this way, a range of about three orders of magnitude in concentration is generated, and 10 calibration points. This number can be reduced if necessary, but less than five calibration points are not recommended. The lowest calibration standard can be selected equal with the limit of quantitation LOQ, or higher, if samples do not contain analytes below the lowest standard. The upper concentration of a standard must be selected at levels of about 1.1e1.4 times higher than the highest concentration of the analyte in the injected sample. However, this upper level can be limited by the upper concentration where the detector response shows sufficient sensitivity (see Fig. 4.4.1). In case of nonlinearity of the detector becoming significant at a certain concentration, the standards must be selected below that value. Each standard must be injected a number of times (e.g., three times) to generate peak areas for each concentration representing a point ½Xi ; Ai . The calibration curve is obtained by plotting the system of points f½Xi ; Ai g and obtaining the parameters for the equation of the trendline using, for example, least-squares fitting. For generating the parameters of the dependence described by Eq. 4.5.7 (½X ¼ a Apeak þ b), the calibration capability of the software of the analytical instrument can be used (e.g., MassHunter program from Agilent), or the LINEST function available in Microsoft Excel (see Section 1.7). For a linear calibration, the parameters a and b are obtained using the expression: ¼ LINESTð½X1 : ½Xn  w1 : wn ; ðA1 : An e 1E⎼99Þ ^f0; 1g  w1 : wn ; FALSEÞ (4.6.1) For a quadratic dependence, parameters a, b, and c (Eq. 4.5.8) are obtained using the expression: ¼ LINESTð½X1 : ½Xn  w1 : wn ; ðA1 : An e 1E⎼99Þ ^f0; 1; 2g  w1 : wn ; FALSEÞ (4.6.2) In Eqs. 4.6.1 and 4.6.2, the values wi represent the weight of importance for each pair of values ½Xi ; Ai . For a calibration representing well the low concentrations, higher values for wi can be selected, while for higher concentrations, the weight can be selected 1.0. Verification of a good calibration and back-calculation of the standard concentration using the calibration equation is an important step in developing a reliable method of analysis. The back-calculation of the concentration of the lower standards should not differ with more than 3%e5% from the real concentration of the standard. In some analyses, even if the calibration generated very good results for the standards with higher concentrations, large deviations are noticed for the lower standards due to the possible nonlinear response (or the response not following the quadratic dependence), as shown in Fig. 4.4.1.

202

Method Development in Analytical HPLC

The establishment of LOQ for analytical method is recommended to be based on the value of 10 s as indicated by Eq. 4.4.14. The precision of measurement of the lowest standard may be in some instances good enough such that the value of 10 s is lower than the lowest standard concentration ½Xn . However, if a large deviation between the back-calculated value of ½Xn based on the calibration and its true value is large, the LOQ is not providing a reliable information regarding the lowest concentration that can be measured accurately. The calibrations can be done with pure solutions of the analyte, although the calibrations can be also obtained by adding different levels of the calibration compound to a blank sample that does not contain the analyte. This procedure makes the analysis of the samples containing the calibration standards as close as possible to the analysis of a real sample and allows the subtraction of the overall influence of the matrix in the analysis. A blank sample is however not always available and a synthetic matrix can be used to mimic the real matrix. Several aspects related to the generation of calibration curves (lines) should be further commented. One is related to the range of concentrations f½Xi i ¼ 1; 2; .:ng utilized for calibration. As indicated in Section 4.4, a detection remains sensitive enough only for a specific concentration range (see Fig. 4.4.1). Calibrations outside this range is not useful. Another aspect is related to the shape of calibration (in the usable calibration range). The most common type follows Eq. 4.5.7 (linear dependence), and the values of the slope a and intercept b can be used for the calculation of an unknown concentration ½X when the corresponding area A for X is measured. Some characteristics of linear dependence may indicate, for example, a loss of analyte when parameter b is negative, or a background interference when b is large (positive). For some calibrations, the value A ¼ 0 for ½X ¼ 0 is included as a “known” calibration point. Forcing the calibration dependence to have b ¼ 0 is also possible, but not recommended. Depending on the requirements of the analysis, the linear range of dependence may remain sufficient for the analysis of all samples. However, quadratic dependencies are also very useful and are commonly used for a better fit of experimental data (the system of points f½Xi ; Ai g), for example, in MS detection. At the same time, quadratic dependencies can be used only as long as sensitivity is still good. Different other dependencies such as log½X=log A can be used in some cases such as for evaporative light scattering detector (ELSD) detection. The stability in time of the calibration standard solutions must be known and the solutions must be perfectly stable during the calibration process. However, the calibration must be repeated from time to time, either with the whole set of calibrants, or only with part of them. In such cases, the stability of the calibration set of solutions must be known. If the calibration solutions are stable (e.g., being kept at low temperature), the same set may be used after a short period of time. Making a new set of calibrants and verification of the calibration is recommended after a certain period of time that should be decided for the laboratory depending on factors such as standards stability, protocols required for the laboratory, etc.

Performance and utilization of HPLC

203

When sample preparation leads to changes in the analyte concentration compared to that from the initial sample and these changes cannot be measured, or when changes in signal intensity are present due to factors such as the variation in the composition of the matrix of the injected sample, the calibration using only the peak areas of the analyte’s standards may not be sufficient for obtaining analyte quantitation in some cases. In such cases, the inclusion of an internal standard (I.S) is necessary for calibration. The I.S. can be utilized for comparison with only one of the analyzed compounds (an I.S. assigned to a specific compound), and in case of a multicomponent sample, more than one I.S. can be used. It is also possible that one I.S. is used for comparison with a group of analytes, or with all analytes in the sample. Internal standards are compounds that are absent in the real samples and are added in a constant amount at a chosen point during the analysis for verifying the reproducibility, accounting for sample losses, and in sample preparation for accounting for changes in the concentration from the raw sample to the processed sample when the I.S. undergoes the same process as the analyte. Together with the measurement of areas Ai of the analyte standards, the measurement of the area AIS of the peak of peak standard should be done in such cases. The use of the I.S. for the correction for changes during sample preparation or to mimic instrument variability is done by replacing the use of peak area Ai in Eqs. 4.5.6 and 4.5.7, or 4.5.8 with the normalized peak area of the analyte Ai =AIS . Since the I.S. is added in a constant known amount, the ratio Ai =AIS remains proportional with Ai . In this way, the quantitation following, for example, the dependence expressed by Eq. 4.5.7 becomes: ½X ¼ a0

Ai þ b0 AIS

(4.6.3)

Similar adjustments are made for other dependencies. The normalization in Eq. 4.6.3 is done with the area of the I.S. designated for comparison of each specific analyte (a unique I.S., or an I.S. assigned to a specific compound). The calculation of parameters a0 and b0 (or a0 , b0 , and c0 in case of quadratic dependencies) can be obtained using Eq. 4.6.1 or Eq. 4.6.2, respectively, where the areas Ai are replaced with Ai =AIS . In some analyses, the best fit for the experimental points is quadratic, although a linear dependence does not deviate to much from the data points. An example of two possible type of dependence is given for an analysis of several sugar substitutes (sweeteners). In this analysis, sucralose and five glycoside sweeteners plus rutin added as I.S. were analyzed using an LC-MS/MS method with the separation on an Acquity UPLC HSS T3 2.1  150 mm column with 1.8 mm particles (from Waters) [74]. In this analysis, the linear calibrations and the quadratic calibrations were not far apart for the analytes. As an illustration, the calibration curves for rebaudioside A are shown in Fig. 4.6.1. Both a linear fit of the experimental data (showing correlation coefficient R2 ¼ 0.9982) and a quadratic fit (showing correlation coefficient R2 ¼ 0.9996) are indicated in the figure.

204

Method Development in Analytical HPLC

Figure 4.6.1 Two possible calibrations, one linear and the other quadratic are shown for the quantitation of rebaudioside A in an analytical method for measuring several sweeteners [75].

In such situation, it is recommended to verify the back-calculated results using each calibration for the low standards, and select the calibration that generated lower differences between the true concentration value (the one taken in the standard) and the back-calculated one. Almost invariable, the quadratic equations provide better results. However, the quadratic dependence should never be used for the calculation of results at higher values than the maximum standard. Regarding the selection of the I.S., the most suitable choice is one that is absent in the sample, is chemically similar to the analyte, and can be selectively separated from the analyte of interest and the matrix. The I.S. must be chosen in such a way to behave in the analytical process as close as possible with the analytes, to not interfere with the analyte determination, and to give a chromatographic peak convenient to integrate. In some instances, the availability of a pure compound X that must be used for creating a calibration curve is not available. When the compound is only available with low purity, the calibration must take into consideration as much as possible the content of X. The true content of a calibrant compound can be affected even by the presence of water. For this reason, very much attention must be paid to the purity of the calibration compound. In cases the calibration compound is not available, an attempt for estimation of the level may be made by using the calibration of a similar compound. However, for many detectors and in particular for MS, the response factor of different compounds can be significantly different. For this reason, only estimation of a compound content can be done in some cases, based on calibration with a different analyte. A different quantitation technique besides that using a calibration curve is the standard addition. Standard addition method can be used to analyze an unknown sample of concentration ½X without the use of a calibration curve. It must be assumed, however, that the relation between the concentration and the peak area follows Eq. 4.5.6. The

Performance and utilization of HPLC

205

linearity between the response and sample concentration and an intercept b ¼ 0 are necessary for successful use of standard addition technique. The unknown concentration ½X that should be analyzed is given by ½X ¼ qX =VX where VX is a known volume of the sample containing the unknown quantity qX . For this sample, a peak area A0 is measured. After the addition of the quantity q1 ¼ ½X1 V1 of the compound to be analyzed, a second peak area A1 is measured. Two equations can now be written: qX =VX ¼ aA0 and ðqX þq1 Þ=ðVX þV1 Þ ¼ aA1 . The ratio of these two formulas leads to the result: ðqX þ q1 ÞVX A1 ¼ ðVX þ V1 ÞqX A0

(4.6.4)

This relation can be easily rearranged to give: ½X ¼

qX q1 A 0 ¼ VX ðVX þ V1 ÞA1  VX A0

(4.6.5)

Other procedures can also be used for quantitation. One such procedure is based on a response factor FX that must be obtained separately. The response factor is generated using a special standard (SS) and it is calculated from the peak area ASS of the special standard and the peak area AX of the compound X, both at equal (or known) concentration present in a blank sample or in pure solvent. The ratio of the two chromatographic peak areas of the standard and analyte, usually obtained as an average of several measurements, gives the response factor: FX ¼

ASS AX

(4.6.6)

Ideally, the value for FX remains constant for an interval of values for the pair of concentrations of the standard and the sample. The concentration of the unknown is then obtained by measuring in the same run the peak area of the compound to be analyzed (at unknown concentration) and peak area of the special standard by using the formula:  ½X ¼ FX

 AX ½SS ASS

(4.6.7)

where AX is the area of the compound X at unknown concentration and ASS is the area of the special standard at the concentration ½SS. In order to achieve a constant value for the response factor FX in a range of concentrations, it is recommended that the two compounds, the special standard and the analyte, be chemically similar or even identical except for use of a labeled compound for the standard. Other calibration procedures are also possible, for example, by using a single concentration of the calibrant and performing the calibration using various injection

206

Method Development in Analytical HPLC

volumes [75]. However, this type of calibration is not frequently utilized. However, as indicated by Eq. 4.5.4, the quantity of the injected analyte is proportional to both the concentration and also to the injection volume of the sample. By injecting a set of samples with increased volume such as between 1 mL and 20 mL, it is possible to obtain a calibration and verify its linearity. The range of such calibration is however much more limited that the calibration obtained by using a fixed injection volume of standards of different concentrations. In this section, the discussion was limited to the concentration determination in the injected solution in the HPLC system, but for practical applications, the concentration measured in the injected solutions must be used to find the concentration (or amount) in the raw sample. From an identified concentration in the injecting solutions into the HPLC system, the corresponding amounts of analyte can be obtained based on Eq. 4.5.4, qinj ¼ ½XVinj . From the value of qinj , the results regarding the content of the analyte in the real sample are obtained considering the amount of raw sample dissolved in the solution injected in the HPLC system. If during the sample preparation steps certain dilutions or concentrations intervene, these must be also considered when calculating the analytes content in the raw sample (see Eq. 6.1.3).

Selection of the internal standards in quantitation using HPLC In chromatographic analysis, the use of internal standards is very common and useful for obtaining more accurate results [76]. For best utilization, the internal standard should behave similarly with the analyte in the sample preparation and in the chromatographic process. More than one internal standard can be selected for a chromatographic method, in case it is necessary to perform analyses for different nonsimilar analytes. The closest behavior to the analyte is typically seen for the same compound as the analyte but isotopically labeled. However, only the use of mass spectrometry as a detection procedure allows the differentiation between isotopically nonlabeled and labeled compounds. Even in this case, the difference in the mass between in compound and its labeled form is recommended to be larger than 2e3 units in order to avoid any interference in the quantitation. Also, the labeled compound must be free of unlabeled compound. The isotopically labeled compounds with deuterium H2, C13, or N15 are frequently used as internal standards [77,78]. Labeling with deuterium is often the most convenient choice. However, when deuterium/hydrogen exchange may occur, the deuterated compounds cannot be used for internal standard [79]. This is, for example, the case of acids, alcohols, or when other exchange reactions may take place such as transesterification [80]. In cases when the detector does not differentiate between the labeled and nonlabeled compounds, or when the labeled compound is not available or it is too expensive, other compounds can be used as internal standards. These compounds are typically selected with properties similar to the analyte (same functional groups, not very different molecular weight, etc.). One additional subject related to the I.S. utilization is its concentration that must be used in an analysis. A good concentration to be selected for the internal standard(s)

Performance and utilization of HPLC

207

should provide Ai =AIS values that are not extreme (very low or very high). Since the values for calibration set fXi gi cover a certain range, an average comparative concentration for the I.S. must be selected such that the extreme values for Ai =AIS are avoided. Several aspects regarding peak shape and separation from other analytes and interferences regarding the selection of an internal standard are further discussed in Chapter 9.

Matrix effects on HPLC quantitation The effect of matrix components coeluting or partly coeluting with analytes may induce modifications in the signal intensity measured by HPLC detectors. The amplitude of this effect (ME) can be calculated from the ratio between the peak areas A measured for the same amount of analyte injected together with the sample matrix analyte (Aanalyte ) of the analyte. Since the with matrix ) and in a pure standard solution (Aas standard pure matrix may also generate some signal Amatrix , the calculation of ME should be performed following the expression:

ME ¼

matrix ðAanalyte with matrix  A

Aanalyte as standard

 100

(4.6.8)

A value of ME ¼ 100% indicates the lack of matrix effect, ME s 100% indicates interference. For example, in MS detection, it can be an ionization suppression, or in fluorescence detection, a quenching of the fluorescence. In such cases, ME < 100. In UV detection, the interference may be caused by the lack of separation between the analyte and a component from the matrix that absorb at the same wavelength as the analyte and in such cases a ME > 100 can be seen. Also, the use of RI detector always requires complete separation of the analyte from the matrix, since this type of detection being nonselective does not provide any possibility of differentiation between the analyte and a matrix component. In cases of selective detection (e.g., a specific mass for MS detection is used, or a specific wavelength for UV detection), it is possible that the HPLC separation does not require to isolate the matrix (or a matrix component) from the analyte, and an accurate measurement is still possible. However, very frequently although the matrix “is not seen,” may still affect the detection.

Utilization of size exclusion HPLC for the evaluation of molecular weight of polymers Size exclusion HPLC indicated as SEC, which can be gel permeation GPC or gel filtration GFC, can be used for the measurement of molecular weight of polymers. Based on Eq. 4.1.22, the retention time in SEC can be written as follows: tR ðXÞ ¼ t0 ½1 þ KSEC ðXÞJ

(4.6.9)

208

Method Development in Analytical HPLC

According to Eq. 2.1.23 for the value of KSEC , Eq. 4.6.9 can be written in the form:



tR ðXÞ ¼ t0 1 þ exp DS0 = R , J

(4.6.10)

and for macromolecules DS0 ¼ S0pore  S0inter < 0 where S0pore is the entropy of the molecule inside the stationary phase pore, and S0inter the entropy outside the pore. As indicated in Section 2.1, the larger is the hydrodynamic volume of the macromolecule the larger is the negative value of DS0 . As a result, the retention time is shorter for macromolecules with the larger dimensions than for those which have smaller dimensions. At the same time, for nonpolymeric molecules, DS0 z 0 and they have the longest retention times in the separation. Because the dimensions of the molecules are related to their molecular weight, it is expected that DS0 will also depend on molecular weight Mw of the molecule. This is verified in practice and an empirical expression can be established between tR ðXÞ and Mw. This dependence can be written in the form: tR ðXÞ ¼ A  B log½MwðXÞ

(4.6.11)

In Eq. 4.6.11, A and B are constants and the equation is utilizable for the evaluation of molecular weight of polymers. The dependence described by Eq. 4.6.11 is valid only for a limited range of Mw values and for nonpolymeric molecules the Mw cannot be measured by this procedure. Also, empirical parameters A and B must be established by calibrations with polymer standards with known Mw and because DS0 depends in fact on hydrodynamic volume of the analytes, the shape of molecules of the polymers used for calibration must be similar to that of analytes. For example, globular polymers generate a different dependence than the polymers with a rodlike shape. Molecular associations which depend on the chemical nature of polymers also influence hydrodynamic volume, and for a more accurate calibration using Eq. 4.6.11, the chemical nature and not only the shape of calibrant must be close to that of the analytes [81].

Key points • • • •

HPLC is used successfully for qualitative and quantitative analysis, as well as for the measurement of molecular weight of polymers. Quantitative analysis by HPLC is utilized for the measurement of concentrations a wide variety of compounds in many types of samples. Quantitative analysis can be performed for some compounds at extremely low levels. The interference on quantitation must be thoroughly evaluated.

4.7

Utilization of HPLC in nontargeted type analysis

HPLC analytical methods are successfully utilized for the targeted types of analysis that focus on the measurement of known analytes in a known sample matrix. However,

Performance and utilization of HPLC

209

a large number of samples are very complex, and the characterization of their chemical composition, identification of unknown compounds, and/or classification of such samples without prior knowledge regarding their chemical content is the task of nontargeted analysis (NTA) also known as nontargeted screening (NTS). NTA includes various analytical protocols also indicated as multiclass, multiplex, mega method, or multicontaminant and multiresidue analyses. Usually, NTS are a combination of sample preparation procedures with HPLC and GC techniques, with the aim of determining, for example, traces of pesticides, veterinary drugs, persistent organic pollutants (POPs), dyes, endocrine disruptors, food contact materials, or polycyclic aromatic hydrocarbons (PAHs) from various natural or biological matrices [82,83].

Application of nontargeted analysis Some typical examples of situations when NTA is carried out for research, industry, and legislative purposes are briefly further discussed. 1) The analysis of contaminants of water resources with pesticides and their degradation products, pharmaceuticals and drug manufacturing by-products, surfactants, and illicit drugs. The task of identification of such man-made compounds released in the environment and of their degradation products is challenging because of both the complexity of such products and because of the complexity of their matrices [84,85]. For analytes not identified from full scan MS acquisition, supplementary experiments using MSn, derivatization methods, H/D exchange combined with information from other analytical methods may be used [86]. 2) The determination of the full composition of natural consumer products such as herbs, foods, coffee, tea, tobacco, nutraceuticals, and others, and their adulteration with unknown substances in the view of verifying their authenticity. Such materials are frequently subject to the measurement of targeted analytes, but nontarget analysis may verify if the quality does not match their description or labeling, and may indicate the illicit source of provenience of the product [87e89]. 3) Analyses for clinical or toxicological studies when the analytical research is focused on the detection and identification of unknown xenobiotics in biological samples. In order to find xenobiotics in acquired untargeted LC-MS datasets, the identification process relies on targeted search in MS spectral libraries created in-house or provided by a third part. The parameters used for detection and structural identification of the compounds include the retention time in specific separations, M=z value, molecular formula, in silico MS/MS evaluation, isotopic pattern, and specific product/fragment ions [90e92]. 4) Evaluation of metabolic changes taking place in biological systems under different conditions when exposed to xenobiotics (exposomic/metabolomics). A critical part of such studies is related to the sample preparation that must enable the extraction of the largest possible variety of molecules from biological matrix. This part of analysis should also take into consideration the stability of metabolites [93,94]. Due to the large amount of analytical data generated from these studies, computational tools are mandatory for faster and in-depth analysis of these data [95].

LC-MS and LC-MS/MS including high-resolution LC-MS/MS are powerful tools for NTA, and the data generated by these methods of analysis are frequently evaluated using chemometric approaches. Examples of chemometrics methods typically applied to evaluate the similarities between experimental HPLC fingerprints

210

Method Development in Analytical HPLC

are pattern recognition methods, principal components analysis (PCA) (see Section 1.7), SIMCA, fuzzy clustering (FC), or multicriteria decision-making (MCDM). In general, samples with similar fingerprints have similar chemical characteristics, and their HPLC profiles are used to establish the identity, authenticity, or batch-to-batch consistency of substances in various fields of application [96e101]. In many cases, beside HPLC, additional analytical techniques are used in NTA, such as CG-MS and NMR. NTA is frequently utilized for providing qualitative information, but quantitative results from NTA can also be obtained. However, quantitative interpretations of NTA data have been largely based on relative quantitation, where measured chemical responses are compared across two or more sample groups [100,101]. The quantitation of one compound based on the quantitation of another compound is questionable because the response factor of different compounds can be different in detection techniques such as UV or MS. For example, even relatively similar compounds, such as the positional isomers 2- and 4-nitrophenol, have a difference of 40 times in ESI-MS response intensity [102,103].

Challenges in NTA analysis NTA encounters a number of challenges. A common such challenge is related to the identification of an unknown compound even using MS detection when this mass spectrum is not available for a reference compound. In GC-MS analysis, large databases of spectra (e.g., NIST 20 and Wiley data bases of mass spectra) are important resources for unknown identification. In LC-MS/MS, this type of identification is much more limited, although various databases are being developed [104]. Identification of the chemical nature from LC-MS data is typically difficult because this method provides only the mass of the molecular ion (and in some cases, the ion after elimination of water or of another fragment). Better information is obtained by using MS/MS detection when a fragmentation pattern can be obtained. However, because the fragmentation in LC-MS/MS analysis depends on experimental MS/MS conditions, including the type of instrument, mass accuracy, resolving power, collision energy, and collision gas pressure, libraries similar to those available for GC-MS (such as NIST 20 or Wiley) are not available. For helping in the interpretation of the spectra obtained in LC-MS/MS analyses, various procedures are utilized such as dedicated programs like Mass Frontier, MassBank data, SmileMS, etc. (see Section 4.6). Some such programs use the “top-down” approach, where the product ions can be predicted in silico MS/MS spectra from an assumed structure [92]. The utilization of highresolution LC-MS/MS data also offer additional help in the unknown compounds identification. Even when a positive compound identification is not possible, highresolution data provide a significant limitation in the number of possible structures for the unknown. Another common procedure used in NTA for the identifications of unknown is the comparison of the profiles of different chromatograms for the evaluated samples with the purpose of studying similarities and differences. One difficult task in such

Performance and utilization of HPLC

211

data processing is the peak alignment and the control of its accuracy, since retention time shifts can be inevitable [105,106]. In a multipeaks chromatogram, the first peaks are considered that keep a constant retention time or show variations that are minor. However, the later eluting sample constituents may show a systematic shift (bias) of their retention time. The comparison of the chromatograms obtained in a sequence is based on peak alignment, which requires the selection of a reference sample, or a reference chromatogram. For automated procedures, this could be chosen by selecting one chromatogram from the middle of the set of chromatograms, or the chromatogram containing the highest number of peaks [107]. One practical procedure for peak alignment is based on correlation optimized warping (COW) algorithm. In this algorithm, the chromatographic peaks are aligned with those in a chromatogram used as a reference, by stretching or compressing some of its segments [108]. Various similarity indices were developed for comparing complex chromatograms (e.g., Ref. [109]). Another challenge in interpretation of complex chromatograms is the lack of complete separation for certain consecutive peaks, even if the separation conditions are modified for this aim. The problem of overlap peaks is theoretically solved by deconvolution procedures, which may generate the initial peaks of the two or more coeluting species from a sample. This deconvolution can be based on the profile of specific ions in case the detection was based on mass spectrometry or on the profile of specific values of wavelength when detection is made using a DAD. Even peaks that are visually unique may be proven to contain multiple compounds [110]. An example of a simulated deconvolution result is given in Fig. 4.7.1. In many cases, the visual inspection of the chromatogram is not sufficient for detecting coeluting compounds. Although the ideal peak shape in a chromatogram should be Gaussian, other effects than peak coelution may influence the peak shape, and differentiation between peak coelution and other effects generating peak shape alteration can be obtained using dedicated capability of computer packages used for data analysis (e.g., MassHunter from Agilent).

Figure 4.7.1 A simulated chromatogram before deconvolution (trace A) and after deconvolution (trace B).

212

Method Development in Analytical HPLC

Key points • •

NTA is an important procedure applied for the evaluation of composition of complex samples. LC-MS/MS is a powerful tool utilized in NTA, along GC-MS and other methods of analysis

References [1] J.S. Baker, J.C. Vinci, A.D. Moore, L.A. Colon, Physical characterization and evaluation of HPLC columns packed with superficially porous particles, J. Separ. Sci. 33 (2010) 2547e2557. [2] J.W. Dolan, How much retention time variation is normal? LCGC North Am. 32 (8) (2014) 546e551. [3] N. Trebel, A. Höltzel, A. Steinhoff, U. Tallarek, Insights from molecular simulations about dead time markers in reversed-phase liquid chromatography, J. Chromatogr. A 1640 (2021) 461958. [4] C.A. Rimmer, C.R. Simmons, J.G. Dorsey, The measurement and meaning of void volumes in reversed-phase liquid chromatography, J. Chromatogr. A 965 (2002) 219e232. [5] R.M. McCormick, B.L. Karger, Distribution phenomena of mobile-phase components and determination of dead volume in reversed-phase liquid chromatography, Anal. Chem. 52 (1980) 2249e2257. [6] F.Z. Oumada, M. Rosés, E. Bosch, Inorganic salts as hold-up time markers in C18 columns, Talanta 53 (2000) 667e677. [7] H.J. Möckel, Influence of temperature on dead volume of ODS columns and on n-alkane retention in high-performance liquid chromatography, J. Chromatogr. A 675 (1994) 13e28. [8] D.V. McCalley, Evaluation of a linear free energy relationship for the determination of the column void volume in hydrophilic interaction chromatography, J. Chromatogr. A 1638 (2021) 461849. [9] L. Redon, X. Subirats, M. Roses, HILIC characterization: estimation of phase volumes and composition for a zwitterionic column, Anal. Chim. Acta 1130 (2020) 39e48. [10] K.B. Sentell, J.G. Dorsey, On the calculation of the stationary phase volume in reversed phase chromatography, J. Liq. Chromatogr. 11 (1988) 1875e1885. [11] D.R. Stoll, G. Kainz, T.A. Dahlseid, T.J. Kempen, T. Brau, B.W.J. Pirok, An approach to high throughput measurement of accurate retention data in liquid chromatography, J. Chromatogr. A 1678 (2022) 463350. [12] S.C. Moldoveanu, V. David, Estimation of the phase ratio in reversed-phase highperformance liquid chromatography, J. Chromatogr. A 1381 (2015) 194e201. [13] E. Caiali, V. David, H.Y. Aboul-Enein, S.C. Moldoveanu, Evaluation of the phase ratio for three C18 high performance liquid chromatographic columns, J. Chromatogr. A 1435 (2016) 85e91. [14] E. Caiali, S.C. Moldoveanu, V. David, Comparison of the phase ratio for C18 HPLC columns using three different organic modifiers (methanol, ethanol, and acetonitrile) in mobile phase composition, Rev. Roum. Chem. 62 (2017) 629e636. [15] S.C. Moldoveanu, E. Caiali, V. David, Phase ratio and equilibrium constant in RP-HPLC obtained from octanol/water partition constant through solvophobic theory, Chromatographia 80 (2017) 1491e1500.

Performance and utilization of HPLC

213

[16] A.C. Soare, V. David, S.C. Moldoveanu, Does phase ratio in reversed phase high performance liquid chromatography vary with temperature? J. Chromatogr. A 1620 (2020) 461023. [17] R.S. Berry, S.A. Rice, J. Ross, Physical Chemistry, Oxford Univ. Press, Oxford, 2000, pp. 526e527. [18] V.I. Smirnov, Kurs Vischei Matematiki, vol. 2, Gosudarstvennoe Izd., Moscow, 1952. [19] B.L. Karger, L.R. Snyder, C. Horvath, An Introduction to Separation Science, J. Wiley, New York, 1973. [20] J.C. Giddings, Dynamics of Chromatography, Part 1. Principles and Theory, Marcel Dekker, New York, 1965. [21] J.J. van Deemter, F.J. Zuiderweg, A. Kinkenberg, Longitudinal diffusion and resistance to mass transfer as causes of nonideality in chromatography, Chem. Eng. Sci. 5 (1956) 271e289. [22] S. Moldoveanu, V. David, Essentials in Modern HPLC Separations, second ed., Elsevier, Amsterdam, 2022. [23] H.G. Barth, Chromatography fundamentals, Part VIII: the meaning and significance of chromatographic resolution, LCGC North Am. 37 (11) (2019) 824e828. [24] Recent developments in LC column technology, Supplement to LCGC North America 38, 2020. [25] U.D. Neue, Theory of peak capacity in gradient elution, J. Chromatogr. A 1079 (2005) 153e161. [26] U.D. Neue, Peak capacity in unidimensional chromatography, J. Chromatogr. A 1184 (2008) 107e130. [27] S.C. Moldoveanu, A. Harrison, Extraction of benzo[a]pyrene from moist snuff with water or artificial saliva (Part 2), Beitr. Tabak. Intern. 29 (2020) 21e26. [28] A. Vailaya, C. Horvath, Retention in reversed-phase chromatography: partition or adsorption? J. Chromatogr. A 829 (1998) 1e27. [29] C.F. Poole, S.N. Atapattu, Analysis of the solvent strength parameter (linear solvent strength model) for isocratic separations in reversed-phase liquid chromatography, J. Chromatogr. A 1675 (2022) 463153. [30] S.C. Moldoveanu, E. Bacalum, T. Galaon, V. David, Revisiting the dependence of retention factor on the content of organic component in the mobile phase in reversedphase HPLC, J. Separ. Sci. 46 (2023) 2300274. [31] E. Bacalum, T. Galaon, A.-C. Soare, V. David, S.C. Moldoveanu, Comparative study for the retention of some benzodiazepines in reversed-phase liquid chromatography using C8, C18, C6H5 and CN stationary phases, Rev. Roum. Chem. 68 (2023) 277e283. [32] U.D. Neue, D.H. Marchand, L.R. Snyder, Peak compression in reversed-phase gradient elution, J. Chromatogr. A 1111 (2006) 32e39. [33] F. Gritti, G. Guiochon, Exact peak compression factor in linear gradient elution, J. Chromatogr. A 1212 (2008) 35e40. [34] F. Gritti, General theory of peak compression in liquid chromatography, J. Chromatogr. A 1433 (2016) 114e122. [35] H. Poppe, J. Paanakker, M. Bronckhorst, Peak width in solvent-programmed chromatography: I. General description of peak broadening in solvent-programmed elution, J. Chromatogr. A 204 (1981) 77e84. [36] M.A. Stadalius, H.S. Gold, L.R. Snyder, Optimization model for the gradient elution separation of peptide mixtures by reversed-phase high-performance liquid chromatography: verification of band width relationships for acetonitrile-water mobile phases, J. Chromatogr. 327 (1985) 27e45.

214

Method Development in Analytical HPLC

[37] W. Hao, K. Wang, B. Yue, Q. Chen, Y. Huang, J. Yua, D. Li, Peak compression in linear gradient elution liquid chromatography, J. Chromatogr. A 1619 (2020) 460908. [38] https://www.chromatographyonline.com/view/smile-ms-small-molecule-identificationsoftware-tandem-mass-spectrometry. [39] T. Kind, J.K. Meissen, D. Yang, F. Nocito, A. Vaniya, Y.-S. Cheng, J.S. Vander Gheynst, O. Fiehn, Qualitative analysis of algal secretions with multiple mass spectrometric platforms, J. Chromatogr. A 1244 (2012) 139e147. [40] S. Chen, Methodology of lipidomics: tandem mass spectrometry of ether phospholipids, in: S. Chen (Ed.), Lipidomics: Sea Food Marine Based Dietary Supplements, Fruit and Seeds, Transworld Res. Network, Kerala, 2012. [41] J. Hollender, et al., NORMAN guidance on suspect and non-target screening in environmental monitoring, Environ. Sci. Eur. 35 (2023) 1e62. [42] Y. Sun, S. Guo, T. Wu1, J. Zhang, L.-Y. Kwok, Z. Sun, H. Zhang, J. Wang, Untargeted mass spectrometry-based metabolomics approach unveils biochemical changes in compound probiotic fermented milk during fermentation, Sci. Food 21 (2023) 1e8. € [43] C. Liteanu, I. Rica, Uber die Definition der Nachweisgrenze, Mikrochim. Acta 61 (1973) 745e757. [44] C. Liteanu, I. Rica, On the frequentometric estimation of the detection limit, Mikrochim. Acta 64 (1975) 311e323. [45] M.D. Nelson, J.W. Dolan, UV detector noise, LC GC 77 (1999) 12e15. [46] J.W. Dolan, Enhancing signal-to-noise, LCGC North Am. 28 (2010) 212e216. [47] U.S. EPA-Title 40: Protection of Environment: Par 136-Gidlines Establishing Test Procedures for the Analysis of Pollutants: Appendix B to Part 136-Definition and Procedures for the Determination of the Method Detection Limit-Revision 1.11. [48] http://www.agilent.com/cs/library/technicaloverviews/public/5990-8341EN.pdf. [49] A.C. Olivieri, Practical guidelines for reporting results in single- and multi-component analytical calibration: a tutorial, Anal. Chim. Acta 868 (2015) 10e22. [50] R.C. Castells, M.A. Castillo, Systematic errors: detection and correction by means of standard calibration, Youden calibration and standard additions method in conjunction with a method response model, Anal. Chim. Acta 423 (2000) 179e185. [51] A. Savitzky, M.J.E. Golay, Smoothing and differentiation of data by simplified least squares procedures, Anal. Chem. 36 (2002) 1627e1639. [52] E. Bosten, P. Van Broeck, D. Cabooter, Automated tuning of denoising algorithms for noise removal in chromatograms, J. Chromatogr. A 1709 (2023) 464360. [53] D.R. Stoll, Effects of flow rate on UV detection in liquid chromatography, LC-GC N. Am. 37 (2019) 846e850. [54] M.F. Wahab, P.K. Dasgupta, A.F. Kadjo, D.W. Armstrong, Sampling frequency, response times and embedded signal filtration in fast, high efficiency liquid chromatography: a tutorial, Anal. Chim. Acta 907 (2016) 31e44. [55] D.S. Bell, New chromatography columns and accessories for 2018, LC-GC N. Am. 36 (2018) 234e247. [56] D.S. Bell, New liquid chromatography columns and accessories, LC-GC N. Am. 37 (2019) 232e243. [57] A. Holm, P. Molander, E. Lundanes, T. Greibrokk, Determination of the pesticide rotenone in river water utilizing large-volume injection packed capillary column switching liquid chromatography with UV and time-of-flight mass spectrometric detection, J. Chromatogr. A 983 (2003) 43e50. [58] A.C. Hogenboom, M.P. Hofman, S.J. Kok, W.M. Niessen, U.T. Brinkman, Determination of pesticides in vegetables using large-volume injection column liquid

Performance and utilization of HPLC

[59] [60]

[61] [62]

[63] [64]

[65]

[66] [67]

[68]

[69]

[70] [71]

[72]

[73] [74]

215

chromatography-electrospray tandem mass spectrometry, J. Chromatogr. A 892 (2000) 379e390. F. Gritti, G. Guiochon, Overload behavior and apparent efficiencies in chromatography, J. Chromatogr. A 1254 (2012) 30e42. J. Dai, P.W. Carr, D.V. McCalley, A new approach to the determination of column overload characteristics in reversed-phase liquid chromatography, J. Chromatogr. A 1216 (2009) 2474e2482. L.R. Snyder, J.J. Kirkland, J.W. Dolan, Introduction to Modern Liquid Chromatography, third ed., Wiley, Hoboken, 2010. E. Krause, S. Rothemund, M. Beyermann, M. Bienert, Monitoring of a-helical secondary structures in peptides by reversed-phase HPLC of replacement sets, Anal. Chim. Acta 352 (1997) 365e374. J. Li, D.B. Hibbert, S. Fuller, G. Vaughn, A comparative study of point-to-point algorithms for matching spectra, Chemometr. Intell. Lab. Syst. 82 (2006) 50e58. S.N. Berger, A.M. Rustum, Comprehensive stress degradation studies of fipronil: identification and characterization of major degradation products of fipronil including their degradation pathways using high resolution-LCMS, Chromatographia 87 (2024) 249e257. K.S. Jewell, U. Kunkel, B. Ehlig, F. Thron, M. Schlusener, C. Dietrich, A. Wick, T.A. Ternes, Comparing mass, retention time and tandem mass spectra as criteria for the automated screening of small molecules in aqueous environmental samples analyzed by liquid chromatography/quadrupole time-of-flight tandem mass spectrometry, Rapid Comm. Mass Spectrom. 34 (2020) e8541. E. Rathahao-Paris, S. Alves, C. Junot, J.C. Tabet, High resolution mass spectrometry for structural identification of metabolites in metabolomics, Metabolomics 12 (2016) 12. D. Petras, I. Koester, R. Da Silva, B.M. Stephens, A.F. Haas, C.E. Nelson, L.W. Kelly, L.I. Aluwihare, P.C. Dorrestein, High-resolution liquid chromatography tandem mass spectrometry enables large scale molecular characterization of dissolved organic matter, Front. Mar. Sci. 4 (2017) 405. D.G. Watson, A rough guide to metabolite identification using high resolution liquid chromatography mass spectrometry in metabolomic profiling in metazoans, Comput. Struct. Biotechnol. J. 4 (2013) e201301005. D. Verdegem, D. Lambrechts, P. Carmeliet, B. Ghesquiere, Improved metabolite identification with MIDAS and MAGMa through MS/MS spectral dataset-driven parameter optimization, Metabolomics 12 (2016) 98. A. Bolje, S. Gobec, Analytical techniques for structural characterization of proteins in solid pharmaceutical forms: an overview, Pharmaceutics 13 (2021) 534. S. Thomas, C.S. Mathela, A. Agarwal, S.K. Paul, Identification and structural elucidation of an unknown impurity in carbamazepine active pharmaceutical ingredient by liquid chromatographyetandem mass spectrometry and semi-preparative chromatographic isolation, J. Pharm. Biomed. Anal. 56 (2011) 423e428. L. Cuadros-Rodríguez, M.G. Bagur-Gonzalez, M. Sanchez-Vi~ nas, A. Gonzalez-Casado, A.M. Gomez-Saez, Principles of analytical calibration/quantification for the separation sciences, J. Chromatogr. A 1158 (2007) 33e46. M. Forina, S. Lanteri, M. Casale, Multivariate calibration, J. Chromatogr. A 1158 (2007) 61e93. S.C. Moldoveanu, A liquid chromatography-tandem mass spectrometry method for the analysis of sucralose and five glycoside sweeteners, Sep. Sci. Plus 6 (2023) 2300068.

216

Method Development in Analytical HPLC

[75] L.D. Asnin, Peak measurement and calibration in chromatographic analysis, Trends Anal. Chem. 81 (2016) 51e62. [76] M. Wang, C. Wang, X. Han, Selection of internal standards for accurate quantification of complex lipid species in biological extracts by electrospray ionization mass spectrometry e what, how and why? Mass Spectrom. Rev. 36 (2017) 693e714. [77] E. Stokvis, H. Rosing, J.H. Beijnen, Stable isotopically labeled internal standards in quantitative bioanalysis using liquid chromatography/mass spectrometry: necessity or not? Rapid Commun. Mass Spectrom. 19 (2005) 401e407. [78] A. De Nicolo, M. Cantu, A. D’Avolio, Matrix effect management in liquid chromatography mass spectrometry: the internal standard normalized matrix effect, Bioanalysis 9 (2017) 1093e1105. [79] L.B. Nilsson, G. Eklund, Direct quantification in bioanalytical LC-MS/MS using internal calibration via analyte/stable isotope ratio, J. Pharm. Biomed. Anal. 43 (2007) 1094e1099. [80] S.C. Moldoveanu, R. Yerabolu, Critical evaluation of several techniques for the analysis of phthalates and terephthalates: application to liquids used in electronic cigarettes, J. Chromatogr. A 1540 (2018) 77e86. [81] A. Striegel, W.W. Yau, J.J. Kirkland, D.D. Bly, Modern Size-Exclusion Liquid Chromatography: Practice of Gel Permeation and Gel Filtration Chromatography, second ed., Wiley, Hoboken, 2009. [82] D. Steiner, A. Malachova, M. Sulyok, R. Krska, Challenges and future directions in LCMS-based multiclass method development for the quantification of food contaminants, Anal. Bioanal. Chem. 413 (2021) 25e34. [83] M.V. Cesio, H. Heinzen, Multiple names for multi-scope analytical methods, more than a semantic issue, J. Chromatogr. A 1717 (2024) 464687. [84] F. Gosetti, E. Mazzucco, M.C. Gennaro, E. Marengo, Contaminants in water: non-target UHPLC/MS analysis, Environ. Chem. Lett. 14 (2016) 51e65. [85] A.A. Bletsou, J. Jeon, J. Hollender, E. Archontaki, N.S. Thomaidis, Targeted and nontargeted liquid chromatography-mass spectrometric workflows for identification of transformation products of emerging pollutants in the aquatic environment, Trends Anal. Chem. 66 (2015) 32e44. [86] A. M€uller, W. Schulz, W.K.L. Ruck, W.H. Weber, A new approach to data evaluation in the non-target screening of organic trace substances in water analysis, Chemosphere 85 (2011) 1211e1219. [87] J. Klikarova, L. Ceslova, Targeted and non-targeted HPLC analysis of coffee-based products as effective tools for evaluating the coffee authenticity, Molecules 27 (2022) 7419. [88] T. Wang, L. Duedahl-Olesen, H.L. Frandsen, Targeted and non-targeted unexpected food contaminants analysis by LC/HRMS: feasibility study on rice, Food Chem. 338 (2021) 127957. [89] G. Campmajo, G.J. Navarro, N. Nunez, L. Puignou, J. Saurina, O. Nunez, Non-targeted HPLC-UV fingerprinting as chemical descriptors for the classification and authentication of nuts by multivariate chemometric methods, Sensors 19 (2019) 1388. [90] C. Chen, A. Wohlfarth, H. Xu, D. Su, X. Wang, H. Jiang, Y. Feng, M. Zhu, Untargeted screening of unknown xenobiotics and potential toxins in plasma of poisoned patients using high-resolution mass spectrometry: generation of xenobiotic fingerprint using background subtraction, Anal. Chim. Acta 944 (2016) 37e43. [91] C. Zhu, G. Lai, Y. Jin, D. Xu, J. Chen, X. Jiang, S. Wang, G. Liu, N. Xu, R. Shen, L. Wang, M. Zhu, C. Wu, Suspect screening and untargeted analysis of veterinary drugs

Performance and utilization of HPLC

[92]

[93]

[94] [95]

[96]

[97]

[98]

[99]

[100]

[101] [102] [103] [104] [105]

[106]

[107]

217

in food by LC-HRMS: application of background exclusion-dependent acquisition for retrospective analysis of unknown xenobiotics, J. Pharm. Biomed. Anal. 210 (2022) 114583. C.M. Fisher, T.R. Croley, A.M. Knolhoff, Data processing strategies for non-targeted analysis of foods using liquid chromatography/high-resolution mass spectrometry, Trends Anal. Chem. 136 (2021) 116188. J. Pezzatti, J. Boccard, S. Codesido, Y. Gagnebin, A. Joshi, D. Picard, V. Gonzalez-Ruiz, S. Rudaz, Implementation of liquid chromatographyehigh resolution mass spectrometry methods for untargeted metabolomic analyses of biological samples: a tutorial, Anal. Chim. Acta 1105 (2020) 28e44. B.L. Milman, I.K. Zhurkovic, Present-day practice of non-target chemical analysis, J. Anal. Chem. 77 (2022) 537e549. L. Goracci, P. Tiberi, S. Di Bona, S. Bonciarelli, G. Ilaria Passeri, M. Piroddi, S. Moretti, C. Volpi, I. Zamora, G. Crucian, MARS: a multipurpose software for untargeted LC-MS-based metabolomics and exposomics, Anal. Chem. 96 (2024) 1468e1477. Y. Ni, Y. Lai, S. Brandes, S. Kokot, Multi-wavelength HPLC fingerprints from complex substances: an exploratory chemometrics study of the Cassia seed example, Anal. Chim. Acta 647 (2009) 149e158. B. Xie, T. Gong, M. Tang, D. Mi, X. Zhang, J. Liu, Z. Zhang, An approach based on HPLC-fingerprint and chemometrics to quality consistency evaluation of Liuwei Dihuang pills produced by different manufacturers, J. Pharm. Biomed. Anal. 48 (2008) 1261e1266. C. Wang, H. He, J.-l. Zhang, X. Li, Z.-l. Ma, High performance liquid chromatography (HPLC) fingerprints and primary structure identification of corn peptides by HPLC-diode array detection and HPLC-electrospray ionization tandem mass spectrometry, J. Food Drug Anal. 24 (2016) 95e104. A. Hawryl, M. Hawryl, A. Hajnos-Stolarz, J. Abramek, A. Bogucka-Kocka, L. Komsta, HPLC fingerprint analysis with the antioxidant and cytotoxic activities of selected lichens combined with the chemometric calculations, Molecules 25 (2020) 4301. J.M. McCord, L.C. Groff II, J.R. Sobus, Quantitative non-targeted analysis: bridging the gap between contaminant discovery and risk characterization, Environ. Int. 158 (2022) 107011. A. Kruve, Making nontargeted liquid chromatographyehigh-resolution mass spectrometry analysis quantitative, LC-GC Eur. 32 (2019) 320e322. M. Oss, A. Kruve, K. Herodes, I. Leito, Electrospray ionization efficiency scale of organic compounds, Anal. Chem. 82 (2010) 2865e2872. A. Kruve, K. Kaupmees, J. Liigand, I. Leito, Negative electrospray ionization via deprotonation: predicting the ionization efficiency, Anal. Chem. 86 (2014) 4822e4830. https://sciencesolutions.wiley.com/solutions/technique/lc-ms/. Y. Wang, L. Ma, M. Zhang, M. Chen, P. Li, C. He, C. Yan, J.-B. Wan, A simple method for peak alignment using relative retention time related to an inherent peak in liquid chromatography-mass spectrometry-based metabolomics, J. Chromatogr. Sci. 57 (2019) 9e16. P.F.M. Sousa, A. de Waard, K.M. Aberg, Elucidation of chromatographic peak shifts in complex samples using a chemometric approach, Anal. Bioanal. Chem. 410 (2018) 5229e5235. T. Skov, F. van den Berg, G. Tomasi, R. Bro, Automated alignment of chromatographic data, J. Chemometr. 20 (2006) 484e497.

218

Method Development in Analytical HPLC

[108] N.P.V. Nielsen, J.M. Carstensen, J. Smedsgaard, Aligning of single and multiple wavelength chromatographic profiles for chemometric data analysis using correlation optimised warping, J. Chromatogr. A 805 (1998) 17e35. [109] E. Tyteca, M. Talebi, R. Amos, S.H. Park, M. Taraji, Y. Wen, R. Szucs, C.A. Pohl, J.W. Dolan, P.R. Haddad, Toward a chromatographic similarity index to establish localized quantitative structure-retention models for retention prediction: use of retention factor ratio, J. Chromatogr. A 1486 (2017) 50e58. [110] T.J. Nelson, Deconvolution method overlapping peak areas for accurate determination in chromatograms, J. Chromatogr. 587 (1991) 129e136.

Samples in high-performance liquid chromatography analysis 5.1

5

Compounds characteristics relevant to HPLC separation

The versatility of high-performance liquid chromatography (HPLC) allows its use for the analysis of a very broad range of molecular types. The main type of application of HPLC is in quantitative analysis, but the technique is also used for qualitative analysis, for the measurement of molecular weight Mw of polymers, as well as for many other applications, some nonanalytical (e.g., preparative). In a method development, the understanding of sample properties is essential for making the best selections of the type of HPLC to be utilized, for the selection of an appropriate HPLC column, of an adequate mobile phase, and of a good method of detection. For this reason, it is important in the development of an HPLC method to understand the relation between specific sample properties and options for the selections for the HPLC features. Most frequently, the HPLC methods are developed for quantitation purposes, and the nature and properties of the samples to be analyzed are not always known. In such cases, prior to attempting to develop a quantitative HPLC method for analysis, it is necessary to perform some exploratory studies for obtaining as much information as possible about the sample characteristics. For samples that are volatile, gas chromatographyemass spectrometry (GC-MS) is a common analytical technique to detect the unknown. To make certain samples amenable for GC-MS analysis, in some cases, derivatization can be used to extend the volatility and thermal stability of sample compounds [1]. Exploratory HPLC with various columns and detector types may be used for providing information regarding the sample constituents, or even separations followed by compound identification using techniques such as infrared (IR) or nuclear magnetic resonance (NMR) (see Chapter 1). In many cases, the sample qualitative composition is basically known and the submission for the analysis is done for specific compound or group of compounds. In such cases, an evaluation of sample characteristics can be performed without direct experimental work, by using the available literature information. The pertinent information about the sample characteristics refer usually to various physicochemical properties of sample constituents and of the estimated proportion of analytes and matrix in the sample. Physicochemical properties related to separation can be either molecular properties, such as ionic charges, dipole moment, polarizability, or they can be “bulk” properties such as sample chemical nature, octanol/water partition coefficient, solubility in various solvents, etc. The importance of several such characteristics for guiding choices in HPLC is further evaluated. In some instances, the HPLC analysis is performed for qualitative and quantitative purposes. In such cases, an important role for the qualitative part is played by the detector. HPLC with mass spectrometric detection (LC-MS), and more frequently with Method Development in Analytical HPLC. https://doi.org/10.1016/B978-0-443-29849-3.00011-7 Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

220

Method Development in Analytical HPLC

tandem mass spectrometric detection (LC-MS/MS), are important techniques used for obtaining qualitative information in addition to quantitative one. Several characteristics of samples that are important in such applications are also discussed in this chapter.

Organic or inorganic character The sample to be injected for analysis with the HPLC system may contain organic compounds, inorganic compounds, or a mixture of them. Compounds containing carbon-hydrogen (CeH) and/or carbon-carbon (CeC) bonds are indicated as organic compounds. A few compounds such as CCl4, CBr4, HCN, ClCO2H, and CS2 can be classified as either organic or inorganic. Organic compounds can also include in their molecule other atoms such as oxygen (O), nitrogen (N), phosphorus (P), arsenic (As), sulfur (S), selenium (Se), as well as halogens, boron (B), etc. A special class is that of organometallic compounds that contain metals bonded to a carbon in their molecule. The selection of a specific type of HPLC for the analysis of organic or inorganic compounds depends on several factors, and there is not a definite choice regarding which HPLC methods must be used. However, in some cases, the selection is more straightforward. For example, small molecules of ionic compounds are mostly analyzed using ion exchange chromatography (see Section 2.3). In this category can be included the analysis of inorganic anions such as Fe, Cle, Bre, Ie, CNe, SCNe, OCNe, CO2 3 , etc., which are analyzed using anion exchange chromatography, or cations such as Liþ, Naþ, Kþ, Ca2þ, Sr2þ, etc., which are analyzed using cation exchange chromatography. Another example is that of small nonpolar organic molecules that are typically analyzed using reversed-phase HPLC (RP-HPLC). The selection of HPLC type as well as other details regarding the HPLC are influenced by a number of other factors, but the organic or inorganic nature of the analytes is one of the factors to be considered in this selection.

Molecular mass A common property that differentiates the molecules is the number of atoms they contain. Some molecules contain a small number of atoms, while others have a large or very large number. When a molecule contains at least 1000 atoms linked by covalent bonds, it is indicated as a macromolecule. However, instead of using the number of atoms as a base for classifying a molecule as small or a macromolecule, a more common classification criterium is the molecular mass. Molecular mass M of a molecule is its mass expressed in unified atomic mass units (1/12 of the mass of one atom of the isotope carbon-12, unit sometimes named dalton, Da). Molecular weight (Mw) of a molecule is the ratio of the mass of the molecule to 1/12 of the mass of isotope carbon-12, and Mw is dimensionless [2]. Except for the different dimensional units, Mw and molecular mass M are numerically equal and the terms are frequently used interchangeably. Molecular weight differentiates analytes as small molecules when Mw is lower than about 2000 Da, and as a macromolecule when it is higher than about 5000 Da. Between these two limits, it is a gray area with molecules indicated as oligomers. The molecular mass M (and Mw) is calculated from the standard atomic

Samples in high-performance liquid chromatography analysis

221

weights of each element which takes into account the isotopic distribution of the elements. However, in mass spectrometric detection, the result for the molecular mass reflects the exact mass as the sum of individual isotopes of the atoms in the molecule. As a result, the mass spectrum for a molecule contains separate exact masses for each isotopic composition, but with intensities corresponding to the isotopic natural distribution. For example, deuterium is present on Earth at a level of 0.0156% of the number of hydrogen atoms and the EIþ spectrum of H2O will show the ion M=z ¼ 18 (99.74% intensity), but also ions M=z ¼ 19 (corresponding to HDO) and M=z ¼ 20 (corresponding to D2O) at very low levels. The difference between the mass and the exact mass of the molecule made from the most abundant isotopes starts to be important when using, for example, multiple reaction monitoring (MRM) or selected ion monitoring (SIM) in mass spectrometric detection for larger molecules. For example, mogroside V, a glycoside of cucurbitane derivative that is present in some plants and is used as a sugar substitute has the Mw ¼ 1287.43, but the exact Mw ¼ 1286.65. This difference can be significant, for example, when setting MRM acquisition parameters for an HPLC analytical method with MS detection when larger nonpolymeric molecules are analyzed [3]. Another such example is the molecule of Bacoside 3A with the exact Mw ¼ 928.50 and reported Mw ¼ 929.11 (see Fig. 9.3.1). For macromolecules, the Mw can be a precise number even for some very large ones such as titin protein (e.g., isoform Q8WZ42-1 having Mw ¼ 3,816,030.05 Da). However, for some synthetic polymers, even for homopolymers, the molecules of a sample may have different degrees of polymerization within a specific range. In such cases, the molecular weight is taken as an average (weight-average molecular weight). Considering the weight fractions wi =W where W is the total weight of a polymer sample and wi the weight of a specific fraction having molecular Pweight Mwi , the average molecular weight is obtained using the formula Mw ¼ ð wi Mwi Þ=W. Similar to the case of being organic or inorganic, the molecular mass (weight) is another criterion for the selection of HPLC type to use for analysis. For small molecules, other criteria than Mw are more important for the selection of HPLC, but for polymers, a common HPLC type utilized for separation is size exclusion (SEC), either gel filtration or gel permeation, where the value of Mw is important in column selection. In some cases, other HPLC types than SEC are applied for the separation of polymers, but even in such cases, the differentiation between small molecules and polymers must be considered. For example, in the case of protein separation using RP-HPLC, a large pore size of the stationary phase must be selected (e.g., 300 Å), while for small molecules, the pore size of the stationary phase is selected in the range 80e120 Å.

Acidic, basic, or amphoteric character Acidic, basic, or amphoteric character (Brönsted-Lowry) was defined in Section 2.4. The dissociation of those compounds is usually characterized using their acidity constant pKa ¼ log Ka . Tables with acidity constants are readily available in the literature for common acids and bases. Similar to any equilibrium constant, Ka is temperature dependent and its value is typically listed for 25 C. Various techniques

222

Method Development in Analytical HPLC

are available for pKa calculation [4]. Computer programs are also available for the calculation of pKa (e.g., MarvinSketch [5]). For macromolecules with multiple groups ionizable in solution, the individual pKa values for each functionality are not relevant parameters, the basic or acidic character of the molecule being a global property. The acidity or basicity constant for a compound is a very important parameter in the decision of the type of chromatography that should be used for its separation. Also, the pH of the mobile phase is selected in many separations as dependent of the pKa of the analytes (mobile phase usually at 2 pH units difference from the pKa ), with the purpose of keeping the analyte in mainly one form or another. In many HPLC separations, buffers are utilized for maintaining a specific pH of the mobile phase.

Some geometric molecular properties An important geometric property of a compound is its molar volume Vmol which is the volume occupied by 1 mole of a compound (or element) at a given temperature and pressure. Molar volume is defined by the formula (r is the density of the compound): Vmol ¼ Mw=r

(5.1.1)

Other geometric molecular properties include van der Waals molecular volume 3 V vdW which is the volume occupied by a molecule (in  A ) and van der Waals molec2 ular surface area A vdW (in  A ). Those parameters are based on the concept that each atoms have a defined van der Waals radius, and the atoms are placed in a molecule at covalent bond distances (shorter than the sum of atomic radiuses of the two connected atoms). Various publications describe the calculation of van der Waals molecular volume V vdW (basically the same as McGowan volume [6]) and area A vdW [7e9], and such values can also be calculated using computer packages such as MarvinSketch [5]. Van der Waals molecular surface area A vdW plays an important role in solvophobic interactions of a molecule as indicated in Section 2.2. An example of van der Waals molecular surface areas A vdW is indicated below in Fig. 5.1.1 for bromazepam showing a “balls and sticks” and a “spacefill” structure.

Figure 5.1.1 Van der Waals molecular surface area A in “balls and sticks” and “spacefill” images.

vdW

¼ 335.33  A2 for bromazepam shown

Samples in high-performance liquid chromatography analysis

223

The shape of molecules is characterized by numerous parameters such as area of projection on different plans, different parameters regarding accessibility to polar groups (polar surface area), solvent accessibility, etc., as well as a number of other topological parameters such as molecular radius of gyration (for polymers), Balaban index, Harry index, eccentricity, etc. [10,11]. The molecular geometry plays a role in various types of separation including RP-HPLC [12]. For polymers, the molecular shape is in particular very important. Polymers are frequently separated using size exclusion chromatography (SEC) and as indicated in Section 2.1, in this technique, the separation is based on hydrodynamic volume of the molecules. In solution, molecules are associated with solvent molecules and hydrodynamic volume includes the volume of the molecule and that of the solvent molecules associated with it.

Isoelectric point The isoelectric point (pI) is the pH value at which the molecule having ionic properties carries no electrical charge. The molecules are positively charged below the pI and are negatively charged above the pI. The concept is particularly important for zwitterionic molecules such as amino acids, peptides, and proteins. For an amino acid, the isoelectric point is the average of pKa values for the amine and the carboxyl group. In the case of amino acids with multiple groups ionizable in solution (e.g., lysine with two amino groups or aspartic acid with two acid groups), the isoelectric point is given by the average of the two pKa of the acid and base that lose/gain a proton from the neutral form of the amino acid. This can be extended to the definition of pI of peptides and proteins. The pI value can be used to indicate the global basic or acidic character of a zwitterionic molecule, and compounds with pI > 7 can be considered basic, and those with pI < 7 can be considered acidic.

Polarity, dipole moment, and polarizability The strict definition of polarity refers to the separation of the center of partial positive charges from that of partial negative charges in the molecule (see Section 2.2). Various approaches are reported in the literature for the calculation of partial charges [13e16] and several computer programs are available for their calculation (e.g., MarvinSketch [5] and other programs [17e19]). However, more general than the evaluation of point charges, the study of electronic population for a molecule is important for the understanding of molecular properties such as dipole moment m (see Eq. 2.2.4) and polarizability a (see Eq. 2.2.12). Values for the dipole moment m for various molecules are reported in the literature (e.g., Ref. [20]). The values for m are usually indicated in specific solvents (e.g., benzene, dioxane, etc.) where the compounds are in molecular form. Computer programs are also available for dipole moment evaluation. Polarizability defined by Eq. 2.2.12, can also be calculated, for example, from the molar refraction (molar refractivity) Rmol based on the following expression: a¼

3Rmol 4pN

(5.1.2)

224

Method Development in Analytical HPLC

Molar refractivity (molar refraction) Rmol is a measure of total polarizability of a mole of substance and it is defined by the formula: Rmol ¼

 2  n 1 Vmol n2 þ 2

(5.1.3)

In Eq. 5.1.3, n is the refractive index of the material, and Vmol is the molar volume (Rmol is also indicated by symbol A). Polarity of the molecule is an important parameter since it determines the energy of intermolecular interactions (as shown in Section 2.2) and also determines the molecular reactivity and hydrogen bond formation. Unfortunately, the calculations of charge distribution in a molecule as well as that of m and a are not always simple, and the correlation with the compound properties is not always straightforward. For this reason, instead of individual molecular property such as electronic population, point charges, dipole moment, and polarizability, a certain bulk property of compounds can be used for the characterization of polarity. This property is the octanol/water partition coefficient Kow .

Octanol/water partition coefficient The use of log Kow as a parameter for evaluating polarity is very common. Octanol/water partition coefficient Kow (sometimes indicated by P in the literature) was previously defined with Eq. 2.2.1 and the definition of this parameter is also given below: Kow ðXÞ ¼

½Xoctanol ½Xwater

(5.1.4)

In Eq. 5.1.4, the molar concentrations ½X in octanol and in water refer to an extraction system in equilibrium where the octanol and water are in contact to each other and the octanol is saturated with water and the water with octanol. Octanol/water partition coefficient describes the opposite of polar character, namely the hydrophobic character, with higher values for Kow indicating hydrophobic character and with lower values for more polar character. It is common for this parameter to be expressed in logarithm form (log Kow ). There is no direct formula giving Kow based on charge distribution in the molecule although log Kow depends on electronic distribution in the molecule. In addition, octanol/water coefficient is a “bulk” property referring to the whole material, while charge distribution, dipole moment, and polarizability are properties at molecular level. The values for log Kow do not correlate with the dipole moment m or polarizability a, but do correlate with van der Waals molecular surface area when specific values for the polar groups are excluded [21]. This indicates that log Kow is a parameter closely related to the effects taking place in a solution where solvophobic effects are important (see Section 2.2). A major component determining the free energy in a dissolution process as explained by solvophobic effects as being the free energy required for the creation of the cavity in the solvent to accommodate the

Samples in high-performance liquid chromatography analysis

225

molecule of the solute plus the energy of the interaction of the solute with the surrounding. The resulting energy depends on the surface area of the molecule. The values for log Kow can be obtained experimentally (e.g., Ref. [22]), and can be calculated using various procedures [23,24]. Most of these procedures are based on the additive fragment methodology (e.g., Ref. [25]). In this procedure, the value of log Kow is obtained based on an expression of the form: log Kow ¼

X

an f n þ

X

bn Fn

(5.1.5)

In Eq. 5.1.5, an is the number of occurrences of a fragment and fn is a constant for the particular fragment, bn is the number of occurrences when a correction factor is needed, and Fn is a correction factor for a structural feature in the molecule. Tables with values for fn and Fn are available in the literature (e.g., Ref. [26]). In such tables, it can be noticed positive values for groups such as C, CH, CH2, CH3, C6H5, etc., and negative values for polar groups such as O, OH, N, NH, NH2, NH3, COO, etc. The resulting log Kow values are increased for molecules with nonpolar moieties and decreased for those containing polar moieties. Also, the larger molecules but having less polar groups have larger log Kow . A few examples of molecules listed in the increasing order of their log Kow values are shown in Fig. 5.1.2. For Kow , extended information is available in the literature. Specific values for Kow are available for many compounds [22,27] and Kow values can also be calculated using computer programs such as MarvinSketch 5.4.0.1 from ChemAxon Ltd. [5], EPI Suite [28], ClogP [29], ACD/logPdb, KowWin [30], and SciLogP/Ultra. Other methods for the estimation of Kow are based on physicochemical molecular properties such as van der Waals molecular surface area [21], solvatochromic parameters [31], etc. The results generated by different methods of log Kow typically generate similar results, and in good agreement with the experimental data. Many compounds can be present in the form of more than one molecular species. These compounds are characterized by a related parameter to Kow , which is the octanol/water distribution coefficient Dow . The distribution coefficient Dow ðXÞ for a given compound X (existent in more than one form X1 ; X 2 ; X 3 ; .Xn ) is defined by a formula similar to Eq. 2.1.13 (o indicates octanol and w water): Dow ðXÞ ¼

½X1 o þ ½X2 o þ ½X3 o þ . þ ½Xn o ½X1 w þ ½X2 w þ ½X3 w þ . þ ½Xn w

(5.1.6)

The case of more forms for the same compound is very common for compounds that can form neutral and ionized forms such as acids, bases, and amphoteric compounds. As an example, the molecule of sildenafil which has an amphoteric character, at very low pH will accept protons and will be present in a small proportion as X 2þ and mainly as X þ , as the pH increases the neutral form of the molecule will be predominant in a solution, and at high pH, the molecule will be mainly in the form X  . Each of these forms is present at a certain proportion in a solution, depending on pH (as discussed in Section 2.4), and depending on pH, the distribution coefficient Dow also will change.

226

Method Development in Analytical HPLC

Figure 5.1.2 A few molecules shown in the order of their increased log Kow values.

This is illustrated in Fig. 5.1.3 where the change of Dow is shown as function of the pH for sildenafil, together with the proportion of different sildenafil forms depending on pH. Octanol/water partition coefficient is a very useful parameter utilized for guiding the selection of an HPLC type and even of a chromatographic column and mobile phase. A good correlation has been reported between log Kow (or log Dow ) and retention factor of many compounds in various HPLC separations [32e34]. An example of 0 the correlation of log k with log Kow for a set of 72 mono and disubstituted aromatic 0 compounds with log k values obtained for a C18 stationary phase (Lichrosorb RP-18)

Samples in high-performance liquid chromatography analysis

Form 3

+

227

Form 4

Form 5

+

Form 2

-

100

1.35 0.85

+

Form 1

+

% Composition

80 0.35

70

log Dow

60 50

log Dow

90

+

-0.15

Form 1 Form 2

-0.65

40 30

-1.15

Form 3

Form 4 Form 5

20 -1.65

10

-2.15

0 0

2

4

6

8

10

12

14

pH Figure 5.1.3 Variation of Dow for the molecule of sildenafil and proportion of different molecular forms as a function of pH.

using water/methanol 50/50 (v/v) as a mobile phase is shown in Fig. 5.1.4, and using water/methanol 40/60 (v/v) as a mobile phase is shown in Fig. 5.1.5 [32]. From Figs. 5.1.4 and 5.1.5, it can be seen that between the retention factor k 0 in RPHPLC and octanol/water partition coefficient log Kow , the following expression can be established: 0

log k ðXÞ ¼ a log Kow ðXÞ þ b

(5.1.7)

In Eq. 5.1.7, parameters a and b depend on the mobile phase and chromatographic column on which takes place the separation. 0 Based on the good correlation between log k and log Kow that is frequently noticed, octanol/water partition coefficient is a useful parameter regarding the choice of the type of HPLC in the development of a quantitation method. In Fig. 5.1.6, a schematic diagram related to the recommended choice of an HPLC type as a function of the log Kow (or log Dow ) of a compound is indicated. The diagram from Fig. 5.1.6 provides only a general guidance regarding the recommended HPLC, and other choices are possible. In addition, many specific details

228

Method Development in Analytical HPLC

2

1.5

y = 0.5199x - 0.3311 R² = 0.9355

log k'

1 0.5 0 -1

-0.5

0

0.5

1

1.5

2

2.5

3

3.5

4

-0.5 -1

log Kow 0

Figure 5.1.4 Correlation between log k and log Kow for 72 compounds on a Lichrosorb RP-18 columns with water/methanol 50/50 (v/v) as a mobile phase.

1.5

1

0.5

log k' -1

y = 0.4765x - 0.5218 R² = 0.9075

0 -0.5

0

0.5

1

1.5

2

2.5

3

3.5

4

-0.5

-1

log Kow 0

Figure 5.1.5 Correlation between log k and log Kow for 72 compounds on a Lichrosorb RP-18 columns with water/methanol 40/60 (v/v) as a mobile phase.

Ion exchange, IPC

log Kow = -5

Reversed- phase (RP)

HILIC, NPC, IPC

-4

-3

-2

-1

0

1

2

3

NARP

4

5

>5

Figure 5.1.6 Types of HPLC recommended as a function of log Kow of the analyte.

Samples in high-performance liquid chromatography analysis

229

regarding each type of HPLC cannot be generated only from the log Kow value. However, the value of this parameter provides useful guidance in the development of an HPLC method of analysis, and compared to many other parameters, the values for log Kow are more readily available. Besides its usefulness in providing guiding information regarding separation, octanol/water partition coefficient is also related to the solubility of an organic compound X and water solubility can be estimated, for example, based on expression of the form: 0

log½X ¼ a log Kow ðXÞ þ b 0

0

(5.1.8)

0

In Eq. 5.1.8, a and b are parameters depending on functional groups in the organic molecules (such as alcohol, ketone, ester, etc.) and such parameters are listed in literature (e.g., Ref. [35]).

Capability of formation of hydrogen bonds The role of hydrogen bond formation in molecular interactions was shortly discussed in Section 2.2, where it was indicated that hydrogen bonds are partly electrostatic and partly covalent, involving an electron donor/electron acceptor interaction. The tendency of forming hydrogen bonds is related to the polarity of a molecule and also on its general chemical structure. Hydrogen bond formation is typical for water, organic acids, alcohols, and to a lower extent to amines, and other compounds having H atoms in a polar group such as thiols. The energies of hydrogen bonds are in the range of 10e40 kJ/mol, as indicated in Table 2.2.1.

Diffusion coefficient One other parameter with importance in HPLC is the diffusion coefficient DX;solv of a compound X in a solvent solv. As indicated in Section 4.1, the plate height H ðHETPÞ for a chromatographic column is not the same for different compounds and H depends on DX;solv (see Eq. 4.1.49). An empirical formula for the value of DX;solv is indicated below [36]: DX;solv ¼ 7:4,108

ðjsolv Mwsolv Þ1=2 T 0:6 h Vmol;X

(5.1.9)

In Eq. 5.1.9 where Vmol;X is the molar volume of solute X (in cm3 mole1), Mwsolv the molecular weight of solvent solv, T is temperature in Kelvin degrees, h the viscosity of the solution (in 10e4 g cm1 s1, or centipoise), and jsolv is an “association” factor for the solvent (jsolv is 1 for nonpolar solvents, 1.5 for ethanol, 1.9 for methanol, 2.6 for water). For mixed solvents, solv1 and solv2 with the corresponding molar fractions xsolv1 for solv1 and ð1  xsolv1 Þ for solv2 the term jsolv Mwsolv is replaced by ½xsolv1 jsolv1Mwsolv1 þð1  xsolv1 Þjsolv2 Mwsolv2 ].

230

Method Development in Analytical HPLC

Other physical characteristics to consider for compounds in HPLC separations Knowledge about some common physical characteristics of the samples is sometimes useful for the HPLC analysis. Among such characteristics are the boiling point B.p., 0 melting point M.p., density r, the superficial tension g , dielectric constant ε, etc. Boiling point (B.p.) of a liquid is the temperature at which the liquid vapor pressure equals the atmospheric pressure. Boiling temperature at different pressures is usually indicated as Tboil , specifying the pressure. The melting point (M.p.) is the temperature at which it changes state from solid to liquid. This property also depends on pressure, but less than B.p. although M.p. is still indicated for 1 atm. Other properties of compounds further listed may be of interest related to an HPLC separation. Among these are: (1) density r which is the mass per unit volume of a compound, (2) surface tension 0 g which is the tendency of a liquid surfaces to shrink into the minimum surface area due to the cohesive forces existent between its molecules (in practice, the importance of this parameter is also related with the formation and stability of emulsions); super0 ficial tension g is measured in force per unit length the SI unit being newton/m (N/m), but other units are also common (e.g., millinewton/m), (3) dielectric constant ε which is the ratio of the permittivity of a material to the permittivity of vacuum ε0 .

General chemical structure of the organic compound and related properties When the chemical structure of the organic compound is known, specific sample characteristics as those previously discussed as having importance in the selection of an HPLC method can be inferred. For example, a large group of organic nonpolymeric molecules are the hydrocarbons that can be saturated hydrocarbons (linear or branched), saturated cyclic hydrocarbons, unsaturated hydrocarbons with one or more double bonds, unsaturated hydrocarbons with triple bonds, aromatic hydrocarbons, or a combination of all these structures. Compounds from this group are typically not water soluble and have large log Kow values. For the separation of such compounds, nonaqueous reversed-phase chromatography (NARP), or normal phase chromatography (NPC) are usually recommended. On the hydrocarbon backbone, various functional groups can be attached. Some such compounds can have only simple groups such as halogen, alcohols, enols, phenols, ethers, peroxy groups, thiols, sulfides, amines, imines, a variety of other nitrogenous compounds (nitro, oximes, etc.), aldehydes, ketones, carboxylic acids, various derivatives of organic acids (such as esters, lactones, acyl chlorides, etc.). Some of those groups such as halogens, alcohol, aldehyde, ketone, esters, etc., do not bring an acidic or basic character to the molecule, but increase molecular polarity. The most common type of HPLC in such cases is reversed phase (RP-HPLC). However, if the polar groups are numerous and the log Kow value for the molecule becomes very low, HILIC or IPC could be the preferred selection. When groups such as carboxyl or amino are present in the molecule, RPHPLC is still a possible choice, but other techniques such as HILIC or even ion

Samples in high-performance liquid chromatography analysis

231

exchange can be selected. In such cases, specific attention must be given to the pH of the mobile phase since the form (ionized or not) of basic or acidic molecules are strongly influenced by pH. In addition to information regarding the core analytical procedure, the information about structure will provide information about solubility. It is common for molecules containing groups such as OH, COOH, NH2, etc., to be water soluble, while those containing long hydrocarbon chains or many aromatic rings to not be soluble in water. One additional advantage of knowing the chemical structure for the analytes is related to the existence of information in the literature about method of analysis commonly used for different molecule types. For example, carbohydrates where frequently analyzed in practice and methods for their analysis are reported in the literature (e.g., Refs. [37,38]). Also, specific chromatographic columns were developed for carbohydrate analysis by different manufacturers (e.g., Dionex CarboPac type columns from Thermo Scientific [39]). Another group of compounds commonly analyzed is that of amino acids where again a large body of information regarding their analysis is available (e.g., Ref. [40]). Similar to the case of carbohydrates, specific chromatographic columns were developed for amino acid analysis (e.g., Intrada Amino Acid from Imtakt [41]). Besides the case of carbohydrates and amino acids where a significant amount of information regarding their analysis is available, this is also true for lipids, steroids, etc., or in the case of polymers for proteins and nucleic acids. Extensive literature containing dedicated information about their analysis also exists for chemical compounds as classified based on the role of those compounds in everyday life. Examples of such groups are samples of biological origin (blood, urine, biological tissues), agricultural products, pharmaceuticals, nutraceuticals, environmental pollutants, pesticides, food and beverages (vegetables, grains, meat, wine, soft drinks), flavoring agents, illicit drugs, cannabis components, tobacco, surfactants, dyes, etc. The compounds present in such groups can be very different and may require different core analytical procedure. However, having similar matrix, the sample preparation for the analytes from those groups is frequently very similar. The variety of nonpolymeric organic compounds is extremely wide, and many classes of molecules could be subject of HPLC analysis. One example is the group of heterocycles which can be classified based on the heteroatoms in the cycle such as oxygen (furans, pyrans), nitrogen (pyrroles, pyrazoles, imidazoles, triazoles, pyridines, pyrazines, etc.), sulfur (thiophenes), or different heteroatoms (oxazoles, thiazoles, oxadiazoles, etc.). These types of molecules are common, for example, as pharmaceutical products and requests for their analysis are frequent. Based on known or estimated log Kow values, the appropriate choice of HPLC can still be done. An additional aspect related to having information regarding the chemical structure is related to the understanding of potential presence in the sample of isomers. These can be structural isomers (chain branching, position of functional groups, functional group type), or steric isomers (cis-trans, syn-anti, chiral, diastereo). In HPLC separations, some isomers can be easily separated, but in particular, the separation of chiral isomers requires special HPLC separations. Chirality is a common property of many compounds present in biological samples, and specific HPLC methods must be developed for differentiating enantiomers. Information about chemical structure of sample

232

Method Development in Analytical HPLC

combined with the information on purpose of analysis are essential for developing chiral separations. Related to the structure of the molecule are also its volume and shape previously described. Molecule structure provides understanding regarding both the volume of the molecule and the capability of association with the molecules of the solvent. Such information can be very useful in selecting an adequate HPLC method of analysis.

Key points • • • •

Properties of the analyte molecule and of matrix important for the separation can be classified as molecular properties (e.g., ionic charges, dipole moment, polarizability), or they can be “bulk” properties (acid-base character, octanol/water partition coefficient). A useful property for estimating a compound behavior in a separation is octanol/water partition coefficient (Kow ). For compounds with acidic, basic, or amphoteric character, octanol/water distribution coefficient (Dow ) which depends on the pH of the medium must be used instead of Kow . The information about the chemical structure of the compound is very useful in selecting an HPLC method of analysis.

5.2

Characteristics of compounds relevant for detection in HPLC

The selection of a detection type and even detection parameters are determined mainly by the properties of the analytes, but also by those of the residual matrix that remained after sample preparation step, and by components of the mobile phase selected for separation. As a result, it is important for developing an analytical method to understand which properties of the analyte are more adequate for detection and how the properties of the residual matrix and of the solvents used in the mobile phase may affect this choice. Some comments regarding those properties are presented in this section.

UV spectra of the analyte The ultraviolet (UV) spectra are generated by the transitions of electrons between different energy levels of a molecule. Electronic levels are classified as bonding, nonbonding, and antibonding depending on their energy (e.g., Ref. [6]). Each electronic level in a molecule is associated with vibrational and rotational levels. When all the electrons of a molecule are in bonding states, the molecule is indicated as being in ground state, and when the molecule has electrons in an antibonding state, the molecule is considered in an excited state. Light absorption is produced when a beam of light produces an electronic transition of the molecule from its ground electronic state into an excited state and this is followed by the dissipation of the energy of the excited state by nonradiative processes such as collisions with other molecules. The change in energy follows Plank’s law DE ¼ hn where DE is the difference in energy between the

Samples in high-performance liquid chromatography analysis

0.7

233

290 nm

0.6

Absorbance

0.5 0.4 0.3 0.2 0.1 250

275

325 350 300 Wavelength nm

375

400

Figure 5.2.1 Range from 250 to 400 nm for the UV spectrum of a-tocopherol.

two energy levels of the molecule. The frequency n (or the corresponding wavelength l ¼ c=n) in UV absorption does not appear as a unique value because the transitions are involving a number of vibrational levels. This explains why the electronic spectra appear as broad bands. An example of a range from 250 to 400 nm of the UV spectrum for a-tocopherol is given in Fig. 5.2.1. The energy levels in molecules are commonly classified based on the symmetry of the molecule (e.g., Ref. [42]). For symmetry reasons, the molecular orbitals can be classified as s; p; d, . for bonding (ground) energy levels, n for nonbonding, and s ; p ; d , . for antibonding (excited) energy levels. The energies exchange can involve s/ s or n/ s transitions that usually generate quite large DE values and correspond to short wavelengths l. These absorption bands commonly fall in the far UV region (below 220 nm). Most molecules have absorption in this range and, as a result, the selectivity of UV detection in this range is not very good. Also, in the range below 220 nm, the molecules of several solvents and of certain compounds that are used as additives may produce adsorption, and this limits their use in the HPLC mobile phase when UV is used as detection. Further discussion regarding the UV absorption of various solvents used for mobile phase in HPLC is given in Section 8.2. The energy involved in n/ p and p/ p transitions also correspond to UV absorptions but at higher wavelength values or may even fall in the visible region (above 360 nm). The n/ p and p/ p transitions are characteristic for organic molecules that contain p bonds such as C]C, C]O, C]S, N]N, C^N, NO2, etc. The presence in a molecule of such groups, indicated as chromophores, is associated with the capability of the molecule to produce a UV absorption that can be utilized for its detection, usually based on the absorption at the maximum value lmax of the absorption band. The chromophore groups are very common in many organic molecules and UV absorption is very frequently utilized for detection in HPLC. Because the absorbance Al for a certain electronic transition depends linearly on the analyte concentration ½X, the UV absorption is perfectly adequate for quantitative analysis (see Eq. 3.3.3, Al ¼ εl ½XL). The molar absorption coefficient εl in Eq. 3.3.3 depends on the whole compound structure, and not only on the existing chromophores. Other

234

Method Development in Analytical HPLC

groups from the molecule such as OH, NH2, and Cl may influence the UV absorption (e.g., Ref. [43]). Such groups are indicated as auxochromes or as hyperchromes when they increase the value of εl . In addition to auxochromes, the conjugation of specific double bonds may influence the UV spectrum both regarding the value of maximum of absorption lmax and the value of εl . Some molecules have a change of structure depending on the pH, and in such cases, the UV spectrum depends on the pH of the mobile phase. The profile of band absorption of a molecule, having specific maxima, can be used for the recognition of a specific molecular species as exemplified by the band with a maximum absorption at 290 nm for a-tocopherol shown in Fig. 5.2.1. The proof of “purity” of a chromatographic peak (i.e., that the peak is generated by a singly compound) is sometimes obtained by comparing the peak spectrum with that of the corresponding standard compound. This cannot be achieved using a single wavelength type detector, but can be performed when a DAD is used as the LC detector. However, the identification of unknown compounds based solely on their UV spectrum is not usually possible as indicated in Section 4.6.

Fluorescence of the analyte In fluorescence by absorbing the radiation with a specific wavelength lex , a fluorescent molecule goes from a lower electronic energetic state (usually ground state) to an excited state, loses part of the energy by a nonradiative process going to a lower excited state and then by emitting a photon having a wavelength lem reaches again the ground state. In this process, a radiation with a lower frequency is generated. Fluorescence can also occur when the initial excited state loses some energy by changing only to a significantly lower vibrational level and from there to a ground state. Fluorescence by emission of radiation at higher frequency than the absorbed one is also possible (anti-Stokes radiation), but it is uncommon. Similar to the case of the UV absorption, because the electronic levels are associated with vibrational levels, the excitation light and the emission light covers a range of wavelengths, and lex and lem are the maxima of those ranges. As an example, the excitation and emission bands for fluorescein are shown in Fig. 5.2.2. 0 As indicated by Eq. 3.3.6, the intensity of fluorescence F int;lem depends linearly on a fluorescent analyte concentration ½X and this property can be used for quantitative analysis in HPLC. In some instances, fluorescence of a compound can also be used for qualitative purpose. The average lifetime of an excited state of a molecule undisturbed by collisions is about 10e8 s, and fluorescence can take place within this length of time. Fluorescence induced by a laser (LIF) can be used as a very sensitive detection technique with application in HPLC as well as in capillary electrophoresis [44]. For some special compounds, the molecules can remain for a longer time in a metastable excited state. In this case, fluorescence can be observed long after the initial radiation is interrupted. This type of fluorescence is commonly called phosphorescence. Fluorescence is less observed than expected because nonradiative losses of energy frequently take place.

Samples in high-performance liquid chromatography analysis

235

494 nm

Excitation band

Intensity

512 nm

Emission band

350

400

500 450 Wavelength nm

550

600

Figure 5.2.2 Excitation band (maximum lex ¼ 494 nm) and emission band (maximum lem ¼ 512 nm) for fluorescein.

Similar to the case of UV absorption where the presence of chromophores generates UV absorption capability, specific structures in a molecule are associated with fluorescence. Those structures are indicated as fluorophores. Several chemical families are recognized as possessing fluorescent capabilities. Among those are derivatives of xanthene, cyanine, squaraine, coumarin, oxadiazole, oxazine, acridine, arylmethine, tetrapyrrole, dipyrromethene as well as certain anthraquinones, dansyl, prodan, and pyrene derivatives, etc. Several factors related to the mobile phase influence fluorescence and the mobile phase pH, solvents nature, presence of impurities such as dissolved O2 may contribute to the decrease (quench) of fluorescence. Quenching of fluorescence detection due to the presence of O2 in mobile phase can be avoided by helium sparging and by in-line vacuum degassing based on a gas-permeable membrane that extracts O2 and N2 from the mobile phase before reaching to the analytical column.

Chemiluminescence of molecules As indicated in Section 3.3, chemiluminescence (CL) is analogous with fluorescence in which the excited state of the molecules is generated following a chemical reaction. There are two general mechanisms of inducing the CL. One mechanism is based on the direct reaction between the reagent R and analyte X, resulting in a reaction product P in an excited state, which returns to a ground state by emitting light (hn): R þ X / P / P þ h n

(5.2.1)

A classic example of this type of CL direct mechanism is the reaction between luminol (o-aminophthalyl hydrazide) and hydrogen peroxide, in the presence of a catalyst. Another example is the reaction between Ru(2,2 -bipyridine)3þ 3 (generated electrochemically from Ru(2,2 -bipyridine)2þ ) and compounds like aliphatic amines, amino 3 acids, and oxalates through the high-energy electron transfer reactions, resulting the reduced species Ru(2,2 -bipyridine)2þ 3 *, which emits light that can be measured.

236

Method Development in Analytical HPLC

The other type is based on inducing CL by the transfer of energy from the product P resulted from a reaction between two reagents R1 and R2 to the analyte species X, by the mechanism described as follows: R1 þ R2 /P

(5.2.2)

P þ X / P þ X  / P þ X þ hn

(5.2.3)

The emitted light has a frequency corresponding to the fluorescence of compound X. An example of reactions 5.2.2e5.2.3 is between an aryl oxalate ester and H2O2 leading to the formation of an energy-rich intermediate capable of exciting a large number of fluorophores X through an electron exchange mechanism. The intermediate P forms a charge transfer complex with a fluorophore X, donating one electron to the intermediate, which is then transferred to the fluorophore raising it to an excited state and subsequently producing light emission (characteristic to X) [45]. CL detection can be applied in HPLC using the direct reaction between a reagent R and analyte X, but more frequently based on inducing CL by the transfer of energy to the analyte species X from the product P resulted from a reaction between two reagents, R1 and R2 (reaction 5.2.3). The reactions known to generate CL are not common and can be applied only for certain analytes. However, when it is possible to be used, CL methods can detect as low as a few hundred amol/mL of compound [46,47].

Characteristics of molecules relevant in LC-MS detection In MS detection, the generation of ions is the first step of importance for this type of detection. As indicated in Section 3.4, in electrospray ionization (ESI), the nature of the analyte but also the water and some additives present in the mobile phase are important for the efficiency of ion formation. The choice between positive or negative ionization mode depends on the nature of the compounds to be analyzed. Molecules with a basic character (e.g., amines, heterocycles containing nitrogen, etc.) have a tendency to form positive ions and are typically analyzed in positive mode. Other molecules such as acids and some oxygenated or halogenated compounds are typically analyzed in negative mode. The main characteristics for the ionization efficiency, as described in Section 4.4, are gas-phase basicity DG0GPB for the formation of positive ions (see Eq. 3.4.2), and gas-phase acidity DG0GPA for the formation of negative ions (see Eq. 3.4.4). These values are, however, not frequently available, and are not commonly used in analytical practice. Besides the chemical nature of the compounds and parameters related to the ionization conditions, the ionization of analytes in MS is also affected by the sample matrix, in case that the matrix is not well separated from the analyte by the chromatographic process. For an analyte X (or XH) and an ionized matrix component MatrixHþ (or Matrix ), a charge transfer process can take place and is described by reactions of the type:

Samples in high-performance liquid chromatography analysis

X þ MatrixHþ $ XHþ þ Matrix

237

(5.2.4)

or in case of negative ionization: XH þ Matrix $ X  þ MatrixH

(5.2.5)

These types of reactions being reversible may favor the formation of the ions of the matrix and not of the analyte (being displaced to the left), and this can contribute significantly to the decrease in ion concentration of the analyte in the LC-MS source. For this reason, even if the matrix compound has a different M=z from the analyte, the response of the detector can be influenced when the chromatographic separation is not very good. In case of large concentration of a matrix constituent, if the apex of the chromatographic peak of the analyte is different but not far from that of the matrix constituent, they elute close to each other and even the tail of matrix constituent may interfere in sensitivity. Besides the role in the formation of the ions, the structure of the analyte molecule is a key factor in the formation of fragments when MS/MS detection is utilized. The fragmentation in MS/MS also depends considerably on the conditions set for the collisioninduced dissociation (CID) cell. The fragmentation pattern of molecules in LC-MS/ MS is in some respects similar to the fragmentation generated by electron impact (e.g., EIþ), but peak intensity and the number and even nature of fragments can be very different. The unimolecular reactions typical for GC-MS with EI þ type ionization are not the same as the one in LC-MS/MS where ion-molecule reactions and other collision reactions take place. The electron impact at 70 EV used in GC-MS has the result of formation of odd electron ions (OEþ) or radical ions of the form X ⦁þ with an open shell electronic structure. This type of ions is not common in LC-MS ionization where mainly ions XHþ having a closed shell electronic structure are formed (EEþ ions), these ions being more stable. In the collision cell of the MS/MS detector, unimolecular reaction also take place for the ions with the type ABHþ which suffer decompositions of the type: ABHþ /Aþ þ BH

(5.2.6)

ABHþ /A þ BHþ

(5.2.7)

ABHþ /AHþ þ B

(5.2.8)

The decomposition of EEþ type ions formed in LC-MS generates usually a fragment positive EEþ type ion and a neutral molecule, while the radical ions of the form ABH⦁þ generated in GC-MS by electron impact typically follow the reaction of the type ABH⦁þ /Aþ þ BH ⦁ generating a positive ion and a neutral radical. The formation of the positive ion creates a similarity with the GC-MS fragmentation, but since the ions ABH⦁þ usually have more internal energy than the ions ABHþ , fragmentation in GC-MS is usually more advanced. Also, similar to the fragmentation in

238

Method Development in Analytical HPLC

GC-MS, the most abundant fragment ion corresponds to the reaction that forms the most stable product ion and neutral fragment in LC-MS or radical in GC-MS. This is the case, for example, when a nonbonding orbital of a heteroatom is present in the molecular ion, and when resonance stabilization occurs (e.g., Ref. [16]). This explains, for example, why many nitrogenous compounds generate intense ions in positive mode ionization. The reactions indicated by Eqs. 5.2.6e5.2.8 can be followed by further unimolecular reactions of the fragments with formation of smaller fragments. In addition to the formation in LC-MS of positive ions XHþ having a closed shell electronic structure, sometimes ions of the form X ⦁þ with an open shell electronic structure can also be formed. However, this type of ions is less stable and although may participate to the formation of fragments, are not commonly found in the mass spectrum in LC-MS. The complexity of possibilities of fragmentation in LC-MS/MS makes however difficult the prediction of fragmentation and even more difficult the reconstruction of a parent molecule from the fragments [48,49]. Various software tools are available for helping both fragmentation prediction, and parent identification such as NIST MS Search, MS-DIAL, Mass Frontier, SmileMS, Massþþ, XCMS2, etc. Also, the development of dedicated libraries allows unknown identification for specific classes of compounds [50]. The use of high-resolution MS/MS instruments that generate ions with a precise mass helps in limiting the alternatives in parent compound identification [51].

The role of compound structure in other detection types As indicated in Chapter 3, in addition to UV, fluorescence, or MS detectors, which are the most commonly utilized, a number of detector types can be used in HPLC. Refractive index, electrochemical, and evaporative light-scattering detectors are also relatively common. Specific analytes properties are related to the detectability using these detectors. For example, from Eqs. 5.1.2 and 5.1.3, it can be seen that refractive index n of a molecular species is related to polarizability a and molar volume Vmol of the molecule. However, it is not common that n is calculated and although lists of refractive indices may be available, and molar refractivity can be estimated (e.g., using MarvinSketch [5]), these values are not usually necessary when using the refractive index for detection. For quantitation based on the detector response, the calibration with the appropriate standard compensates for differences in the n values. However, the differences in the n value from molecule to molecule indicates that calibration for one compound cannot be utilized for another compound quantitation, unless it is verified that the n values of the two compounds are not practically different. Regarding electrochemical detectors, the conductivity detectors are typically used in ion chromatography and the nature of the ions to be detected influences the resulting conductivity. For example, in the analysis of various anions, molar conductivity of a specific ion as well as dissociation constant of the acid HX generated by the ion suppressor (see Section 3.3) may lead to different responses of the detector. Similar to the case of refractive index, the differences in conductivity of the analytes require proper

Samples in high-performance liquid chromatography analysis

239

calibration for performing quantitation. In case of using an amperometric detector, a specific oxidation potential (higher than E1=2 ) must be utilized for achieving oxidation. This potential depends on the nature of the analyte and specific electroactive groups must be present in molecule for successful utilization of this technique. Tables of standard electrochemical potentials in specific media are available in the literature [52] and the half-reactions are expressed as reductions. Molecules such as neurotransmitters, catecholamines, aminoglycosides, carbohydrates, thiols, and phenols can be measured using an amperometric detector (e.g., Ref. [53]). For evaporative light-scattering detectors as well as for charged aerosol detectors (CADs), the molecules of the analyte should be nonvolatile. The response factor for those detectors is relatively uniform for many analytes [54]. The CAD detection is very useful, in particular, for molecules that do not have chromophores, are not fluorescent, and have low ionization yield in MS [55]. Purity and volatility of the mobile phase are also important factors in CAD response. In some cases, when specific compounds such as polyfluorinated long-chain carboxylic acids have to be detected with CAD, it is recommended to use CH3COONH4 as buffer in mobile phase in order to allow the formation of nonvolatile ion-pairs with ions generated from analyte molecules [56].

Key points • •

The detection of analytes can be performed based on various physicochemical properties that are different from those of the mobile phase. The selection of a detection type and even detection parameters are determined mainly by the properties of the analytes, but also buy those of the residual matrix and by components of the mobile phase.

5.3

Complexity of the injected sample

The development of an HPLC analytical method typically starts with the development of the method using pure compounds and not including a matrix. This step makes sure that the method can analyze the target analytes, and establishes a specific range of concentrations for the calibration. After the method is verified that works for the specific set of analytes and it is capable to provide reliable results, it is also useful to test if some potential interferences from the matrix require or not to be eliminated by sample preparation. Sample preparation may consist of a simple sample dissolution, but in many cases, the raw sample is subject to cleanup, analyte concentration, or even derivatization that changes the chemical nature of analytes and/or the matrix. It is common that not all matrix is eliminated in the sample preparation step (see Chapter 6). After sample preparation step, the sample may contain mainly the analytes in a solution, but it is also possible to contain a significant level of matrix (heavy matrix) made from components that do not need to be measured. After sample preparation, the components of the sample to be injected in the HPLC are: (1) the sample solvent (diluent), (2) the analytes,

240

Method Development in Analytical HPLC

and (3) the remaining matrix present in the injected sample even after sample preparation. Sample solvent is selected by the analyst and specific recommendations are important for this selection. A full discussion regarding the selection of an appropriate sample solvent and its influence on HPLC separation is presented in Section 8.4. The effect on the analysis of the content of analytes and remaining matrix is summarized in the present section. The characteristics of analyte and matrix are referred to the ones in the processed sample, which is utilized as solution to be injected in the HPLC and not to the ones in the raw sample. The samples injected in the HPLC can be rather complex and possibly with a low content of analytes. Also, in some cases, the number of targeted analytes is high, and sample preparation cannot be used to reduce this number. In such cases, the HPLC separation may become more complicated, and the degradation of HPLC columns may be more rapid if the correct precautions are not taken (e.g., using guard columns). Also, the analysis of samples with complex and/or heavy matrix may be subject to more errors. For this reason, the understanding of the complexity of the injected sample is important in developing an HPLC method.

Sample complexity regarding the chemical nature of its constituents The overall qualitative sample composition is usually known from preliminary information and preliminary experiments with the sample. Because the nature of analytes and of the remaining matrix of the sample have significant importance in the separation and the measurement of the analytes, the chemical nature of its components should be known before staring a quantitative analysis. Complexity of the sample regarding chemical nature of its constituents (both analytes and matrix) can be loosely classified as follows: (1) samples containing only organic small molecules (nonpolymeric), (2) sample containing both small and polymeric organic molecules, (3) sample containing both organic and inorganic small molecules, (4) complex samples containing small and polymeric molecules, both organic and inorganic. 1) In case of samples containing only small organic molecules, the focus should be placed toward evaluating the polarity of sample components. This can be done, for example, based on octanol/water partition coefficient (log Kow ). For compounds with log Kow values in a limited range, the separation is usually performed on a unique type of chromatographic column (e.g., RP type or HILIC type, etc.). Separation of close eluting peaks is typically achieved by using gradient which may improve separation. The separations on a single column can be more difficult when the log Kow of the analytes to be separated are very close. In some cases, the significant change in gradient may solve the problem. Such gradients may even use a range of solvents with very different elution strength. In some cases, the difference in log Kow values is only between the analytes and the matrix. If the matrix has very low log Kow values, it can be made to elute at the beginning of the separation (eluted with the dead volume of the column) with virtually no retention. However, when the matrix has very large log Kow values, the matrix component may accumulate in the column if proper measures are not taken to eliminate them at the end of the chromatographic run.

Samples in high-performance liquid chromatography analysis

241

2) In case that small and polymeric organic molecules are present in the sample, it depends if the analytes are of both types, or only the matrix contains polymers (a common case when biological samples must be analyzed for metabolites or exobiotic molecules). In such cases, the separation can be adjusted to have the matrix eluted at the beginning of the separation with virtually no retention (for hydrophilic matrix). In case of a strongly hydrophobic matrix, the care should be to eliminate it at the end of the chromatographic run. In case the analytes are polymeric, their separation is usually achieved using SEC. In SEC, the small molecules are strongly retained and their elimination from the column must be verified. In SEC, the use of gradients is not common and the mobile phase is important mainly to assure the analytes solubility. Using long enough retention times, the small molecules are usually eliminated toward the end of the run. 3) For samples containing both organic and inorganic small molecules as analytes, it is frequently necessary to use different chromatographic separations for each group since simultaneous analysis is not usually possible. In case of inorganic compounds in the matrix, they rarely affect the separation of the organic compounds and common columns such as RPtype do not retain these compounds. In case the inorganic compounds are strong acids or strong bases, they may affect the stability of the chromatographic column, and it is recommended that such compounds to be eliminated during sample preparation. 4) Complex samples regarding the chemical nature of their constituents are usually more difficult to analyze using a single chromatographic run. The effort in method development is usually to use a procedure as simple as possible, but in some instances, a single separation is not sufficient for capturing the separation of all analytes. This also depends on which components are analytes and which are matrix. If the matrix can be grouped and eluted either with the dead volume or at the end of the separation, the analytical method can be achieved using a single separation. Other separations can be more difficult.

Sample complexity regarding the number of compounds Regarding the number of compounds, the samples can be loosely classified as follows: (1) sample with one or few analytes and simple (or nonexisting) matrix, (2) sample with one or few analytes but with a rather complex matrix or a large content of matrix (heavy matrix) that should be separated from the analytes, (3) sample with a larger number of analytes and simple matrix, (4) sample with a larger number of analytes and also with a complex matrix. 1) For samples with few analytes and simple matrix, the separation can be usually performed on shorter and narrower columns. The use of gradients is common in HPLC and with this type of samples. The initial development of the methods, as obtained on pure compounds can be focused toward having a short run time for increasing efficiency, and can be applied without modifications to the real samples. 2) For samples with a small number of analytes but with a complex matrix or a matrix in high concentration, the focus in initial method development is to design a separation such that the detection and measurement of the analytes to not be affected by the matrix. For this purpose, the main matrix components must be evaluated before starting to apply the method on real samples. In particular for the quantitation of analytes at low level, a matrix at a significantly higher level can strongly influence the analysis. An example is the measurement of tobaccospecific nitrosamines (TSNAs) in a nicotine sample. A chromatogram generated by LC-MS/ MS for the standards of deuterated TSNAs at the level of 1 ng/mL in absence of nicotine is

242

Method Development in Analytical HPLC

Figure 5.3.1 Chromatogram of a standard mixture of deuterated TSNAs at 1 ng/mL level (trace A) and the chromatogram of the same standard mixture in a sample with 5% nicotine (trace B). shown in Fig. 5.3.1A. When nicotine is present at the level of 5%, although the nicotine peak is chromatographically separated, the extremely modified resulting profile for the same deuterated standards is shown in Fig. 5.3.1B. The separation of analytes in the chromatograms shown in Fig. 5.3.1 was performed on a Kinetex 1.7 mm EVO C18 column (100  2.1 mm) with elution using gradient with solution A aqueous 10 mM CH3COONH4, and solution B 0.1% CH3COOH in acetonitrile [57]. In this separation, although the nicotine peak has a different retention time than the TSNAs, it is still close to the analyte peaks. Also, the Kinetex 1.7 mm EVO C18 column is a coreshell type column with a relatively low loading capacity such that the tail of nicotine peak is still extending over the TSNAs peaks and interferes with the detection. The use of a larger column, Kinetex 2.6 mm EVO C18 column (150  4.6 mm) resolved the interference problem. In some instances, when the matrix of the sample is in large quantity and has strong eluting capability, it may have the same influence on the separation of analytes as a “strong” sample solvent (see Section 8.4). The solvent of the sample, when it has weak eluting properties and is injected at volumes in the range 10e20 mL, may leave all the analytes at the head of the chromatographic column and passes rapidly through the column eluting close to t0 . This

Samples in high-performance liquid chromatography analysis

243

produces a “focusing” of the analytes at the head of the column and can improve the peak shape (see Section 8.4). A sample solvent with strong eluting properties (and at injection volumes larger than e.g., 5.0 mL) may produce the reverse effect. The analytes can be fast eluted from the column head for a short period of time until the mobile phase starts diluting the sample solvent. This short period can be sufficient for distorting the peaks of the analytes. The same effect can be seen with a matrix component in large proportion present in the sample, when the matrix component may act as a solvent until it is diluted with the mobile phase. In such situations, the dilution of sample with the mobile phase (at initial composition) usually resolves the problem. 3) As sample complexity increases, larger columns must be utilized, and longer run times for the separation. Also, complex gradients can be utilized for enhancing the separation. The purpose in such cases is to develop a separation with higher resolution Rs. As indicated in Section 4.2, an increase in the resolution Rs of an analytical column can be achieved by increasing theoretical plate number N, retention factor k', and separation factor a (see Eqs. 4.2.8 and 4.2.9). Column construction can help increasing N by using longer columns with smaller particles that have spherical shape and homogeneous size distribution, as well as possibly using columns with core-shell particles. The decrease in particle size and increased length of the column are however limited by the increase in the column backpressure as indicated by Eq. 4.1.57. The retention factors k' depend on the nature of the compounds to be separated, selection of the stationary and mobile phase (through K), and also on phase ratio J. Phase ratio can be increased by using columns containing more stationary phase (e.g., of wider diameter or a larger carbon load) as indicated by Eq. 4.1.23. However, an increase in k' leads to longer retention times in the chromatogram which may be undesirable because longer run times lead to a decrease in productivity. The increase in a can be achieved by using an appropriate column and mobile phase, but finding such column and mobile phase may require various trial and error attempts. In summary, for more complex sample analysis, the method development of an analytical method requires more effort. 4) Samples with complex set of analytes and complex matrix are the most difficult to analyze. Different strategies can be applied in such cases. First option is to modify the sample preparation and simplify the matrix. Regarding the choice of column, mobile phase, and separation conditions, the recommendation is to make effort for achieving a higher Rs as applied to sample with many analytes but simple matrix, as indicated previously. In some cases, it is possible that the chromatographic separation cannot be performed successfully on all sample components, and selective detectors can be used (such as MS or MS/MS) that allow separate measurements even for compounds not well separated chromatographically. This can be achieved, for example, in MS/MS based on differences in the transitions of product ion (parent) /fragment (daughter) for each analyte. An example of an analysis without complete separation of analytes but still successful regarding the measurement is a separation of 20 amino acids of a standard mixture containing 100 nmol/mL in 0.1 N HCl of each analyte. The separation was obtained on an Intrada Amino Acid column 50  3 mm and 3 mm particles. The mobile phase for the separation was component A: CH3CN/THF/25 mM in water HCOONH4/HCOOH ¼ 9/75/16/0.3 (v/v/v/v) and component B: CH3CN/100 mM HCOONH4 ¼ 20/80 (v/v). The gradient for the separation was 0% B (0e3 min), 1e17%B (3e6.5 min), and 100% B (6.5e10 min) with a flow of 0.6 mL/min and a column temperature of 40 C [58]. The detection was performed using MS/MS in ESI type ionization in positive mode, using MRM with retention times and transitions indicated in Fig. 5.3.2.

1

2

4

6

7

8

9.194, Arg, 175.1Æ 70.1

8471, Lys, 147.0Æ 84.1

8.125, His, 156.1Æ 110.1

7.125, Cys-Cys, 241.0Æ 152.0

5.312, Gln, 147.1Æ 84.1

5 Time min

5.487, Gly, 75.8Æ 30.1 5.665, Asn, 133.1Æ 74.1

4.715, Ala, 90.1Æ 44.0

5.125, Ser, 106.1Æ 60.2

3.812, Pro, 116.1Æ 70.1

4.125, Asp, 134.1Æ 73.9 4.185, Thr, 120.1Æ 74.0

3.050, Val, 118.1Æ 72.1

3

3.500, Glu, 148.1Æ 84.1

Response 0

2.213, Leu, 132.1Æ 86.1 2.325, Met, 150.1Æ 56.1 2.475, Ile, 132.1Æ 86.1

Method Development in Analytical HPLC

Æ 188.1 1.625, Trp, 205.1Æ 1.775, Phe, 166.1Æ 120.1 1.913, Tyr, 182.1Æ 136.2

244

9

10

Figure 5.3.2 LC-MS/MS chromatogram of a mixture of 20 amino acids separated on an Intrada amino Acid column. The differentiation of analytes using a combination separation þ selective detection can be used for the analysis of very complex samples, mainly when the matrix does not produce peak distortion or interferences. A very similar separation as the one for 20 amino acids shown in Fig. 5.3.2 can be extended to the analysis of 55 amino acids, but also with the use of differentiating capability of the MS/MS detection. In another example, up to 500 pesticides at 10 ng/mL in an olive oil matrix matched mix of standards (and sample preparation) can be analyzed in a chromatographic run of only 10 min using LC-MS/MS. The analyte differentiation was achieved both chromatographically on a Accucore aQ C18 column, 100  2.1 mm, 2.6 mm particles, and using high-resolution MS/MS detection [59].

Analytes level in the injected sample HPLC analysis is most frequently performed for the determination of the precise level of the analytes in samples. However, in many cases, a ballpark knowledge of this level can be known and the development of the method on pure standards of the analyte is able to establish a calibration range that accommodates the range of concentrations in the real samples. The analytes level in the injected sample can be described as follows: (1) all analytes at high level (e.g., percent or mg/mL concentration), (2) all analytes at mg/mL level, (3) all analytes at ng/mL or below ng/mL level (trace), (4) large difference between the analytes’ levels in the sample. It must be indicated that in a set of samples, the concentration of the analytes may vary considerably. Overall, the main concern should be that most of the samples have a response within the calibration range for the analysis. Unique samples that fall outside the calibration range are either diluted if they are too concentrated, or subject to additional sample preparation for being concentrated. In some instances, even the ballpark level of the analytes in the sample is completely unknown. In such cases, preliminary experiments must be performed to obtain information about the levels that must be measured.

Samples in high-performance liquid chromatography analysis

245

1) For samples having a high concentration of the analytes, the main precaution should be to not overload the separation capability of the column (see Section 4.5) and to not overload the detector. When the response linearity of the detector starts to be affected, the precision of the results diminishes. A common solution when analyzing too concentrated samples, and not having also other analytes at much lower level, is to perform a sample dilution and reanalyze the sample. Samples with concentration of analytes beyond the level of the highest calibration standard are not uncommon in a series of analyses. In such cases, either the specific sample is diluted and reanalyzed, or, in cases when the concentrated sample exceeds only slightly the calibration range, an attempt can be made to extended the calibration and include in it the sample with slightly larger analyte concentration. 2) Samples with all analytes at mg/mL level are usually analyzed without many problems using various detectors. Many detectors are usually capable to measure sample in this range of concentration (in the injected sample). In some instances when samples are at low mg/mL level (1e3 mg/mL), some detectors with lower sensitivity such as refractive index detector (RID), evaporative light-scattering detector (ELSD), or conductometric detectors may have some problems to measure precisely the analytes, but in most cases, even those detectors are adequate for measurement. Also, for some MS, MS/MS, or fluorescence detectors, concentrations higher than 3e4 mg/mL may be too high for maintaining the linearity of calibration. When the detector is not sensitive enough, either a concentration step should be included in the sample preparation or another detector should be used (if possible). In case of detectors that are too sensitive, either a dilution step for the more concentrated samples is necessary or extension of the calibration range similar to the case of slightly too concentrated samples previously described. 3) When analytes are at low level in the range of ng/mL or below, in many cases, only specific detectors can be used for analysis. These include MS, MS/MS, fluorescence, and CL. The selection of the type of detection is determined by the response of the analyte to the specific detector. An increase in the capability to measure very low concentration of analyte can be obtained, for example, by increasing (electronically) the sensitivity of the detector. The increase in the background noise may limit this alternative. Another alternative to increase sensitivity is the injection of a large volume of sample, if the separation is not badly affected. Sample preparation with various procedures for concentrating the analytes is applied when the increase in injection volume, and in the sensitivity of the detectors is not possible. 4) A more difficult situation to solve is the case when either the level of an analyte is at very different levels in different samples, or when more analytes are present, when they are at very different levels. One such example is the analysis of vitamins in nutraceuticals. Common multivitamin gummies, for example, may contain vitamin B12 at levels of few mg/g, while other vitamins such as B3, B5, or B6 in the range of several mg/g. The difference in concentration of vitamins in those samples can be of four orders of magnitude in the same sample, or in different samples, the lowest level of some analytes and the highest level of other analytes can be even more distant. For the analysis of such sample, it is frequently necessary to perform two separate dilutions, one low and another high, and analyze the sample twice. One such example is the analysis of drotaverine and its impurities. For the analysis of impurities, a concentrated sample is utilized generating, for example, the chromatogram from Fig. 5.3.3. For the measurement of drotaverine, a different dilution is necessary. The separation for the impurities of drotaverine was performed on a Zorbax XDB C18 column (150  4.6 mm and 3.5 mm particle size) in experimental conditions described in the literature [60].

246

Method Development in Analytical HPLC

Figure 5.3.3 UV-HPLC chromatograms showing impurities (indicated as 1, 2, . ,9) in a drotaverine sample separated on a Zorbax XDB C18 column.

The analysis of samples with a wide range of concentration depends on the dynamic range of the detector, and in some cases, sample with large difference in the concentration can be analyzed in one run, but when requirements of high productivity are not imposed, the use of two dilutions may be preferred.

Complexity of the matrix The matrix is an important part of the sample that must be considered when developing a new analytical method, or when adopting an analytical method from the literature, mainly when the method was applied for the analytes of interest but with a different matrix. The matrix characteristics of the raw sample are very important when developing a sample preparation procedure which in most cases is addressing the simplification or the quantity reduction of the matrix in the sample. This subject is further presented in Chapter 6. The matrix may consist of a unique material, for example, when the analytes are impurities in a main compound, or may consist of a complex mixture. In all cases, special attention must be given to both the separation of the analytes from the matrix and also to the influence of the remaining matrix on analytes quantitation. For the matrix remaining in the processed sample, several aspects must be considered. 1) In some cases, the matrix of the sample is chemically different from the analytes, but is left as the main component of the injected sample. Such cases are very frequent, for example, in the analysis of many biological samples for drug metabolites, in the analysis of pesticides in the environment, in the analysis of additives to food or consumer products, etc. In such cases, the chromatographic process is directed toward the separation of the matrix from the analytes in addition to the analytes’ separation among themselves. In case of a large remaining matrix, this can influence detection, even if the chromatographic separation shows different retention times for the matrix and for the analytes. Large peak(s) of the matrix can still be tailing toward the analytes. This is the case, for example, of many biological samples where some

Samples in high-performance liquid chromatography analysis

247

proteins and phospholipids (e.g., from plasma) remain in the injected sample and it is eluted at the front of the chromatogram and still may influence the quantitation of the analytes. An example was shown in Fig. 5.3.1 when the tail of a major peak can influence the analysis of trace compounds. This type of potential interference must be evaluated based on the influence of the matrix on spiked samples with pure analytes and verify the matrix influence as previously described using Eq. 4.6.8 regarding the matrix influence on quantitation. Special quantitation techniques were also developed in case the matrix has effects on the quantitation and this effect must be eliminated [61]. One other case of a large matrix separated from the analyte which still can affect the application of the analytical method is that when the matrix components are very late eluting compounds and are not eliminated from the column at the end of a chromatographic run. Such compounds are accumulated and alter the column characteristics after a low number of analyzed samples. In such cases, the separation must use a strong “cleaning” step in the separation, or cleaning of the column should be performed regularly after a number of runs. Particular attention must be given to the part of matrix that is not detected in the method although it is present. If the detection is performed using a selective detector, for example, a UV detector set to measure absorbance at a specific wavelength, or an MS detector set to measure a specific mass (molecular ion in MS or product ion corresponding to a specific transition molecular ion / product ion in MS/MS), it is possible that the presence of matrix in the eluate at the same time with the analyte is not noticed as a chromatographic peak, but still may affect the quantitation. For this reason, effort must be dedicated to verify that the matrix effect ME value is around 100%. (see Eq. 4.6.8). In many analyses, the presence of the matrix even if not subject of detection may interfere with the quantitation. For evaluating the influence of matrix on the quantitation, it can be experimented to evaluate how an increase (decrease) in the matrix concentration affects the quantitation. For example, if ME is affected by a certain matrix, a more diluted matrix may be evaluated with a low analyte spiking and verify the level at which the matrix is acceptable in the analyzed sample. Sample preparation must be utilized to reduce the matrix to an acceptable level of presence. 2) In case that the matrix contains compounds that are similar in chemical nature with the analytes, special attention must be given to the separation process. The same attention as given to the separation of the analytes should be given to the separation of such matrix components. In case the separation is not entirely possible, selective detection may be utilized (e.g., MS/MS) to differentiate the coeluting analyte/matrix component, and verification of lack of interference in the measurement of the analytes must be performed. 3) It is possible that the matrix of the samples to be analyzed for a specific set of analytes can vary from sample to sample (is not always the same). In such cases, the task of the method development is more complicated and it is recommended that each potential interfering compound is evaluated individually for good separation from the analytes. 4) In most analyses that are affected by a matrix problem, it is advisable to attempt to improve the method by eliminating systematically the problems related to the matrix, instead of attempting an alternative method that will very likely encounter the same problems.

248

Method Development in Analytical HPLC

Key points • •

Sample complexity and the ballpark level of its components can be known either from the preliminary information about the sample or from preliminary experiments. The decisions in developing an HPLC analytical method are influenced by the sample complexity regarding the chemistry of the analytes, number and levels of analytes, and matrix components in the sample.

5.4

Reference materials, commercial reagents, solvents, and additives

Development, validation, and application of an HPLC method need reagents and solvents of high purity, and also reference materials (RMs). In some circumstances, such as for the interlaboratories comparisons, the use of certified RMs and the same source for reagents and solvents is necessary. According to ISO 17034, the definition of RM for HPLC analysis refers to a mixture that is sufficiently homogeneous and stable for being used during a specified time frame. A certified reference material (CRM), also known as standard reference material (SRM), must be characterized by a metrologically valid procedure for one or more specified properties, being accompanied by a certificate providing information on its specified property, as well as the associated uncertainty including a statement of metrological traceability. More details regarding terminology on this topic can be found on Ref. [62]. CRMs are produced by academic, industrial, and government sectors for various purposes, including analytical applications in research and development activity. Production of these materials is basically dependent on the reliable and thoroughly validated analytical methods, but these methods at their turn are usually based on CRMs. Such vicious circle may occur especially in new areas of analysis. This is the case of CRMs for persistent organic pollutants in water, the case of fine particle air pollutants (PM2.5) certified for specific ions, or the case of microplastic for environmental monitoring where validated analytical methods have not been entirely demonstrated by interlaboratories comparison regarding the values of the certified parameters [63]. Many CRMs are produced by National Metrology Institutes (NIMs), such as the National Institute of Standards and Technology (NIST) in the United States [64], the European Commission Joint Research Centre of Directorate F, Health Consumers and Testing (BAM) in Germany; National Measurement Institute of Australia (NMIA), National Council of Canada (NRCC), National Institute of Metrology from China (NIMC), National Metrology Institute of Japan (NMIJ), Korea Research Institute for Science and Standards (KRISS), and others [65]. Producers of these standards are required to publish a certification report describing the preparation mode, homogeneity testing, stability and purity assessment, analytical measurements, and value assigned approach for each standard. Certification must include the approach by which the materials were collected, their preparation, and how they were homogenized, as well as the statistical methods used for data interpretation of the analytical results.

Samples in high-performance liquid chromatography analysis

249

Fields of applications of the produced CRM are environmental and food control, pharmaceutical analysis, bioanalytical research, tobacco industry, and others. NIST has already updated documents concerning various CRMs produced by this authority [66]. As an example, NIST in collaboration with the National Institutes of Health’s Office of Dietary Supplements (NIH-ODS) have developed SRM 972 for the determination of 25-hydroxyvitamin D, 25(OH)D, in serum samples. This standard consists of four serum pools of vitamin D metabolites at different concentration levels, certifying the reference values for 25(OH)D2, 25(OH)D3, and 3-epi-25(OH)D3 using a combination of three isotope-dilution MS methods [67]. In another example, a new tobacco filler SRM has been issued by NIST in 2016 with certified and reference mass fractions for nicotine, N-nitrosonornicotine, 4-(methylnitrosoamino)-1-(3-pyridyl)-1-butanone, and volatiles [68]. Standard materials prepared in-house (e.g., contaminated test batches and target analytes spiked into blank matrices) are often used as quality control (QC) samples. One of the main conditions is the blank matrices to not contain the target analyte(s), which may be a challenge in some fields of application, such as in the determination of toxins in food samples [69]. QC standards are very important in the assessment of accuracy, precision, uncertainty, and traceability of HPLC methods. They are selected prior to the analysis and analyzed under the same analytical conditions with the calibration standards and real samples. Solvents, additives, and reagents are available commercially under different purity grades in order to meet the requirements for different HPLC analyses. Generally, the purity of HPLC solvents is higher than 99.9%, and the levels of few impurities are indicated by the manufacturer on the bottle label. In practice, some of the solvents should be filtered through 0.22 mm filter and degassed before use, but these procedures may not be necessary when the solvent provider assures a specific quality. Depending on the purpose of HPLC analysis, the solvents can be classified in three classes: HPLC grade, gradient grade, and LC-MS grade. Their qualities can be observed mainly in the background noise of the chromatogram. In gradient elution, it is possible that some impurities to be accumulated in the head of the column and may elute during the increase of the organic solvent content in the mobile phase, resulting in socalled ghost peaks that can interfere with other peaks. Water and organic solvents with LC-MS grade are characterized by low mass noise level, and minimal organic and metal contamination [70]. Various impurities, such as alkali salts, plasticizers (such as phthalates), or surfactants, that can be commonly found in lower-grade solvents may pose problems to the sensitivity in LC-MS, due to the high background noise and formation of adducts [71]. For example, a recommended standard for checking the quality of solvents in LC-MS is reserpine, whose MHþ ¼ 609.1 signal in ESI or in APCI is suppressed in accordance to the levels of the interfering trace contaminants that can be present in acetonitrile. Storage of RMs, solvents, additives, and reagents must be performed using recommendations and/or specific protocols. These may indicate the storage at a specific temperature, in absence of light, or protected from moisture. Absorption of water, for example, for a standard material will generate incorrect results when standard solutions

250

Method Development in Analytical HPLC

are made using such a standard by weighing. Even during the weighing process, the absorption of water can be a problem when the standard is a hygroscopic material. When weighing volatile compounds, the preparation of solutions that are assumed to have a precise concentration may be a problem. Such problems are avoided by performing weighing of standard in carefully established conditions such as using closed containers. Quality of water is essential for all types of HPLC analysis. HPLC-grade water is ultra-pure water prepared by a combination of methods including reverse osmosis, ion exchange, distillation, filtration, UV irradiation, and other methods. The American Society for Testing and Materials (ASTM) recommended the quality for water in HPLC analysis represented by high resistivity (>18 mU), total organic content (TOC)d below 10 ppb, and bacteria level below 1 CFU/mL.

Key points • •

Reference materials are important in sample preparation for assessing the functionality of an analytical method. Solvents and reagents purity must be always considered when developing an analytical method.

References [1] S.C. Moldoveanu, V. David, Derivatization methods in GC and GC/MS, in: P. Kusch (Ed.), Gas Chromatography: Derivatization, Sample Preparation, Application, IntechOpen, London, 2018. [2] N. Kuramoto, New definitions of the kilogram and the mole: paradigm shift to the definitions based on physical constants, Anal. Sci. 37 (2021) 177e188. [3] S.C. Moldoveanu, A liquid chromatography-tandem mass spectrometry method for the analysis of sucralose and five glycoside sweeteners, Sep. Sci. Plus 6 (2023) 2300068. [4] J. Reijenga, A. van Hoof, A. van Loon, B. Teunissen, Development of methods for the determination of pKa values, Anal. Chem. Insights 8 (2013) 53e71. [5] http://www.chemaxon.com. [6] J.C. McGowan, Molecular volumes and structural chemistry, Rec. Trav. Chim. Pays-Bas 75 (1956) 193e208. [7] M.L. Connolly, Computation of molecular volume, J. Am. Chem. Soc. 107 (1985) 1118e1124. [8] M. Ptitejean, On the analytical calculation of van der Waals surfaces and volumes: Some numerical aspects, J. Comput. Chem. 15 (1994) 507e523. [9] Y.H. Zhao, M.H. Abraham, A.M. Zissimos, Fast calculation of van der Waals volume as a sum of atomic and bond contributions and its application to drug compounds, J. Org. Chem. 68 (2003) 7368e7373. [10] A.T. Balaban, J. Devillers, Topological Indices and Related Descriptors in QSAR and QSPAR, CRC, Boca Raton, 2000. [11] B. Lucic, I. Lukovits, S. Nikolic, N. Trinajstic, Distance-related indexes in the quantitative structure-property relationship modeling, J. Chem. Inf. Comput. Sci. 41 (2001) 527e535.

Samples in high-performance liquid chromatography analysis

251

[12] P.W. Carr, J.W. Dolan, U.D. Neue, L.R. Snyder, Contributions to reversed-phase column selectivity. I. Steric interaction, J. Chromatogr. A 1218 (2011) 1724e1742. [13] S.C. Moldoveanu, A. Savin, Aplicatii in Chimie ale Metodelor Semiempirice de Orbitali Moleculari, Edit. Academiei RSR, Bucuresti, 1980. [14] H. Heinz, U.W. Suter, Atomic charges for classical simulations of polar systems, J. Phys. Chem. B 108 (2004) 18341e18352. [15] A.K. Rappe, W.A. Goddard III, Charge equilibration for molecular dynamics simulations, J. Phys. Chem. 95 (1991) 3358e3363. [16] S. Fliszar, Atomic Charges, Bond Properties, and Molecular Energies, Wiley, Hoboken, 2009. [17] J.J.P. Stewart, MOPAC-7, QCPE 113, Indiana Univ, Bloomington, 1994. [18] M.J. Frisch, A. Frisch, J.B. Foresman, Gaussian 94, Gaussian Inc., Pittsburgh, 1995. [19] T. Racek, O. Schindler, D. Tousek, V. Horský, K. Berka, J. Koca, R. Svobodova, Atomic Charge Calculator II: web-based tool for the calculation of partial atomic charges, Nucleic Ac. Res. 48 (W1) (2020) W591eW596. [20] A.L. McClellan, Tables of Experimental Dipole Moments, Freeman & Co., San Francisco, 1963. [21] S.C. Moldoveanu, V. David, Dependence of the distribution constant in liquideliquid partition equilibria on the van der Waals molecular surface area, J. Sep. Sci. 36 (2013) 2963e2978. [22] C. Hansch, A. Leo, D. Hoekman, Exploring QSAR. Hydrophobic, electronic, and steric constants, ACS Prof. Ref. Books, ACS, Washington, DC (1995). [23] V.N. Viswanadhan, A.K. Ghose, G.R. Revankar, R.K. Robins, Atomic physicochemical parameters for three-dimensional structure direct quantitative structure-activity relationships. 4. Additional parameters for hydrophobic and dispersive interactions and their application for an automated superposition of certain naturally occurring nucleoside antibiotics, J. Chem. Inf. Comput. Sci. 29 (1989) 162e172. [24] T. Cheng, Y. Zhao, X. Li, F. Lin, Y. Xu, X. Zhang, Y. Li, R. Wang, L. Lai, Computation of octanol-water partition coefficients by guiding an additive model with knowledge, J. Chem. Inf. Model. 47 (2007) 2140e2148. [25] C. Hansch, A.J. Leo, Substituent Constants for Correlation Analysis in Chemistry and Biology, J. Wiley, New York, 1979. [26] S.C. Moldoveanu, V. David, Sample Preparation in Chromatography, Elsevier, Amsterdam, 2002. [27] C. Hansch, A. Leo, Exploring QSAR. Fundamentals and applications in chemistry and Biology, ACS Prof. Ref. Books (1995). ACS, Washington DC. [28] http://www.epa.gov/oppt/exposure/pubs/episuite.htm. [29] http://www.daylight.com/. [30] S.G. Machatha, S.H. Yalkowsky, Comparison of the octanol/water partition coefficients calculated by ClogP, ACDlogP, and KowWin to experimentally determined values, Int. J. Pharm. 294 (2005) 185e192. } [31] D. ErTs, I. Kövesdi, L. Orfi, K. Takacs-Novak, G. Acsady, G. Kéri, Reliability of log P predictions based on calculated molecular descriptors. A critical review, Curr. Med. Chem. 9 (2002) 1819e1829. [32] N. El Tayar, H. van de Waterbeemd, B. Testa, The prediction of substituent interactions in the lipophilicity o disubstituted benzenes using RP-HPLC, Quant. Struct.-Act. Relat. 4 (1985) 69e77. [33] A. Sandi, A. Bede, L. Szepesy, G. Rippel, Characterization of different RP-HPLC columns by a gradient elution technique, Chromatographia 45 (1997) 206e214.

252

Method Development in Analytical HPLC

[34] S. Moldoveanu, V. David, Essentials in Modern HPLC Separations, second ed., Elsevier, Amsterdam, 2022. [35] W.J. Lyman, W.F. Reehl, D.H. Rosenblatt, Handbook of Chemical Property Estimation Methods, ACS, Washington, 1990. [36] C.R. Wilke, P. Chang, Correlation of diffusion coefficients in dilute solutions, AIChE J. 1 (1955) 264e270. [37] Z.E. Rassi (Ed.), Carbohydrate Analysis by Modern Liquid Phase Separation Techniques, second ed., Elsevier, Amsterdam, 2021. [38] G.J. Gerwig, The Art of Carbohydrate Analysis, Springer Nature, London, 2021. [39] J. Rohrer, Carbohydrate Analysis by High-Performance Anion-Exchange Chromatography with Pulsed Amperometric Detection (HPAE-PAD), Technical Note 20. [40] M.A. Alterman, P. Hunziker (Eds.), Amino Acid Analysis; Methods and Protocols, Springer Science, Berlin, 2012. [41] https://www.imtaktusa.com/wp-content/uploads/2015/04/Intrada-Amino-Acid.pdf. [42] S. Moldoveanu, Aplicatile Teoriei Grupurilor in Chimie, Stintifica & Enciclopedica, Bucuresti, 1975. [43] R.M. Silverstein, G.C. Bassler, T.C. Morrill, Spectrometric Identification of Organic Compounds, fourth ed., Wiley, New York, 1981. [44] D. Roy, C.L. Colyer, Nitrogen-doped carbon dots aid in the separation of ssDNA molecules of different length by capillary transient isotachophoresis (ctITP) with laser-induced fluorescence (LIF) detection, J. Chromatogr. A 1641 (2021) 461990. [45] L. Gamiz-Gracia, A.M. García-Campana, J.F. Huertas-Perez, F.J. Lara, Chemiluminescence detection in liquid chromatography: applications to clinical, pharmaceutical, environmental and food analysis - a review, Anal. Chim. Acta 640 (2009) 7e28. [46] T.J. Novak, M.L. Grayeski, Acridinium-based chemiluminescence for high-performance liquid chromatography detection of chlorophenols, Microchemical J. 50 (1994) 151e160. [47] H. Kodamatani, H. Yamasaki, T. Sakaguchi, S. Itoh, Y. Iwaya, M. Saga, K. Saito, R. Kanzaki, T. Tomiyasu, Rapid method for monitoring N-nitrosodimethylamine in drinking water at the ng/L level without pre-concentration using high-performance liquid chromatography-chemiluminescence detection, J. Chromatogr. A 1460 (2016) 202e206. [48] G. Graça, Y. Cai, C.-H.E. Lau, P.A. Vorkas, M.R. Lewis, E.J. Want, Automated annotation of untargeted all-ion fragmentation LCeMS metabolomics data with MetaboAnnotatoR, Anal. Chem. 94 (2022) 3446e3455. [49] O. Baars, D.H. Perlman, Small molecule LC-MS/MS fragmentation data analysis and application to siderophore identification, in: J. Valdman (Ed.), Applications from Engineering with MATLAB Concepts, IntecOpen, London, 2016. [50] A.W. Schultz, J. Wang, Z.-J. Zhu, C.H. Johnson, G.J. Patti, G. Siuzdak, Liquid chromatography quadrupole time-of-flight mass spectrometry characterization of metabolites guided by the METLIN database, Nat. Protoc. 8 (2013) 451e460. [51] T. Kind, H. Tsugawa, T. Cajka, Y. Ma, Z. Lai, S.S. Mehta, G. Wohlgemuth, D.K. Barupal, M.R. Showalter, M. Arita, O. Fiehn, Identification of small molecules using accurate mass MS/MS search, Mass Spectrom Rev 37 (2018) 513e532. [52] B.E. Conway, Electrochemical Data, Elsevier, Amsterdam, 1952. [53] https://antecscientific.com/products/techniques/electrochemical-detection/. [54] J.P. Hutchinson, J. Li, W. Farrell, E. Groeber, R. Szucs, G. Dicinoski, P.R. Haddad, Universal response model for a corona charged aerosol detector, J. Chromatogr. A 1217 (2010) 7418e7427. [55] M. Eckardt, M. Kubicova, T.J. Simat, Universal response quantification approach using a Corona Charged Aerosol Detector (CAD) e application on linear and cyclic oligomers

Samples in high-performance liquid chromatography analysis

[56] [57]

[58] [59] [60]

[61] [62] [63]

[64] [65] [66] [67]

[68]

[69] [70] [71]

253

extractable from polycondensate plastics polyesters, polyamides and polyarylsulfones, J. Chromatogr. A 1572 (2018) 187e202. K. Schilling, U. Holzgrabe, Recent applications of the Charged Aerosol Detector for liquid chromatography in drug quality control, J. Chromatogr. A 1619 (2020) 460911. S.C. Moldoveanu, J. Zhu, N. Qian, Analysis of traces of tobacco-specific nitrosamines (TSNAs) in USP grade nicotine, e-liquids, and particulate phase generated by the electronic smoking devices, Beitr. Tabak. Intern. 27 (2017) 86e96. Imtakt Technical Report, TR06A. F. Barbetti, C. Yang, D. D’Addonna, C. Klaas, Thermo Scientific, Application Note 65901. T. Galaon, V. David, Influence of mobile phase pH on the retention and selectivity of related basic compounds in reversed-phase liquid chromatography, Rev. Roum. Chim. 57 (2012) 131e140. M.C. Ortiz, L. Sarabia, Quantitative determination in chromatographic analysis based on nway calibration strategies, J. Chromatogr. A 1158 (2007) 94e110. https://www.nist.gov/srm/srm-definitions. H. Emteborg, J. Seghers, S. Garcia-Ruiz, S. Elordui-Zapatarietxe, A. Breidbach, K. Labibes, J. Charoud-Got, R. Koeber, Paving the way for new and challenging matrix reference materials-particle suspensions at the core of material processing providing RMs for method development and method validation, Anal. Bioanal. Chem. 416 (2023) 2079e2088. https://www.nist.gov/srm. S.A. Wise, What is novel about certified reference materials? Anal. Bioanal. Chem. 410 (2018) 2045e2049. https://www.nist.gov/system/files/documents/2023/05/15/SRM%20Catalog.pdf. K. Phinney, M. Bedner, S. Tai, V. Vamathevan, L. Sander, K. Sharpless, S. Wise, J. Yen, R. Schleicher, M. Chaudhary-Webb, C. Pfeiffer, J. Betz, P. Coates, M. Picciano, Development and certification of a standard reference material for vitamin D metabolites in human serum, Anal. Chem. 84 (2011) 956e962. L. Sander, J. Pritchett, Y. Daniels, L. Wood, B. Lang, S. Wise, J. Yen, T. Johnson, M. Walters, T. Phillips, M. Holman, G. Lee, J. Lisko, B. Lane, L. Valentin-Blasini, C. Watson, Development of a cigarette tobacco filler standard reference material, Anal. Chem. 89 (2017) 10461e10467. K. Zhang, M. Phillips, Opinion: multi-mycotoxin references materials, Foods 11 (17) (2022) 2544. A. Lenk, Solvents: an overlooked ally for liquid chromatographyemass spectrometry, Column 14 (2018) 19e21. T. Cajka, J. Hricko, L.R. Kulhava, M. Paucova, M. Novakova, O. Fiehn, O. Kuda, Exploring the impact of organic solvent quality and unusual adduct formation during LCMS-based lipidomic profiling, Metabolites 13 (2023) 966.

Sample preparation for highperformance liquid chromatography analysis 6.1

6

The role of sample preparation for HPLC

Sample preparation is the process of making the raw sample amenable for highperformance liquid chromatography (HPLC) analysis. This process consists of several operations that modify the initial (raw) sample such that the final result is a solution that can be injected in the HPLC instrument for performing a successful analysis. Sample preparation is intercalated between the HPLC method developed for pure analytes and the application of the HPLC method to real samples. For this reason, it must be adjusted based on both the characteristics of the sample, and the conditions developed in the core HPLC method initially proved to be adequate only for the pure set of analytes (e.g., Ref. [1]). Besides placing the sample in a solution, sample preparation attempts to solve the two most common problems encountered when performing an analysis, the presence of interferences from the matrix (or among the analytes), and the lower concentration of analytes than the detection limit of HPLC detector. The first problem, the cleanup of the sample, is solved by different procedures including the elimination of the matrix components that interfere in the analysis, the fractionation of the sample to generate simpler samples for analysis, or the modification by derivatization of the chemical nature of the analytes (or of the matrix) to enhance the differences between the analytes and the matrix. Most commonly, the role of cleanup is to reduce the matrix of the sample. The second problem is solved by procedures such as the concentration of the analytes, and the enhancement of a specific property of the analyte by derivatization such that the detection technique can measure it better. For improving the sensitivity, concentration and derivatization are typically used, but in some cases, simple cleanup may improve sensitivity. The field of sample preparation is covered by a large volume of publications including dedicated journals, such as Advances in Sample Preparation (Elsevier), peer-reviewed papers in journals, such as Journal of Chromatography A and B, Journal of Separation Science, Analytica Chimica Acta, dedicated books [2e5], as well as information on the web. In this chapter, only a summary of sample preparation procedures is presented. Sample preparation is frequently a process requiring a certain amount of manpower and can be time consuming. The HPLC analysis being mainly an automated process (in particular when autosamplers are used for injection), minimizing the work on sample preparation and directing the task of handling samples loaded with more matrix to the HPLC is usually more economical. For this reason, in many analytical methods, the Method Development in Analytical HPLC. https://doi.org/10.1016/B978-0-443-29849-3.00009-9 Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

256

Method Development in Analytical HPLC

sample preparation step is reduced as much as possible. As the samples must be in a liquid solution for being used for HPLC (except some unique cases such as samples adsorbed on SPME fiber that are desorbed in a special injector for HPLC analysis), in some methods, the only sample preparation step is the raw sample dissolution. This procedure leaves in the sample all the components of the matrix of the raw sample, and the HPLC core analysis must handle those components. The correct balance must be achieved between how extensive should be the sample preparation and when to use “dissolve and inject sample.” In the development of an analytical method, the right level of sample preparation can be tested only after real samples start to be analyzed, and in case the results of the analysis are not satisfactory, the sample preparation (or in some cases even the core analytical method) must be reevaluated. For this reason, the whole process of developing an analytical method can be iterative.

Initial sample processing and sample dissolution The first step in many cases of sample processing is sample homogenization. This is frequently necessary for samples that are solid, semisolid, or mixtures of solid and liquid. Most such samples are not homogeneous, although they may be large enough to be representative (see Section 1.2). However, even for nonhomogeneous samples, a decision must be made whether or not they should be homogenized before analysis, or if specific components must be separated and analyzed independently. A simple such case is a pharmaceutical tablet that has a coating. It can be necessary in some cases to analyze the whole tablet and in other cases, the coating and the tablet interior are analyzed separately. Also, very frequently in the analysis of art or historic objects, no homogenization is utilized and only a specific small part of the sample is used for analysis. Homogenization is always applied when resampling is necessary to use in the analysis only a smaller part of the initial sample. Liquid samples pose, in general, fewer problems regarding homogenization, but some liquids may contain solid particles and then the solids are either separated or the homogenization is necessary. After sample homogenization (if necessary), the next step in sample preparation is either dissolution or extraction. The step of obtaining the analytes in solution is very important and errors in this operation can affect significantly the results of the analysis. In the dissolution process for the purpose of quantitative analysis of the analytes, it must be assured that the following steps are done correctly: (1) the weight of the raw sample is correctly measured, (2) the volume of extracting solvent is known with accuracy and precision, (3) the transfer of the analytes from the sample into a solution is done correctly, and (4) the sample solvent is miscible with the mobile phase of the HPLC developed for the analysis of samples. 1) The precise weighing of the raw (representative) sample may be just a simple operation, but in some cases, problems may occur, for example, due to the presence of moisture in the sample. If the moisture is present, the final results can be reported to the sample “as is” or on a “dry weight basis.” In cases where the sample must be dried, the loss by evaporation or decomposition of the analytes must be avoided. Even if the drying is performed by freezedrying (see Section 6.3), specific attention must be paid to the potential contamination and losses through evaporation (e.g., Ref. [6]).

Sample preparation for high-performance liquid chromatography analysis

257

2) Liquids can be dissolved more easily in specific solvents, making sure that no precipitation occurs. The proper dissolution of liquids is usually only a matter of selecting properly a specific solvent. When the liquid sample is not homogeneous and contains particles, these can be separated using filtration, centrifugation, etc. Specific care must be taken for the possibility to have the analyte distributed in two phases. Solids present in a liquid may adsorb selectively part of the analyte and modify the result of an analysis if the analytes are not redissolved. Specific analytes in solution can be separated by extraction with nonmiscible liquids. Extraction must be selected in such a way to completely recover the analytes. In many cases, at least part of the matrix is extracted together with the analytes. 3) When the dissolved sample is intended to be used for HPLC injection, the solvent dissolving the sample (sample diluent) must be selected to be miscible with the mobile phase of the HPLC method. A discussion of the role of sample solvent in the chromatographic process is presented in Section 8.4

Solid samples can also be either dissolved or extracted. Extractions are applied only when specific analytes are planned to be measured and not the whole sample composition. For either complete dissolution, or for extraction, specific solvents must be used as in the case of liquids samples. When sample are extracted for obtaining the analytes in solution and leaving part of the sample as a solid matrix, special care must be taken for assuring complete extraction of the analytes. This is the case, for example, when samples of a plant material are analyzed and the cellulose (in some cases associated with lignin) is not soluble in common solvents. For assuring complete extraction, such materials must be in a fine ground form. However, extraction of solid compact samples is not recommended to live undissolved material if possible. In case the analytes are extracted from a solid raw sample, the recovery of the analytes must be verified. For this purpose, it is recommended that the material left after extraction is reextracted. A low level of analytes may still be present in the second extract if some liquid from the first extraction remains in the extracted sample. However, for good recovery, the level of analyte in the second extract should not exceed 1%e2% from the concentration of the first extract. In case a larger amount of analyte is present in the second extract, the two extracts can be combined (considering the dilution factor). Special problems appear when the solid sample is nonhomogeneous such as some pharmaceutical products like tablets, jellies, gummies, and various encapsulated types of samples. Also, materials such as protein powders, chocolate, various type of grease, etc., pose various problems when they must be placed in solution (e.g., Ref. [7]). In such cases, for the dissolution of the initial sample, it must be used a solvent that completely dissolves both the matrix and the analytes, even if part of the matrix is further precipitated, for example, by changing the solvent (e.g., Ref. [8]). The precipitation of matrix from a solution containing the analyte may trap some of it and the evaluation of any analyte loss in the precipitated matrix must be evaluated. One procedure for this evaluation is the spiking of the solution containing the whole matrix with a known amount of analyte, and measurement of its recovery after the matrix is precipitated. In some cases, simple extraction in water can be sufficient, such as for the analysis of soluble ionic species from solid samples. Other cases require organic solvents, mixture of solvents, and in some cases, mixture of solvents and additives (e.g., buffers).

258

Method Development in Analytical HPLC

Besides dissolution in an appropriate solvent, some solid samples require special chemical reactions to be transferred into solution. These chemical reactions may be applied to modify the analytes, the matrix, or the entire sample. Some aspects of the use of chemical reactions with the sole purpose of taking the sample into solution are further discussed.

Sample cleanup and sample fractionation One common purpose of sample preparation is the elimination of some and sometimes all of the compounds that form the matrix that may interfere with the analyte measurement, or that affect the HPLC column. The interference in the measurement can be caused by some specific compounds that are difficult to separate by HPLC or by the matrix that interferes nonspecifically with analytes quantitation. Not only the separation or the detection of the analyte may suffer because of a certain matrix but also problems may arise with the deterioration of the chromatographic column caused by the accumulation of the matrix components in the column, or by chemical reactions with the stationary phase. The process of elimination of undesired compounds from the matrix is commonly known as sample cleanup. A variety of sample cleanup techniques and procedures are reported in the literature (e.g., Ref. [3]). Most cleanup procedures are based on separation techniques. For the separation, simple mechanical procedures such as filtration, sedimentation, sieving, membrane filtration, and gel permeation can be used. Other types of separation that are very common are using the differences between analytes and matrix in the partition between two phases. Among these are solvent extraction, solid-phase extraction (SPE), supercritical fluid extraction, accelerated solvent extraction (ASE), microwaveassisted extraction, solid-phase microextraction (SPME), matrix solid-phase dispersion, and chromatography itself (such as preparative chromatography). Separations based on transport rate such as various types of electrophoresis, ultracentrifugation, etc., are also utilized. Some cleanup and fractionation operations involve chemical modifications. One cleanup operation is not always able to eliminate completely the undesired matrix components. These may include compounds interfering with the analytes or other materials that affect negatively the analysis. This calls for further sample preparation operations. Also, many compounds from the matrix may remain in the processed sample after cleanup, and these are separated by the core HPLC process. A different process that simplifies the composition of a sample is sample fractionation. This is the process of separating the sample into a number of subsamples with different chemical composition. In this way, the matrix may be simplified, and also some analytes that are difficult to separate in the HPLC process may be separated during sample preparation. A clear distinction between sample cleanup and fractionation is not always possible, because fractionation, by separating parts of the matrix from the initial sample, can be considered just a specific cleanup process. Sample fractionation is also achieved by the same techniques as sample cleanup with the difference that in sample cleanup only one fraction is prepared for being analyzed and the other is discarded, while in fractionation, different fractions are further analyzed.

Sample preparation for high-performance liquid chromatography analysis

259

Concentration of analytes in the processed sample A distinction must be made between “the increase in analytes concentration” (enrichment), which is performed by sample preparation with the goal of improving the detection and measurement of the analytes, and “the value of analyte concentration.” The increasing in the concentration of a sample consists in obtaining a processed sample with a higher concentration of the analytes as compared either to the raw sample or to a solution of the raw sample. The concentration of the analytes in the processed sample, even when it is higher than in a solution of the raw sample, can be in some instances lower than the one in the initial raw sample, because the dissolution lowers the analyte concentration. The increase in analyte concentration can be achieved by the same procedures as sample cleanup. The elimination of a part of matrix leads to the increase of the ratio analyte versus residual matrix and this leads to analyte concentration. This step may increase the concentration of both the analytes and of some of the remaining compounds from the matrix. If the analysis of the concentrated sample is adversely affected by the matrix, an additional cleanup operation may be necessary. In addition to solvent extraction, SPE, etc., analyte concentration can be obtained after cleanup by solvent partial evaporation, or by solvent drying and redissolving in a smaller solvent volume. The value of the concentration of the analytes in the processed sample is always changed in comparison to that in the raw sample during sample preparation (dissolution, cleanup, concentration). The transfer of one analyte from the raw sample to the processed sample is described by a parameter known as analyte recovery Rec% (expressed in percent) defined by the ratio: Rec% ¼

qproc 100 qraw

(6.1.1)

In Eq. 6.1.1, qproc is the amount of analyte in the whole processed sample, and qraw the amount of analyte in the raw sample. Good analyte recovery is considered when Rec% > 95%. The measurement of analyte recovery in an analytical method is typically done by adding a known amount of analyte in a blank sample and measuring it in the processed sample. This procedure requires the availability of a blank sample. When a blank sample is not available, the amount of analyte qraw is first measured in the raw sample, then a specific quantity qadd is added, and the total amount of analyte qfin is finally measured. The recovery can be calculated as follows: Rec% ¼

qfin  qraw 100 qadd

(6.1.2)

This procedure requires that the analyzed samples are clean enough to not interfere with the analytical measurement. Also, the recovery should not vary significantly when the level of analyte is higher or lower in the sample, except for common variations due to random error affecting the analysis. This should be verified by repeating the process of measuring recovery with different level of added analyte.

260

Method Development in Analytical HPLC

Analyses with low recoveries may be utilized for qualitative analysis but in some cases even for quantitative analysis. Quantitative analysis can be performed using internal standards that have the same recovery as the analyte. When both the analyte and the standard have the same recovery, their ratio still can be used for quantitation. However, it is not always possible to have standards with the same recovery as the analyte, even when using labeled compounds as standards. The addition of an internal standard in the sample processing is always recommended for the control of errors, even when the recovery of the analyte is good. Another parameter that must be considered in sample preparation is the factor of concentration change F (enrichment factor) between the processed sample and the raw sample. This factor is defined by the expression: F¼

½Xproc

(6.1.3)

½Xraw

where ½Xproc is the concentration in the processed sample and ½Xraw the concentration in the initial sample. The value of F depends on the steps in the sample processing (e.g., how much raw sample is taken for processing and how much processed sample results). The factor F is used to convert the concentration ½Xproc which is measured by the analytical instrument into the concentration of interest ½Xraw for the initial raw sample. The higher is the factor F, the higher has been the increase in the concentration of the analyte during sample preparation. Based on the enrichment factor F, the detection limit LOD in the initial sample is modified following the expression: LODinitial ¼

1 LODprocessed F

(6.1.4)

Eq. 6.1.4 shows that the higher is the factor F for concentration change, the lower is the detection limit of an analyte in the initial sample. Therefore, a sample preparation technique that increases the concentration of the analytes has the effect of decreasing the detection limit relative to the initial sample. In sample preparation, the value of F can be lower than 1.0, for example, because a solid raw sample is dissolved in a solvent for being injected in HPLC. In such cases, LODinitial > LODprocessed . A similar expression as that indicated for the limit of detection LOD is valid for the limit of quantitation LOQ.

Sample derivatization Chemical modifications of the sample can be done in relation to sample dissolution, cleanup, enhancing detection, and improving the HPLC separation. A large variety of procedures are used for chemical modification of the sample. The modification can be performed before injection into the column (precolumn derivatization) or after separation (postcolumn derivatization). Some chemical modifications are applied only to the analytes, some are applied only to the matrix, and some affect the whole sample.

Sample preparation for high-performance liquid chromatography analysis

261

Sample derivatization is a common procedure in HPLC, and numerous publications present the subject (e.g., Refs. [2,9]).

Key points • • • •

Sample preparation is the process of making the raw sample amenable for HPLC analysis. Sample preparation must be adjusted to both the HPLC method developed for pure analytes and the composition of raw samples. Different operations such as sample dissolution, cleanup, concentration, and derivatization can be applied for sample preparation. The final solution in which the analytes are ready for injection in the HPLC must be made in a solvent miscible with the mobile phase used in the HPLC method.

6.2

Mechanical processing in sample preparation

Several operations that may take place in sample preparation involve mechanical processing. The main such operations are: (1) grinding and sieving, (2) weighing and volume measuring, (3) filtration, and (4) centrifugation. Each such operation is described in details in the dedicated literature (e.g., Ref. [3]).

Grinding and sieving Grinding and sieving are operations applied to solid samples. The purpose of these operations is the reduction of particle size of the solid sample, and the homogenization of the raw sample (the operation can also be performed separately). The grinding can be performed at laboratory scale using a simple mortar and pestle, but a variety of dedicated instruments also are available for particle size reduction. The models can differ in the technique used for grinding (e.g., cutting, crushing by pressure, friction, and impact of two surfaces), the material used as tool for grinding (e.g., tungsten carbide, agate, or steel), capacity of the grinder, and particle size distribution obtained after grinding (e.g., coarse size >5 mm, fine size >63 mm, and extrafine size 0. Second term in Eq. 6.3.11 is negative which decreases the value of ln Sol which becomes negative and the compound is not soluble. 2) For a polar solvent and a polar solute, DH cav (positive) is large and DH vdW (negative) is even larger and as a result DHcav þ DH vdW < 0. Second term in Eq. 6.3.11 is positive which increases the value of ln Sol and the compound is highly soluble. 3) For a nonpolar solvent and a nonpolar solute, DH cav (positive) is small and DHvdW (negative) is also small and as a result DH cav þ DHvdW z 0. Second term in Eq. 6.3.11 is negligible and the first term in Eq. 6.3.11 prevails and ln Sol remains positive and the compound is soluble

270

Method Development in Analytical HPLC

4) For a nonpolar solvent and a polar solute, DH cav (positive) is small and DHvdW (negative) is also very small and as a result DHcav þ DHvdW > 0. Second term in Eq. 6.3.11 is negative which decreases the value of ln Sol and when the value becomes negative, the compound is not soluble.

These considerations explain the principle of solubility of “like into like” regarding polarity. Because the evaluation of DH cav and of DH vdW is not straightforward, a specific parameter was developed for helping in the estimation of solubility of nonelectrolytes. This parameter is known as Hildebrand solubility parameter d. The value for d for a compound (solute or solvent) can be estimated by the formula:  vap  DH  RT 1=2 dz V

(6.3.12)

In Eq. 6.3.12, V is the molar volume of the solute, and DH vap is its heat of vaporization. Assuming that DH mix z DH vap and obtaining DH vap from the value of d, Eq. 6.3.11 is written as follows:   13 T fus Vsolute 1000 rsolvent ðdsolute  dsolvent Þ2 þ ln ln Sol ¼ 1  R T RT Mwsolvent

(6.3.13)

The maximum value for Sol is obtained when dsolute ¼ dsolvent . This indicates that the closer is the nature of the solute to that of the solvent, the higher is its solubility (principle of solubility of “like into like”). Values for molar volume V and Hildebrand parameter d are reported in the literature (e.g., Ref. [3]). A convenient way to estimate solubility of many nonionic compounds, in particular in water, is based on octanol/water partition coefficient Kow (Hildebrand solubility parameter d has only a poor and negative correlation with Kow ). The evaluation based on Kow is done using expressions of the form: log SolðXÞ ¼ a log Kow;X þ b

(6.3.14)

where a and b are empirical parameters. A list of such empirical parameters for the estimation of water solubility of organic compounds based on log Kow is given in Table 6.3.1. Since log Kow is higher for more hydrophobic compounds, the dependence is inversely proportional for solubility of organic compounds in water (the dependence between log Sol in water and log Kow has a negative proportionality coefficient). Some computer programs also provide calculated solubility values, for example, for solubility in water [18,19]. The dissolution process of a solid material frequently comes to equilibrium slowly, because it takes time to transfer material across the phase boundary. This delay in attaining the equilibrium is a factor to consider in sample preparation using dissolution

Sample preparation for high-performance liquid chromatography analysis

271

Table 6.3.1 Regression equation for the estimation of water solubility of organic nonionic compounds based on log Kow [17]. a

b

Units for Sol

Used for chemical class

1.37 1.113 1.229 1.013 1.182 1.221 1.294 1.294 0.966 1.214 1.237

þ7.26 þ0.926 þ0.720 þ0.520 þ0.935 þ0.832 þ1.043 þ0.248 þ0.339 þ0.850 0.248

mmol/L mol/L mol/L mol/L mol/L mol/L mol/L mol/L mol/L mol/L mol/L

Aromatics, chlorinated hydrocarbons Alcohols Ketones Esters Ethers Alkyl halides Alkynes Alkenes Aromatics Various Alkanes

of solids. The rate of dissolution (without agitation) can be expressed by an equation of the form [20]:  dqðXÞ D ¼A ½Xsurf  ½Xbulk dt L

(6.3.15)

In Eq. 6.3.15, qðXÞ is the mass of the dissolving material (solute), t is time, A is the surface area of the interface between the dissolving material and the solvent, D is the diffusion coefficient of compound X in the solvent, L is the thickness of the boundary layer of the solvent at the surface of the dissolving material, ½Xsurf is the concentration of solute on the surface of the dissolving material, and ½Xbulk is the concentration of the substance in the bulk of the solvent. Eq. 6.3.15 indicates that a larger area A of the dissolving material is favorable to a more rapid dissolution. This explains why larger crystals are more difficult to dissolve than fine powders. Various procedures are used to accelerate the dissolution such as heating, agitation, sonication, size reduction of the particles to be dissolved in order to increase the contact surface between the solid and the liquid, etc. Sonication, for example, uses vibrations with frequencies higher than 16 kHz to produce mechanical stress, heating, and cavitation, all contributing to the acceleration of the dissolution process. Polymer dissolution can pose in many cases specific problems. One such problem is, for example, protein solubilization without degradation [21]. Special procedures have been developed for this purpose.

Crystallization and precipitation Crystallization is the process in which a compound from a solution is separated in the form of crystals. This can be achieved in several ways, such as: (a) partial or total removal of solvent, (b) change of the solvent composition by mixing it with other

272

Method Development in Analytical HPLC

liquids, (c) change in the temperature of the solution which modifies the solubility of the solute, (d) induction of the precipitation using seeding, or (e) modifying the ionic strength of the solution with additives. Details about crystal formation can be found in the dedicated literature [22]. Crystallization that allows the formation of a solid material from a liquid can be followed by a mechanical separation of the solid from the liquid. Because only the compound that crystallizes is assumed to have reached the saturation concentration, the crystal formation is also a purification procedure. However, in the formed crystals, it is possible to remain included some solvent or other impurities. Precipitation is a form of rapid crystallization and usually involves a chemical reaction that produces an insoluble solid compound. The solid can be amorphous but also microcrystallin. In addition to chemical reactions that form precipitates, other procedures include the change in solvent composition by adding another solvent, solvent evaporation, the decrease of the solution temperature, stirring, addition of a salt, etc. Precipitation can be used in sample preparation for the elimination of an undesirable matrix component or for the separation of specific analytes. A common application of precipitation is deproteinization of biological samples. In current applications, this procedure proved more efficient and leading to better precision than liquid extraction in the removal of proteins from biomatrix [23]. The formation of crystalline precipitates in the removal of inorganic matrices is desirable because they include less foreign material from the solution.

Key points • • •

Physical dissolution of samples is a common procedure in HPLC since the samples are placed in the mobile phase flow using an injection. Selection of a proper solvent for dissolving the sample and of standards for an HPLC analysis is important not only for allowing the injection but can affect separation. Solubility of various compounds is usually reported in the literature, but also can be estimated from other properties such as octanol/water partition coefficient.

6.4

Liquideliquid extraction in sample preparation

Liquideliquid extraction (LLE) is a technique using a solvent to extract a compound from another solvent (nonmiscible with the first). The process is based on a partition equilibrium. LLE is utilized in many applications, including in sample preparation for chromatography.

Equilibrium in a simple liquideliquid extraction For a compound X, distributed between two nonmiscible liquid phases from which one is organic “o” and the other is water “w”, the following equilibrium can be considered:

Sample preparation for high-performance liquid chromatography analysis

Xw % Xo

273

(6.4.1)

Equilibrium 6.4.1 is characterized by the equilibrium constant KX;ow described by the formula: KX;ow ¼

½Xo ½Xw

(6.4.2)

In Eq. 6.4.2, ½Xw is the molar concentration of compound X in solvent “w” and ½Xo the molar concentration in solvent “o”. The value of the equilibrium constant KX is given by a formula of the type:

DG0 ðXÞ KX ¼ exp  RT

(6.4.3)

In Eq. 6.4.3, DG0 is the change in free enthalpy during extraction and, at a constant volume, free enthalpy is equal with free energy (DG0 ¼ DA0 ). As indicated in Section 2.2, solvophobic theory shows that the changes in free energy during dissolution of compound X in a solvent are given by the expression (symbol D and the index 0 for standard expressions of free energy are omitted for simplifying the notation): cav Asol þ AvdW X ¼A

(6.4.4)

In Eq. 6.4.4, the values of Acav are positive (energy is consumed to form a cavity in the solvent) and AvdW are negative (energy is generated due to interactions). Based on Eq. 6.4.3 and 6.4.4, the equilibrium constant KX will have the expression: KX ¼ exp 

cav vdW vdW Acav X;o  AX;w þ AX;o  AX;w

RT

! (6.4.5)

The values of KX based on Eq. 6.4.5 have been successfully calculated and reported in the literature for the system water and octanol for a number of compounds X [24]. Qualitative comments can be made based on Eq. 6.4.5 for the case of any pair of solvents and compounds X. Assuming that solvent “w” is polar (e.g., water) and solvent “o” is organic (e.g., hexane, methylene chloride, benzene, etc.), the cavity term will be cav negative Acav X;o  AX;w < 0 since the interactions between the molecules in a polar solvent are stronger than in an organic solvent. Regarding the solute X, two cases are basically possible: 1) The compound X is an organic molecule with low polarity (hydrophobic character). In this case, the cavity formation term in Eq. 6.4.5 will be larger than the van der Waals term (in absolute value) since the interaction of solute X with each of the two solvents will be small. cav vdW vdW cav The term Acav X;o  AX;w þ AX;o  AX;w will be negative (dominated by the value of AX;w ). As

274

Method Development in Analytical HPLC

a result, the value under the exponent will be positive, leading to a high value for KX and the compound will go preponderantly to the organic phase. 2) The compound X is an organic molecule with high polarity. In this case, the term for van der Waals forces (in absolute value) will be larger than the term for cavity formation and the term cav vdW vdW vdW Acav X;o  AX;w þ AX;o  AX;w will be positive (dominated by the value of AX;w which is negative). As a result, the value under the exponent will be negative and polar molecules will remain in the polar solvent the value of KX being small.

The previous results indicate that the extraction in a specific solvent is higher when the polarity of the solvent is similar to that of the extracted compound X, nonpolar compounds being efficiently extracted in nonpolar solvents from the polar ones, and polar compounds remaining in the polar solvent. For the estimation of polarity of different solvents and compounds X, a helpful guidance can be obtained from the octanol/water partition coefficients Kow of each participant in the equilibrium. Other properties and parameters of solvents that allow for the estimation of their polarity and suitability for an LLE are further presented in this Section. In addition to the equilibrium constant K, the extraction can also be characterized by the extracted fraction EX (equivalent to recovery Rec for the component defined by rel. 6.1.1). For example, for an extraction in an organic phase “o” of compound X from an aqueous phase “w”, the extracted fraction EX is given by the formula: EX ¼

½Xo Vo qo KX b 1 ¼1 ¼ ¼ 1 þ KX b qtotal ½Xo Vo þ ½Xw Vw 1 þ KX b

(6.4.6)

In Eq. 6.4.6, q0 is the amount of the extracted solute (e.g., analyte) in the organic phase and qtotal is the total amount of solute, Vo and Vw are the volumes of organic and aqueous phase, respectively, b is the phase ratio with b ¼ Vo =Vw , ½Xo is the molar concentration of the analyte in organic phase, ½Xw is the concentration in water, and KX is the equilibrium constant given by Eq. 6.4.2. The extracted fraction can also be expressed in %. The value for KX is determined by the nature of solute X and that of the two solvents o and w, while b is selected by the operator. The contribution of KX or of b to the extracted fraction EX is the same, and since KX cannot be modified for a given system since the nature of components cannot be changed, the increase in b can improve the extraction efficiency. For achieving the concentration of the solute (analyte) in solvent o (organic phase), the value of b is usually kept relatively small (e.g., b ¼ 0.1). The improvement of the extraction yield can be obtained using repeated extractions, for example, using portions of pure solvent “o” to extract the same water solution a number of times. In this case, for n extractions, the extracted fraction EðnÞX from the total amount, assuming that the volume Vw remains constant after n extractions each with the volume Vo of fresh organic solvent, is given by the expression:  EðnÞX ¼ 1 

n

1 1 þ KX b

(6.4.7)

Sample preparation for high-performance liquid chromatography analysis

275

Figure 6.4.1 Variation of extraction fraction EðnÞX with KX b for n ¼ 1, 2, and 4.

The variation of EðnÞX with KX b for n ¼ 1, 2, and 4 is shown in Fig. 6.4.1. The graphs from Fig. 6.4.1 show that the extracted fraction tends to EX ¼ 1 for indicating recovery of compound X when KX b increases and when the number of repeated extractions increases. The extraction can also be characterized by the enrichment ratio F (enrichment factor has been discussed in Section 6.1). This parameter shows the ratio of the concentration of compound X in organic phase versus the initial concentration of the compound in the aqueous solution ½Xw;init and is given by the expression: F¼

½Xo 1 ¼ EX b ½Xw;init

(6.4.8)

Eq. 6.4.8 shows that the enrichment ratio depends on both EX and b such that F is larger when EX is closer to 1 (maximum possible value) and when Vo is smaller and Vw is larger (smaller b).

Equilibrium in liquideliquid extraction involving multiple solute species The previous discussion regarding distribution of component X between the two nonmiscible liquid phases can be extended from a single compound X to the case when X has more than one form. This can be caused by the fact that X can dissociate in the polar solvent (e.g., in water) and have more than molecular form (e.g., HX can be also present as X  ) or by the presence of tautomerism, dimerization, ion-pairing, or complexation. For a compound X present in solution in a number of forms, X1 ; X 2 ; .:Xn , all being identified as compound X, the equilibrium between phase w

276

Method Development in Analytical HPLC

Figure 6.4.2 Equilibrium between phase w and phase o for a compound X present in two forms, as X1 and X2 , and subject of a chemical equilibrium in phase w.

and o for X is described by the distribution coefficient DX with an expression similar to Eq. 2.1.13: DX ¼

½X1 o þ ½X2 o þ .::½Xn o ½X1 w þ ½X2 w þ .::½Xn w

(6.4.9)

A simplified system in which the compound X is present in equilibrium of only two forms, indicated as X1 and X2 , and is subject of a chemical change in equilibrium in phase w is shown in Fig. 6.4.2. Examples of such equilibria are the isomerization reactions, such as keto-enolic, oxime-nitroso, or even cis-trans transformation, which are influenced by the characteristics of the two nonmiscible liquids. In Fig. 6.4.2, the constants KX1 , KX2 , and K1;2 are given by the expressions: KX1 ¼

½X1 o ½X1 w

KX2 ¼

½X2 o ½X2 w

w K1;2 ¼

½X2 w ½X1 w

(6.4.10)

For the equilibrium shown in Fig. 6.4.2, the distribution coefficient can be written in the form: DX ¼

w ½X1 o þ ½X2 o KX1 þ KX2 K1;2 ¼ w ½X1 w þ ½X2 w 1 þ K1;2

(6.4.11)

In case that an equilibrium between X1 and X2 also takes place in organic phase, an o must be considered, and the value for D will additional equilibrium constant K1;2 X have the expression: DX ¼ KX1

o 1 þ K1;2



w K1;2

where

o K1;2 ¼

½X2 o ½X1 0

(6.4.12)

In cases of LLE of acid or basic compounds, the value of DX depends on pH of the aqueous solution. For example, for a weak acid analyte HX with the acid dissociation constant Ka , the dependence of DX on pH is given by the expression:

Sample preparation for high-performance liquid chromatography analysis

DHX ¼ KHX

1 1 þ 10pHpKa

where KHX ¼

½HXo ½HXw

277

(6.4.13)

Eq. 6.4.13 shows that the acid HX is more extracted in the organic phase when pH < pKa (acidic conditions of the aqueous phase). For a basic analyte, the extraction in the organic phase is favored by higher pH values (basic conditions of the aqueous phase). The distribution coefficient DX is the parameter that describes the extraction equilibrium of many systems involving compounds with acidic, basic, or amphoteric character, as well as in other situations when more than one species of an analyte is present in one or both phases and when the utilization of equilibrium constant KX is not appropriate. For such cases, in the expression of extracted fraction EX , the value of KX must be replaced with DX .

Properties of solvents used in liquideliquid extraction A number of properties of a solvent play a role in its selection as an extracting agent. One important property of extracting solvents is polarity and this is characterized by several parameters such as octanol/water partition coefficient Kow , Hildebrand solubi0 lity parameter d, polarity parameter P that characterizes solvents based on liquidegas partition [25e27], solvatochromic ET ð30Þ and Kamlet-Taft p , a, and b solvatochromic parameters [28,29], and various other types of characterization [30,31]. A set of solvent properties also related to polarity include some molecular characteristics such as dipole moment m, polarizability a, and ionization potential I (e.g., Ref. [32]). Such properties can be utilized, for example, for the estimation of other properties such as solvatochromic parameters. Another series of properties include boiling point (B.p.), density (b), viscosity (h), superficial tension (g), and solubility in other solvents (and miscibility). These properties are related to the possibility of a solvent to be separated from other solvents, how easily it evaporates, what is its tendency to form emulsions, etc. These properties are important for selecting a solvent that leads to a better extraction of specific analytes from a given sample placed in another solvent. Such properties are amply described in literature for a variety of solvents (e.g., Ref. [3]). Solvent properties selected as an extracted agent can be made based on the values of different parameters with the goal of producing a close to complete extraction of the analytes of interest. Criteria of selecting the solvent can be obtained by considering the observation regarding the variation of equilibrium constant K, but also how much matrix the solvent extracts, what evaporative properties the solvent has, if it is toxic or flammable, etc. The use of the extracting solvent may involve either the extraction of the analyte or the separation of specific matrix components. For HPLC applications, it is common that an extracting solvent is further used for injection of the sample in the HPLC system. In such case, additional properties of the extracting solvent must be considered, including compatibility with the mobile phase, potential to influence separation as a sample solvent, and even potential to affect detection.

278

Method Development in Analytical HPLC

In some cases, the solvent extract is either inappropriate for the chromatographic analysis, or the solution is too diluted and its volume must be reduced. In such cases, the organic solvent can be evaporated (in part or completely). After this, the processed sample can be reconstituted in the desired solvent and volume. The evaporation must be done without losses of analytes (by their evaporation or decomposition). In such cases, the boiling point of the solvent must be considered, and volatile solvents are preferred to be easily evaporated. The degree of miscibility of the two phases involved in the extraction, viscosity, and tendency to form emulsions also should be taken into consideration in the choice of a solvent.

Types of liquideliquid extraction LLE has been utilized in the form of various techniques. A common procedure is the conventional LLE performed by stirring two nonmiscible solvents, the one initially containing the compound(s) to be extracted and the one in which these compounds are transferred, followed by the separation of the two phases (e.g., with the help of a separatory funnel). Other procedures use specially constructed extractors that allow distillation and condensation of the extracting solvent which is continuously percolated through the initial solvent (similar to Soxhlet extraction for solids), various countercurrent extraction devices, etc. A number of special LLE procedures are also used. Among these, liquid-phase microextraction (LPME) or liquid-liquid microextraction (LLME) procedures have been frequently utilized, in particular for analyte concentration. The main difference from conventional procedures consists in the fact that the volume of aqueous solution (the donor phase) in LLME is the same as in conventional laboratory extractions (in the range of mL), but the extracting solvent (acceptor phase) is the range of mL. Some LLME types are listed in Table 6.4.1.

Table 6.4.1 Several types of liquideliquid microextraction techniques. Liquid-phase microextraction technique

Acronym

Single drop-phase microextraction Membrane-assisted solvent extraction Microporous membrane liquideliquid extraction (LLE) Hollow fiber liquid-phase microextraction Liquideliquideliquid-microextraction Dispersive liquideliquid microextraction Salting-out assisted LLE Liquideliquid extraction with low temperature partitioning Cloud-point extraction Electrochemically modulated LLE Continuous flow microextraction Drop to drop microextraction Directly suspended droplet microextraction Supported liquideliquid microextraction

SDME MASE MMLLE HF-LPME LLLME DLLME SALLE CPE CFME DDME DSDME SLLME

Sample preparation for high-performance liquid chromatography analysis

279

Additional information regarding LLME can be found in the literature [33e38].

Separations using liquideliquid extraction LLE as a sample preparation technique can also be used for separation purposes between analytes. The selectivity of the extraction process between two analytes X and Y can be characterized by a separation factor a. This factor (for one step extraction) can be defined by the ratio: a¼

EX KX ð1 þ bKY Þ ¼ EY KY ð1 þ bKX Þ

(6.4.14)

where EX and EY are the extracted fractions for the two compounds, KX and KY are the equilibrium constants, and b ¼ Vo =VW . The ratio of the concentrations of the compounds remaining in the aqueous phase can be characterized by the ratio ð1  EX Þ= ð1  EY Þ. Larger values for b correspond to a better separation between the two species, with the extraction of species X in the organic phase. LLE can also be used in various separations necessary in the analysis of complex biological samples. As an example, for the analysis of nucleic acids are used detergents, chaotropes, and/or heat to release the nucleic acids from the biological material into a solution [39]. Following this step, LLE procedure using a mixture of phenol and chloroform can be applied for the extraction of the denatured proteins, lipids, and other cellular components of biological matrix, while nucleic acids remain in the aqueous phase and can be further analyzed [40].

Key points • • •

LLE is a technique that uses a solvent to extract a compound of interest from another solvent, and can be used as sample preparation for HPLC. LLE is useful for analytes concentration. Various alternatives of LLME are also applied in sample preparation for HPLC.

6.5

Liquidesolid extraction in sample preparation

Liquidesolid extraction (LSE) is the process of using a solvent for extracting compounds from a solid sample. The compounds to be extracted can be the analytes that must be separated from the matrix or, less often, specific matrix compounds that must be eliminated from the sample. LSE is a very common procedure to place samples in a solvent which is necessary for injection in HPLC. In many cases, the extract can be directly used for the HPLC injection, but it can also be further concentrated or diluted, or even subject to additional sample preparation.

280

Method Development in Analytical HPLC

Extraction efficiency in liquidesolid extraction In solideliquid extraction, it can be assumed that the solubility of the solute X in the extracting solvent is high. In such case, the rate of extraction in LSE is controlled by diffusion and can be approximated based on Fiks’s law using the expression: d½X A$D$Ve ¼ ð½Xs  ½Xe Þ dt d

(6.5.1)

In Eq. 6.5.1, ½Xs is the concentration of the solute in the sample, ½Xe is the concentration in the extracting solution, A is the surface area of the solid sample, D is the diffusion coefficient of the solute in the sample soaked with solvent and depends on temperature and the nature of solute and solvent, Ve is the volume of extracting solution, and d is the diffusion layer thickness inside the solid in which the concentration of X reaches ½Xs (closer to the surface the concentration of X decreases in the solid). The integration of Eq. 6.5.1 leads to the expression: ½Xe ¼ ½Xs ½1 expðkSLE tÞ where kSLE ¼ A D Ve = d

(6.5.2)

Extraction efficiency is described by a parameter E with the formula: E¼

½Xe ¼ ½1  expðkSLE tÞ ½Xs

(6.5.3)

Eq. 6.5.3 indicates that extraction efficiency E increases in time and for a given time, is larger when kSLE is larger. The value of kSLE is directly proportional with the area A of the sample, with the volume of extracting solution, and with the diffusion coefficient of the solvent. This indicates that a finely ground sample has a higher kSLE . Also, it indicates that a larger volume of the extracting solution and a higher diffusion coefficient for the solvent improve extraction. A graph showing the variation with t of extraction efficiency for three kSLE values is given in Fig. 6.5.1.

Figure 6.5.1 Variation with t of extraction efficiency E for three kSLE values.

Sample preparation for high-performance liquid chromatography analysis

281

Solvents used in liquidesolid extraction An important requirement for a solvent selection is the high solubility of the solute X in the solvent [41]. In Section 6.3, some criteria for solubility were presented and those criteria should be considered when selecting a solvent for extraction. For the extraction of analytes with hydrophilic character, water or aqueous buffers are generally used for extraction. For analytes with limited solubility in pure water, it is possible to use aqueous buffers with pH < 7 to increase the extraction yield of basic analytes, and aqueous buffers with pH > 7 to increase the extraction yield of acidic compounds. In this way, the analyte molecules become dissociated and their solubility in aqueous solvent is enhanced. Common organic solvents are also used for the extraction of a wide range of organic molecules. The most frequently used such solvents are methanol, acetonitrile, ethanol, and isopropanol. These solvents are also compatible with most mobile phases used in reversed-phase HPLC (RP-HPLC) and hydrophilic interaction liquid chromatography (HILIC). Chlorinated solvents, aliphatic or aromatic solvents, and upper alcohols can also be used for the extraction. Some analytes, such as fats, are soluble only in strongly hydrophobic solvents. In some analytical procedures, the extraction solvent can be evaporated completely or partially, using, for example, a stream of N2, and the analytes can be redissolved in another solvent, or even in the same solvent but using a small measured volume (solvent reconstitution). This procedure can be used for increasing the sensitivity of an analytical method by starting with a larger volume of solvent that contains the extracted analyte and ending with small volume of solvent with a higher analyte concentration. This process should be performed taking all the precautions to not decompose the analytes because of overheating, and to not lose analytes because of their own evaporation.

Types of liquidesolid extraction LSE has a variety of forms of utilization. The most common is simple mechanical shaking. The shaking can be done manually but more frequently is done using mechanical shakers. The shakers are of various types, such as platforms with gyratory movement, roller bottles, wrist action, 3D-type, vibrational, etc. During the extraction, parameters such as temperature, shaking range and speed, etc., can be controlled. Extraction can also be performed using with a heated solvent, either in closed vials or using a boiling solvent with reflux and a special extractor known as Soxhlet is commonly used for this purpose. The use of heat during extraction is based on the increase in solubility of analytes in hot solvents, and also to the decrease in solvent viscosity and increase in diffusion coefficient D in Eq. 6.5.2). Sonication is another common procedure for increasing the rate of dissolution and of solvent penetration in the solid matrix. After the extraction, the solvent containing the compounds of interest must be separated from the remaining solid material. This can be done using filtration, decantation, or centrifugation. The solution remaining soaked in the solid material must be recovered as much as possible. The concentration of these analytes is the same as in the

282

Method Development in Analytical HPLC

whole solution and the recovery is important only for obtaining a larger amount of analytes and not for correcting the concentration of the solution. The ratio of the amount of sample and that of extracting solution must be taken into consideration for quantitative measurements. Repeated washes may solve the problem of analyte loss in the solid material. Losses of analytes during the separation of the extracting solution must also be accounted for. This can be achieved by measuring the initial volume of solvent and that of the recovered solvent, or by using an internal standard in order to track the change in concentration of the analytes during the extraction and solvent separation. Besides simple extraction by mechanical shaking or by heating the sample in the presence of a solvent (or both shaking and heating), several other procedures of extraction are utilized in laboratories. The more common such procedures include: (1) accelerated solvent extraction (ASE), (2) microwave-assisted solvent extraction (MASE), (3) ultrasound-assisted extraction (UAE), and (4) supercritical solvent (fluid) extraction (SFE). 1) ASE can be applied to solids and semisolid samples and is using an increase in the temperature and also the pressure of the extracting solvent such that the solvents can be used at temperatures above the boiling point at atmospheric pressure. This increases the solubility of analytes and diminishes the viscosity of the solvent which improves the migration of solvent in the sample. The solvents used in ASE are the same as those used for other liquid extraction as Soxhlet or sonication such as water, methanol, or other common solvents [42]. When water is used as extraction solvent at pressure, above its atmospheric boiling point but below the critical point of water, the technique is known as pressurized hot water extraction (PHWE). In this case, the dielectric constant of water is decreased, and water becomes less polar being able to extract less polar compounds [43]. PHWE is also important for green considerations [44]. Due to higher extraction efficiency, the extraction times are also reduced in ASE. 2) MASE is based on extracting the sample in a microwave oven that heats the solvent and possibly the sample. Because microwave radiation is not significantly absorbed by nonpolar materials, the solvents must contain a polar component such as water, methanol, acetone, acetonitrile, etc. [45,46]. Several alternatives are known for microwave-assisted extraction instrumentation [47], the more common type consisting of simply placing the sample in sealed vessels in a microwave oven, with the capability of monitoring the temperature and the pressure in the vessels. Other systems allow, for example, the circulation of the extracting solvent. 3) The use of simple sonication of solid samples for enhancing solubility is a common practice in the laboratory. More elaborate sonication systems, with specific sonication frequency are also available [48]. 4) SFE is an extraction procedure that uses a supercritical fluid as the extraction solvent. This is a fluid at temperature and pressure above the critical point (critical temperature Tc and critical pressure pc ). Different compounds can be used as fluid in SFE such as CHClF2, SF6, C3H8, but the most common is CO2. CO2 has significant advantages regarding range of temperatures that can be used during extraction (from ambient up to 100e120 C) and pressures utilized (from tank pressure about 870 psi up to 5000 psi), lack of toxicity, lack of reactivity, etc. SFE is typically applied on solid or semisolid samples. The extracting properties of CO2 can be controlled by varying the temperature and pressure, and because the polarity of supercritical CO2 is low, polar modifiers such as CH3OH or C2H5OH can be added to the supercritical CO2 for increasing its polarity. The variation of extracting properties of CO2 can be controlled in such a way that certain compounds can be extracted with high efficiency and

Sample preparation for high-performance liquid chromatography analysis

283

Figure 6.5.2 Schematic diagram of a supercritical fluid extractor using CO2. other with low efficiency leading to a selective extraction (efficiency of extraction in SFE is described in the same manner as for other extraction procedures using parameter E given by Eq. 6.5.3). Several advantages are related to SFE extraction, such as the use of an environmentally friendly extracting agent and ease of recovering of the extract, the exposure of the sample only to low temperatures avoiding any potential decomposition, the use of an extracting agent that is not toxic or reactive, etc. The extraction instrumentation in SFE is specially designed and a schematic diagram of an SFE extractor is shown in Fig. 6.5.2. The instrument has basically the following components: (1) a component allowing the delivery of liquid CO2 and of the adequate modifier, (2) an extraction cell that is placed in an oven that heats the sample and allows the formation of a supercritical fluid, (3) a separator where the supercritical fluid is released, expands to generate gaseous CO2 which can be recirculated, deposits the extracted material on a specific support (e.g., stainless steel spheres, silica, Tenax, etc.), and where a rinsing solvent is added to collect the extract. Extensive literature is available describing various parameters related to temperatures and pressures of the fluid used for extraction, the use of modifiers, and the conditions used in the separator [49e51].

A comparison of several extracting techniques regarding typical volume of solvent utilized for extraction and typical extraction time is given in Table 6.5.1.

Table 6.5.1 Comparison of several extraction techniques of solid samples. Extraction technique

Typical volume of solvent (mL)

Typical extraction time

Mechanical shaking Soxhlet Sonication Supercritical fluid Accelerated solvent Microwave

1e50 300 200 8e50 15e40 25e50

30e60 min 4e48 h 30e60 min 30e120 min 12e18 min 30e60 min

284

Method Development in Analytical HPLC

As indicated in Table 6.5.1, mechanical shaking is suitable for small samples and even small volume of extracting solvent. For this reason, it is the most common type of extraction used for analytical purposes. However, extraction efficiency may vary by this procedure depending on the nature of sample and of extracting solvent. Other extraction types are use as necessary for various samples such as some more difficult to extract.

Key points • •

Liquidesolid extraction is a very common procedure to place samples in a solvent which is necessary for injection in HPLC (with or without additional sample preparation). Several procedures are known for liquidesolid extraction, the most common can be considered simple mechanical shaking of the sample with the extracting solvent.

6.6

Solid-phase extraction in sample preparation for HPLC

SPE is a sample preparation technique that uses the capability of a solid material (sorbent) to retain selectively compounds dissolved in a liquid (or present as vapors in a gas). The retaining mechanism in SPE sorbent (partition in the sorbent, adsorption, ion exchange, etc.) depends on the nature of the compound to be retained, and of the medium in which the retention takes place. The retention being a reversible process, the retained compound (e.g., the analyte) can be released from the sorbent by changing the nature of the liquid solution that is in contact with the sorbent and its retained compound. The technique has been frequently utilized successfully for the concentration and the cleanup of samples in HPLC. A significant volume of information is available about SPE, including many peer-reviewed papers, books (e.g., Refs. [3,38]), manufacturers catalogs, and on the web.

Practice of solid-phase extraction Conventional SPE is commonly performed on a sorbent placed in a small column (syringe tube), made usually from PP. A variety of formats are available for the columns that may contain from 10 mg up to 10 g absorbing material (typical amounts 50e500 mg). The sorbent material has the shape of small particles loaded in the column between two permeable disks (frits, e.g., made from filter paper or porous PP). Other formats containing sorbent are also available including cartridges, 96 well plates that are useful in high-throughput applications, pipette tip loaded sorbent used for processing very small sample volumes, etc. For performing the sample preparation using SPE, the solution containing the sample is added into the column containing the sorbent. The sorbent from the column is selected to retain specific compounds of interest. The solution flows through the sorbent bad which is packed such that the liquid can flow freely or using a small pressure differential. The pressure differential can be

Sample preparation for high-performance liquid chromatography analysis

285

Figure 6.6.1 The use of SPE sorbent to retain interferences from a sample.

achieved either applying a small gas pressure on the top of the column, but more frequently using a weak vacuum at the base of the column. The SPE sorbent is usually conditioned, for example, with water and methanol before utilization. There are several procedures to use the SPE columns, but two are basically common. One procedure is based on the retention of the interferences from the sample on the sorbent, and it is schematically shown in Fig. 6.6.1. In this procedure, the sorbent is selected to have higher affinity for the matrix components than for the analytes such as the interferences are retained while the analyte molecules are not. The analytes are eluted (and also rinsed) from the SPE column, and a cleanup of the sample takes place. The second common procedure is based on the retention of the analytes on the solid phase, in a specific solution composition. In this procedure in step A, the sample solution is placed into the SPE column, the interferences are eluted and thoroughly eliminated by rinsing (step B). After these operations, the solid phase material is usually dried using a stream of air or N2, and in step C, an eluting solution is added such that the analytes are not anymore retained on the solid support and are released from the sorbent. This procedure is schematically shown in Fig. 6.6.2.

Figure 6.6.2 The use of SPE sorbent to retain and then elute the analytes on an SPE column.

286

Method Development in Analytical HPLC

By this procedure, cleanup as well as concentration of the analytes can be achieved. For the concentration, the volume of the sample solution should be larger than the volume of the eluting solution. Besides those two basic utilizations of the SPE, various other procedures to use the retention-elution capability of SPE are known. In samples with a complex matrix, for example, it is possible that some matrix components are not retained on the sorbent, but other components are retained even stronger than the analytes. A combination of partial cleanup of the sample in the first stage and a second cleanup by using an eluent which elutes only the analytes but not the remaining matrix on the sorbent is possible. SPE is utilized not only based on column retention and elution. One utilization takes advantage of the possibility to make magnetic sorbent particles. Such particles can be mixed with the sample, separated with a magnet, and either the remaining solution is cleaned if the magnetic particles are retaining the matrix, or the analyte is retained on the magnetic particles which are separated and further eluted with an appropriate eluant to recover the analytes. Because during SPE sample preparation the concentration of the analytes in the processed solution is different from that of the initial sample solution, this change must be known for quantitative purposes. A recovery parameter for the analyte as described by Eq. 6.1.1, and an enrichment factor as described by Eq. 6.1.3 are applicable for SPE.

Materials used as SPE sorbent A wide variety of materials were used as solid sorbents, including inorganic porous materials such as carbon, silica, silica with an organic bonded surface, organic synthetic polymers, organic natural polymers, etc. The functionality of the SPE sorbents can also be very specific. Some sorbents may have a nonpolar character (e.g., silica with bonded C18 groups on its surface, or various synthetic polymers such as polystyrene-divinylbenzene). Such phases are designed to retain hydrophobic compounds from aqueous solutions. Sorbents with some polarity, or even strongly polar are also available. For example, a common weakly polar sorbent is a copolymer styrene-divinylbenzene-vinylpyrrolidone, where the pyrrolidonyl groups add a polar character to the hydrophobic polystyrene-divinylbenzene moiety. Such materials are used for the processing of a wide variety of sample and are commercialized under different names (e.g., HLB from Waters or Strata X from Phenomenex). More polar material can be generated by functionalizing different substrates with polar groups such as diol, amino, etc. Ion exchange sorbent can also be used in various SPE types. Similar to the other types of sorbents, the ion exchange materials can be silica-based, polymeric-based, and other types of ion exchangers. Ion exchange sorbents are classified as cation ex change SPE containing anionic bonded groups such as SO 3 or COO , and anion exþ þ change SPE containing cationic groups such as NH3 and NR3 . Also, amphoteric SPE ion exchangers are known. Ion exchange sorbents can be strong, medium, or weak, depending on their capability to retain compounds with low, medium, or high pKa . Sorbents with chelating capability, affinity, immunoaffinity and aptamer sorbents, molecular imprinted polymers, restricted access materials, and mixed mode sorbents

Sample preparation for high-performance liquid chromatography analysis

287

are also used in SPE. More recently materials such as electrospun nanofibers, specific fabrics, and metal organic frameworks were used as sorbents in SPE [52e54]. For the elution of the retained compounds on SPE sorbents, a variety of solvents are utilized. The two operation modes, “retain interferences” or “retain analyte,” require the choice of a solvent to fit the purpose of the SPE operation. The solvent used for elution in SPE is typically different from the one used to retain the analytes, and gradient elution is not common. The solvents are selected depending on retention mechanism on the SPE. For example, hydrophobic compounds are retained on nonpolar SPE (or weakly polar ones) from water and are eluted with more hydrophobic solvents such as benzene, toluene, ethanol, methanol, acetonitrile, ethyl acetate, etc. From anion exchange SPE, the retained analytes such as organic acids are retained using diluted solution of strong acids (e.g., HCl). From cation exchange SPE, the basic compounds are eluted with solutions in water or water/alcohol of ammonium hydroxide or other bases. Several similarities exist between HPLC and SPE retention and elution. Similar to HPLC, the SPE is retaining specific compounds (analytes or matrix components) from a solution passing the stationary phase, and similar to HPLC, the retained compounds can be eluted with a specific eluant. However, besides the use of much larger particles of sorbent in SPE compared to HPLC, the retention in SPE takes place in a “yes/no” manner with the intention to have the compounds of interest either strongly retained or not retained at all. The retention and elution processes in HPLC are gradually allowing a much more precise separation. Also, it is common that the SPE devices are used only once. The column or cartridges can be cleaned and regenerated, but it is uncommon to use them a significant number of times. On the other hand, the HPLC columns are used on a large number of samples (depending on application, several hundred injections can be performed on an HPLC column).

Special types of SPE techniques SPE has been utilized in the form of several other techniques with specific particularities. The main such techniques are: (1) SPME, (2) stir-bar sorptive extraction (SBSE), (3) matrix solid-phase dispersion (MSPD), and (4) QuEChERS. 1) Typical SPME is used mainly in connection with GC (or GC-MS) core analysis. The technique is a miniaturization-type SPE that uses a small amount of the solid phase coated on a silica fiber that can be exposed or retracted in a syringe needle. The fiber containing the sorbent in SPME can be used to collect the analytes from the headspace of a volatile sample followed by the introduction in the hot injection port of a GC instrument where the analytes are desorbed and further analyzed. The collection of the analytes can be also made by immersing the coated fiber in a solution containing the analytes, followed by the same desorption procedure. A variety of fibers are commercially available and the technique has numerous applications (e.g., Refs. [55,56]). A variation of SPME is modified for use on line with HPLC. This technique is indicated as in-tube SPME (or IT-SPME) [57]. IT-SPME uses a short capillary column coated inside with a sorbent. The capillary column is initially connected to a syringe for aspirating and removing the sample solution for the analytes’ retention. Following the analyte retention, the capillary is inserted in a flow generated

288

Method Development in Analytical HPLC

by an HPLC pump such that the analytes are carried to an HPLC chromatographic column where they are separated and further measured. The selection of the inside capillary coating and of the mobile phase composition is adjusted according to the analytes that must be retained and later eluted [58]. 2) SBSE uses the partition of the analytes from a solution into a sorbent covering a small glasscoated magnetic stir-bar (also called “twister”). After the retention of the analytes, a desorption step follows. For use with a GC (or GC-MS) instrument, the desorption is typically done using heating in a dedicated desorber on-line with the GC instrument. For use in HPLC, the desorption can be obtained using the extraction of the stir-bar with an adequate solvent. The adsorbing coating of the stir-bar usually consists of common adsorbing materials such as polydimethylsiloxane (PDMS) [59,60]. 3) MSPD is an analytical technique developed mainly for the extraction of analytes from solid or semisolid and viscous samples. The principle of this technique is based on the use of a sorbent similar to those used in SPE that is mixed with the sample and a grinding material for producing the disruption of sample matrix. After extraction, the mixture of sorbent, sample, and grinding material are transferred into an empty SPE column. Using an adequate solvent, the analytes can be eluted from the column [61]. 4) Quick, easy, cheap, effective, rugged, and safe or QuEChERS technique is a technique similar to conventional SPE, but which is following some specific steps [62]. These steps include the extraction of the sample in a solvent, followed by the addition of a salt such as MgSO4 and/or CH3COONa and further extraction. The solids from the liquid phase are then separated by centrifugation and the resulting solution is further subject to a cleanup step by adding, for example, a primary-secondary amine sorbent (PSA) or other sorbents. The resulting solution is then subject to the core analytical analysis. Special kits for use in QuEChERS are commercially available [63].

Key points • • •

Solid phase extraction, SPE, is a sample preparation technique using a solid sorbent that can be used either for the retention from the sample solution of some of the matrix components or of the analytes. The retained analytes can be further eluted from the sorbent. A variety of sorbent materials are commercially available. Modifications of SPE to be suitable for miniaturization or other purposes lead to specific sample preparation techniques such as SPME, SBSE, or QuEChERS.

6.7

Sample derivatization

Derivatization in sample preparation is the process of changing the chemical structure of the analytes and in some cases of the matrix or both analytes and matrix by using specific reagents. A change in the chemical structure can also be applied to macromolecules and can be in the form of chemical degradation when small molecules are generated from the polymer. Derivatization takes advantage of the presence in the analyte molecule of various functionalities such as hydroxyl, thiol, amino, carbonyl, carboxyl, etc. that can react with the derivatization reagents. The subject of derivatization and polymer degradation as sample preparation procedure is covered in many publications

Sample preparation for high-performance liquid chromatography analysis

289

including books (e.g., Refs. [2,9,64,65], chapters in monographs (e.g., Ref. [66]), papers in peer-reviewed journal (e.g., Ref. [67]), and information on the web.

The role of derivatization in sample preparation for HPLC Derivatization is used in HPLC sample preparation for several purposes, including: (1) sample dissolution, (2) improvement of separation, (3) improvement of detection, (4) improvement of both detection and separation at the same time, (5) improvement in qualitative analysis, (6) improvement of quantitation accuracy, and (7) improvement of stability of the analyte. Except for reagents used in sample dissolution, the reagents used for derivatization must have both a functionality capable to react with the analyte and a moiety carrying a desired characteristic such as a nonpolar moiety, a chromophore group, a fluorophore group, etc. The process of derivatization can be described schematically as follows (Fig. 6.7.1). Derivatization plays an important role in sample preparation for HPLC. However, the use of mass spectrometric detection which can be performed in many cases directly on the analyte, and the tendency to eliminate as much as possible sample preparation that requires manpower, the utilization of derivatization tends to be less common. 1) Sample dissolution necessary for being injected in HPLC cannot be always achieved using physical dissolution. In some cases, chemical modifications of the sample are necessary to place it in a solution. For many samples, a simple use of a solvent with a specific pH obtained by using an acid or a base will be sufficient for sample dissolution. For example, compounds having an acidic character can be easily dissolved in aqueous solution of a base, following a reaction that generates a soluble salt: RH þ NaOH $ RNa þ H2 O

(6.7.1)

A very common procedure used for the pH change of a sample is the addition of buffer solutions. Among these, acetate buffer, borate buffer, citrate buffer, glycine buffer, phosphate buffer, pyrophosphate buffer, 2-amino-2-methyl-1-propanol buffer, and tris(hydroxymethyl)-aminomethane buffer are common. It is also possible to use a combination of neutralization and buffer addition for achieving a desired solution pH. A number of other chemical modification procedures can be done on samples to obtain a better solubility. For example, compounds having active hydrogens can be derivatized (methylated, silylated, etc.) such that they become less polar and soluble in specific organic solvents. Special procedures are sometimes utilized for polymer solubilization.

Figure 6.7.1 Schematic reaction of an analyte with a derivatization reagent.

290

Method Development in Analytical HPLC

2) Derivatization can be used in some instances for the improvement of separation. For example, highly polar compounds containing groups such as OH or COOH are sometimes difficult to separate using RP-HPLC which is a very common and efficient type of separation for less polar compounds. By derivatization, the polar groups can be blocked and the derivatized compounds become less polar and amenable to be separated on C8 or C18 type columns. Derivatization can also be used to avoid matrix effects due to coeluting components. A special role of derivatization for improving separation is performed for the separation of enantiomers. Separation of enantiomers is not possible on nonchiral stationary phases (and nonchiral mobile phase). However, diastereoisomers can be separated on nonchiral phases. Since the use of chiral chromatographic columns requires various restrictions regarding the nature of mobile phase, derivatization can be applied to change the enantiomers into diastereoisomers that can be separated on common columns. One such example is the derivatization of a-substituted organic acids, with a specific enantiomer of an amino acid ester. The reaction is shown in Fig. 6.7.2 for derivatization with L-alanine ethyl ester ((S)- alanine ethyl ester). The reaction shown in Fig. 6.7.2 is catalyzed by the presence of benzotriazol-1-yl-oxytris(dimethylamino)-phosphonium hexafluorophosphate (BOP) as a peptide coupling reagent. As a result of the reaction, from the enantiomers (S and R), two diastereoisomers (S,S) and (S,R) are formed, and the separation is possible on a nonchiral HPLC column. 3) Improvement of detection sensitivity is one of the most common reasons for the use of analytes derivatization. Many analytes do not have chromophores or fluorophores in their molecule and as a result the utilization of common UV or of fluorescence detection is not possible. Derivatization with reagents containing strong chromophore groups or fluorophores has the result of generating a compound that can be easily detected, and in many cases with high sensitivity. Dedicated literature (e.g., Ref. [2]) describes many reagents that are used for derivatization to generate compounds with high UV absorbance, high fluorescence, or even chemiluminescence. Derivatization for improving MS detection is also used for improving detectability of certain analytes. In such cases, the derivatization enhances the ionization efficiencies of

Figure 6.7.2 Reaction of the enantiomers of an organic acid with L-alanine ethyl ester (which is (S)) in the presence of BOP.

Sample preparation for high-performance liquid chromatography analysis

291

analytes leading to lower detection limits. For electrospray ionization MS, the introduction of easily ionizable moieties can effectively increase the sensitivity of detection of target analytes. The introduction of moieties with proton affinity or electron affinity enhances the analyte signals in positive and negative atmospheric pressure chemical ionization MS, respectively [68,69]. 4) In most cases, the modification of a molecular structure for achieving better detection also leads to the necessity to change the separation conditions and usually the derivatization is selected to improve both. One such example is in the analysis of glyphosate and its metabolite aminomethylphosphoric acid (AMPA) after derivatization with FMOC (e.g., Ref. [70]). The reaction of glyphosate with FMOC is shown in Fig. 6.7.3. Following this derivatization, glyphosate and AMPA can be analyzed on a C18 column while underivatizated compounds must be separated on an ion exchange type column that has lower resolution, and in addition can be analyzed with much better sensitivity using UV absorption, fluorescence detection as well as MS/MS detection. 5) The improvement in the results of qualitative analysis is also possible after derivatization. One example is the identification of the position of the C¼C double bonds in unsaturated fatty acids [71]. Another example of a derivatization applied for structural determination is the use of p-bromophenyl isothiocyanate for the derivatization and analysis of peptides primary structure (sequence of amino acid). In this reaction, a peptide (generated, for example, from the hydrolysis of a protein with trypsin) reacts with p-bromophenyl isothiocyanate in basic medium followed by hydrolysis in acidic conditions and heat of the terminal bromophenylthiohydantoin that can be analyzed by HPLC using UV detection [72,73]. This reaction can be successfully used in identifying the sequence of amino acids from peptides by detecting the liberated N-terminus phenylthiohydantoin product. The use of HPLC with MS and circular dichroism (CD) detection allows simultaneous determination of the sequence and absolute configuration of peptide amino acids [74]. The reaction is shown in Fig. 6.7.4. 6) The improvement of quantitation accuracy by derivatization can be the result of increasing sensitivity of analysis, but also with the help of a procedure consisting in derivatizing the sample with a specific reagent and derivatizing the standards with the same reagent but in the isotopically labeled form. For example, a mixture of amino acids from a sample can be derivatized with 2-(4-methylpiperazine)acetic acid N-hydroxysuccinimide ester (NHS). A mixture of amino acids to be used as internal standards can be derivatized with the same reagent having the methylpiperazine group 13C6, 15N2 labeled. The samples with added internal standard can be analyzed by LC-MS/MS, the internal standards showing a difference of 8 amu compared to the analytes [75]. By using the labeled derivatized internal standard, the analysis can be performed more accurately.

Figure 6.7.3 Reaction of glyphosate with FMOC.

292

Method Development in Analytical HPLC

Figure 6.7.4 Reaction of a peptide with p-bromophenyl isothiocyanate followed by hydrolysis utilized for analysis of peptides. 7) Improvement of analytes stability may be necessary in some cases when, for example, the analytes can be easily oxidized. Examples of such analytes are compounds containing SH groups and b-lactam antibiotics. Adding to the sample specific reagents the oxidation process can be avoided and the derivatized compounds can be analyzed without taking special precaution to avoid oxidation [76].

Procedures for performing derivatization for liquid chromatography Derivatization in HPLC can be performed either before the chromatographic analysis (precolumn), or after the column separation and before detection (postcolumn derivatization). The precolumn derivatization can be performed off-line following a specific protocol, but some commercially available autosamplers (e.g., Ref. [77]) have the capability to perform operations such as reagent mixing, heating or cooling, and specified waiting times necessary for a reaction to take place. Such autosamplers can be used for derivatization followed by sample injection. Postcolumn derivatization is usually performed on line and requires some adjustment of the HPLC equipment. This includes the capability of adding the derivatizing reagents, and because the derivatization reaction may require some time to complete, delays coils may be necessary. These coils may also have the capability of heating in case the reaction needs to be accelerated by heat [78]. A problem with postcolumn derivatization by this procedure is the potential peak broadening due to the addition of reagents, and fluctuation in the analyte signal. Another reported possibility to perform a postcolumn derivatization is in source postcolumn derivatization, which is suitable for MS detection [79].

Sample preparation for high-performance liquid chromatography analysis

293

Types of chemical reactions used in derivatization A variety of chemical reactions can be used for derivatization. Among these are the reactions with the formation of alkyl or aryl derivatives. Some of these reactions such as alcohol-acid esterification reactions are facilitated with specific reagents such as dicyclohexylcarbodiimide (DCCI), N,N0 -carbodiimidazole (CDI), etc. Another type of reactions is the formation of silyl derivatives. This type of reaction is very useful in GC and GC-MS analysis, but the propensity of silyl derivatives to hydrolyze diminishes the importance of this type of derivatization in HPLC analysis [80]. Acylation is another common derivatization type of reaction, frequently utilized in the acylation of amines and of compounds containing amino groups such as amino acids. Special compounds are sometimes utilized for facilitating the formation of peptide bonds. Among these can be indicated benzotriazol-1-yl-oxy-tris(dimethyl-amino)phosphonium hexafluorophosphate (BOP), 1-(3-dimethylaminopropyl)-3ethylcarbodiimide (EDAC), diethylcyanophosphonate, etc. Another useful type of derivatization that can be considered acylation is the one using chloroformates (oxycarbonyl chlorides). Chloroformates are used, for example, for the amine derivatization with the formation of urethanes. The radical of the chloroformate group may contain chromophores or fluorescent moieties and the derivatized compound can be used in HPLC leading to a sensitive detection. Among the chloroformates commonly used with a fluorescent moiety is fluorenylmethyloxycarbonyl chloride (FMOC). Sulfonyl derivatives are also used for derivatization, reacting with alcohols, phenols, amines, etc. Such reagents can also contain chromophores or fluorophores. Other types of compounds can be used as derivatization reagents, such as those involving addition to hetero multiple bonds present in the functional groups C¼O, C¼S, C¼N, or ChN. These reactions are used in two manners. One is the derivatization of analytes with functional groups containing active hydrogens such as OH, NH2, SH, etc., using reagents containing hetero multiple bonds. The other is the derivatization of analytes with hetero multiple bonds such as aldehydes or ketones using reagents that are able of causing addition reactions. For example, an amine as an analyte may be derivatized with a ketone reagent, or a ketone as an analyte may be derivatized with a primary amine as a reagent. In addition to small molecules used as derivatization reagents, derivatization can also be achieved using reactions on solid support [81]. Dedicated literature covers many other types of derivatization reactions, some with general applicability and some used only for very specific applications (e.g., Ref. [9]).

Key points • • •

Derivatization is a sample preparation technique that uses a reagent to modify the chemical structure of the analyte (and in some applications of the matrix), with the goal of obtaining better results of the analysis. For HPLC, derivatization can be achieved before the chromatographic analysis (precolumn), or can be performed after the separation (postcolumn). A large variety of chemical reactions and reagents can be used for derivatization.

294

6.8

Method Development in Analytical HPLC

Automated sample preparation and on-line coupling with the HPLC

Sample preparation can be a time-consuming process when performed manually. For this reason, effort has been made to develop automated sample processing. As a result, a variety of automated systems are commercially available for performing different types of operations such as dissolution, extraction, sample cleanup systems using SPE, solution concentration, solvent evaporation, etc. Also, dedicated instruments capable of performing multiple operations are available. For example, all stages of the automated SPE process can be performed with an automated sample preparation instrument (e.g., RapidTrace from Biotage [82]). More difficult to perform automatically are LLE operations, although even for this type of sample processing, automated systems are available [83]. Some of such systems can also be used for the optimization of the sample preparation after setting specific parameters for the method (e.g., SPE cartridges to be tested, solvents to be used, necessary buffers, etc.) [23]. Following an automated sample preparation step, the solution of the processed sample can be loaded in a sample vial and submitted for HPLC analysis. Although the sample preparation performed automatically can improve significantly the productivity, the connection between sample preparation and the HPLC analysis still remains an “off-line” activity. An intermediate stage between the use “off-line” of the samples generated by an automatic sample preparation instrument and a truly “on-line” sample preparationHPLC system is offered by robots that perform the transfer of the vials filled with an automatically processed sample to an HPLC autosampler. Such operations can be performed using computers with dedicated software that controls the sample preparation and the sample transfer (e.g., Ref. [84]). Systems performing a truly “on-line” sample preparation and HPLC analysis have been also developed. As an example, a system using an SPE sample cleanup cartridge on line with an HPPC separation is schematically indicated in Fig. 6.8.1.

Figure 6.8.1 Schematic diagram of an on-line system using an SPE cartridge for sample cleanup/concentration and an HPLC core analysis.

Sample preparation for high-performance liquid chromatography analysis

295

A system such as the one schematically indicated in Fig. 6.8.1 works in two stages. In stage A, the sample is first loaded in an SPE cartridge. The sample may be concentrated in the cartridge, and can be subject to cleaning. In stage B, the flow from an HPLC pump is directed through the SPE cartridge, eluting the analytes and further separating them using an analytical HPLC column. Various challenges are posed by such “on-line” systems, including the need for the SPE cartridge to stand the backpressure created by the HPLC column, requiring for this a special housing, and the adjustments of flow volumes. However, such systems can be very useful in specific applications (e.g., Ref. [85]).

Key points • •

Sample preparation can be performed manually or can be automated to save time and manpower. Sample preparation can be used off-line related to the core HPLC analysis, on-line with the HPLC, or can be performed off-line, but samples can be transferred to the HPLC instrument using a robot.

References [1] S.C. Moldoveanu, Solutions and challenges in sample preparation for chromatography, J. Chromatogr. Sci. 42 (2004) 1e14. [2] S.C. Moldoveanu, V. David, Sample Preparation in Chromatography, Elsevier, Amsterdam, 2002. [3] S.C. Moldoveanu, V. David, Modern Sample Preparation for Chromatography, second ed., Elsevier, Amsterdam, 2021. [4] J. Pawliszyn (Ed.), Comprehensive Sampling and Sample Preparation, vol. 1, Elsevier, Amsterdam, 2012. [5] R.E. Majors, Sample Preparation Fundamentals for Chromatography, Agilent Technologies, Wilmington, 2014. [6] S.C. Moldoveanu, W.A. Scott, D.M. Lawson, Nicotine analysis in several non-tobacco plant materials Contrib, Tob. Nicotine Res. 27 (2016) 54e59. [7] Y. Yang, J. Zhang, B. Shao, Quantitative analysis of fourteen synthetic dyes in jelly and gummy candy by ultra performance liquid chromatography, Anal. Methods 6 (2014) 5872e5878. [8] S.C. Moldoveanu, Comparison of several HPLC methods for the analysis of vitamin C, Biomed. Chromatogr. 38 (2024) e5753. [9] G. Lunn, L.C. Hellwig, Handbook of Derivatization Reactions for HPLC, J. Wiley, New York, 1998. [10] T. Butt, Reproducible sample preparation for reliable food analysis, in: M.T. Stauffer (Ed.), Ideas and Applications toward Sample Preparation for Food and Beverage Analysis, IntechOpen, Rijeka, 2017. [11] K. Liu, Some factors affecting sieving performance and efficiency, Powder Technol. 193 (2009) 208e213.

296

Method Development in Analytical HPLC

[12] D.A. Skoog, D.M. West, F.J. Holler, Fundamentals of Analytical Chemistry, seventh ed., Sounders College Pub., Fort Worth, 1996. [13] D.J. Scott, S.E. Harding, A.J. Rowe (Eds.), Analytical Ultracentrifugation: Techniques and Methods, Royal Soc. Chem. Publ, Cambridge, 2006. [14] D.R. Linde (Ed.), Handbook of Chemistry and Physics, eighty third ed, CRC Press, Boca Raton, 2003. [15] D. MacKay, W.Y. Shiu, K.C. Ma, Henry’s law constant, in: R.S. Boethling, D. MacKay (Eds.), Handbook of Property Estimation Methods for Chemicals: Environmental and Health Sciences, CRC Press, Boca Raton, FL, 2000, p. 69. [16] J.H. Hildebrand, R.I. Scott, The Solubility of Non-electrolytes, Dover, New York, 1964. [17] W.J. Lyman, W.F. Reehl, D.H. Rosenblatt, Handbook of Chemical Property Estimation Methods, ACS, Washington, 1990. [18] http://www.epa.gov/oppt/exposure/pubs/episuite.htm. [19] http://www.chemaxon.com. [20] A. Dokoumetzidis, P. Macheras, A century of dissolution research: from Noyes and Whitney to the biopharmaceutics classification system, Int. J. Pharm. 321 (2006) 1e11. [21] D.H. Johnson, W.W. Wilson, L.J. DeLucas, Protein solubilization: a novel approach, J. Chromatogr. B 971 (2014) 99e106. [22] R.F. Strickland-Constable, Kinetics and Mechanism of Crystallization, Academic Press, New York, 1969. [23] M.H. Sarafian, M. Gaudin, M.R. Lewis, F.-P. Martin, E. Holmes, J.K. Nicholson, M.-E. Dumas, Objective set of criteria for optimization of sample preparation procedures for ultra-high throughput untargeted blood plasma lipid profiling by ultra performance liquid chromatographymass spectrometry, Anal. Chem. 86 (2014) 5766e5774. [24] S.C. Moldoveanu, V. David, Dependence of the distribution constant in liquideliquid partition equilibria on the van der Waals molecular surface area, J. Separ. Sci. 36 (2013) 2963e2978. [25] L. Rohrschneider, Solvent characterization by gas-liquid partition coefficients of selected solutes, Anal. Chem. 45 (1973) 1241e1247. [26] L.R. Snyder, Classification of the solvent properties of common liquids, J. Chromatogr. 92 (1974) 223e230. [27] L.R. Snyder, Classification of the solvent properties of common liquids, J. Chromatogr. Sci. 16 (1978) 223e234. [28] R.W. Taft, M.J. Kamlet, The solvatochromic comparison method 2. The alpha scale of solvent hydrogen-bond donor (HBD) acidities, J. Am. Chem. Soc. 98 (1976) 2886e2894. [29] B.P. Johnson, M.G. Khaledi, J.G. Dorsey, Solvatochromic solvent polarity measurements and retention in reversed-phase liquid chromatography, Anal. Chem. 58 (1986) 2354e2365. [30] M.H. Abraham, A. Ibrahim, A.M. Zissimos, Determination of sets of solute descriptors from chromatographic measurements, J. Chromatogr., A 1037 (2004) 29e47. [31] C.F. Poole, Solvent selection for liquid-phase extraction, in: C.F. Poole (Ed.), LiquidPhase Extraction, Elsevier, Amsterdam, 2019. [32] S. Moldoveanu, A. Savin, Aplicatii in Chimie ale Metodelor Semiempirice de Orbitali Moleculari, Editura Academiei RSR, Bucharest, 1980. [33] J. Dean, Extraction Methods for Environmental Analysis, J. Wiley, New York, 1998. [34] A. Handley (Ed.), Extraction Methods in Organic Analysis, Sheffield Academic Press, Sheffield, 1999. [35] J.M. Kokosa, A. Przyjazny, M.A. Jeannot, Solvent Microextraction: Theory and Practice, J. Wiley, Hoboken, 2009.

Sample preparation for high-performance liquid chromatography analysis

297

[36] A. Sarafraz-Yazdi, A. Amiri, Liquid-phase microextraction, TrAC, Trends Anal. Chem. 29 (2010) 1e14. [37] J.M. Kokosa, Solvent microextraction, in: J. Pawliszyn (Ed.), Comprehensive Sampling and Sample Preparation, vol. 2, Elsevier, Amsterdam, 2012. [38] C.F. Poole (Ed.), Liquid-Phase Extraction, Elsevier, Amsterdam, 2020. [39] K.D. Clark, C. Zhang, J.L. Anderson, Sample preparation for bioanalytical and pharmaceutical analysis, Anal. Chem. 88 (2016) 11262e11270. [40] C.G. Huber, H. Oberacher, Analysis of nucleic acids by on-line liquid chromatographymass spectrometry, Mass Spectrom. Rev. 20 (2001) 310e343. [41] S. Nyiredy, Solid-liquid extraction strategy on the basis of solvent characterization, Chromatographia 51 (2000) S288eS296. [42] B.E. Richter, B.A. Jones, J.L. Ezzell, N.L. Porter, N. Avdalovic, C. Pohl, Accelerated solvent extraction: a technique for sample preparation, Anal. Chem. 68 (1996) 1033e1039. [43] C.C. Teo, S.N. Tan, J.W.H. Yong, C.S. Hew, E.S. Ong, Pressurized hot water extraction (PHWE), J. Chromatogr. A 1217 (2010) 2484e2494. [44] M. Plaza, M.L. Marina, Pressurized hot water extraction of bioactives, TrAC, Trends Anal. Chem. 166 (2023) 117201. [45] V. Lopez-Avila, R. Young, W.F. Beckert, Microwave-assisted extraction of organic compounds from standard reference soils and sediments, Anal. Chem. 66 (1994) 1097e1106. [46] I. Fernandez-Pastor, A. Fernandez-Hernandez, S. Perez-Criado, F. Rivas, A. Martinez, A. Garcia-Granados, A. Parra, Microwave-assisted extraction versus Soxhlet extraction to determine triterpene acids in olive skins, J. Separ. Sci. 40 (2017) 1209e1217. [47] C.S. Eskilsson, E. Björklund, Analytical-scale microwave-assisted extraction, J. Chromatogr. A 902 (2000) 227e250. [48] F. Chemat, N. Rombaut, A.-G. Sicaire, A. Meullemiestre, A.-S. Fabiano-Tixier, M. AbertVian, Ultrasound assisted extraction of food and natural products. Mechanisms, techniques, combinations, protocols and applications. A review, Ultrason. Sonochem. 34 (2017) 540e560. [49] R. Smith, H. Inomata, C. Peters, Introduction to Supercritical Fluids: A Spreadsheet-Based Approach, Elsevier, Amsterdam, 2013. [50] Z. Huang, X. Shi, W. Jiang, Theoretical models for supercritical fluid extraction, J. Chromatogr. A 1250 (2012) 2e26. [51] L.T. Taylor, Supercritical Fluid Extraction, Wiley, Hoboken, 1996. [52] J. Plotka-Wasylka, M. Marc, N. Szczepanska, J. Namiesnik, New polymeric materials for solid phase extraction, Crit. Rev. Anal. Chem. 47 (2017) 373e383. [53] N. Fontanals, R.M. Marce, F. Borrull, Materials for solid-phase extraction of organic compounds, Separations 6 (2019) 56. [54] S. Chigome, N. Torto, Electrospun nanofiber-based solid-phase extraction, TrAC, Trends Anal. Chem. 38 (2012) 21e31. [55] C.L. Arthur, J. Pawliszyn, Solid-phase microextraction with thermal desorption using fused silica optical fibers, Anal. Chem. 62 (1990) 2145e2148. [56] J. Pawliszyn, Solid-phase Microextraction, Theory and Practice, Wiley-VCH, New York, 1997. [57] Y. Moliner-Martinez, A. Ballester-Caudet, J. Verd u-Andrés, R. Herraez-Hernandez, C. Molins-Legua, P. Campins-Falco, In-tube solid-phase microextraction, in: C.F. Poole (Ed.), Solid-Phase Extraction, Elsevier, Amsterdam, 2020, pp. 387e427.

298

Method Development in Analytical HPLC

[58] S. Fang, Y. Huang, Q. Ruan, C. Chen, S. Liu, G. Ouyang, Recent developments on solid phase microextraction (SPME) coatings for in vivo analysis, Green Anal. Chem. 6 (2023) 100069. [59] E. Baltussen, P. Sandra, F. David, C. Crammers, Stir-bar sorptive extraction (SBSE), a novel extraction technique for aqueous samples: theory and principles, J. Microcolumn Sep. 11 (1999) 737e747. [60] Z. Wang, M. He, B. Chen, B. Hu, Azo-linked porous organic polymers/polydimethylsiloxane coated stir bar for extraction of benzotriazole ultraviolet absorbers from environmental water and soil samples followed by high performance liquid chromatography-diode array detection, J. Chromatogr. A 1616 (2020) 460793. [61] A.L. Capriotti, C. Cavaliere, P. Foglia, R. Samperi, S. Stampachiacchiere, S. Ventura, A. Lagana, Recent advances and developments in matrix solid-phase dispersion, TrAC, Trends Anal. Chem. 71 (2015) 186e193. [62] M. Anastassiades, S.J. Lehotay, D. Stajnbaher, F.J. Schenck, Fast and easy multiresidue method employing acetonitrile extraction/partitioning and “dispersive solid-phase extraction” for the determination of pesticide residues in produce, J. AOAC Int. 86 (2003) 412. [63] S.J. Lehotay, K.A. Son, H. Kwon, U. Koesukwiwat, W. Fu, K. Mastovska, E. Hoh, N. Leepipatpiboon, Comparison of QuEChERS sample preparation methods for the analysis of pesticide residues in fruits and vegetables, J. Chromatogr. A 1217 (2010) 2548e2560. [64] D.R. Knapp, Handbook of Analytical Derivatization Reactions, J. Wiley, New York, 1979. [65] T. Toyo’oka (Ed.), Modern Derivatization Methods for Separation Sciences, J. Wiley, Chichester, 1999. [66] S.C. Moldoveanu, V. David, Derivatization methods in GC and GC/MS, in: P. Kush (Ed.), Gas Chromatography, Derivatization, Sample Preparation, Applications, IntechOpen, London, 2018. [67] V. David, S.C. Moldoveanu, T. Galaon, Derivatization procedures and their analytical performances for HPLC determination in bioanalysis, Biomed. Chromatogr. 35 (2021) e5008. [68] T. Higashi, K. Shimada, Derivatization of neutral steroids to enhance their detection characteristics in liquid chromatography-mass spectrometry, Anal. Bioanal. Chem. 378 (2004) 875e882. [69] T. Santa, O.Y. Al-Dirbashi, T. Fukushima, Derivatization reagents in liquid chromatography/electrospray ionization tandem mass spectrometry for biomedical analysis, Drug Discov. Ther. 1 (2007) 108e118. [70] J. Garba, A.W. Samsuri, R. Othman, M.S.A. Hamdani, Simplified method for derivatization of extractable glyphosate and aminomethylphosphonic acid and their determination by high performance liquid chromatography, Environ. Sci. Technol. 1 (2018) 19e30. [71] T.-Y. Zhang, S. Li, Q.-F. Zhu, Q. Wang, D. Hussain, Y.-Q. Feng, Derivatization for liquid chromatography-electrospray ionization-mass spectrometry analysis of small-molecular weight compounds, TrAC, Trends Anal. Chem. 119 (2019) 115608. [72] P. Edman, Preparation of phenylthiohydantoins from natural amino acids, Acta Chem. Scand. 4 (1950) 277e282. [73] P. Edman, G. Begg, A protein seqenator, Eur. J. Biochem. 1 (1967) 80e91. [74] D. Hahn, W. Wang, H. Choi, H. Kang, Determination of sequence and absolute configuration of peptide amino acids by HPLC-MS/CD-based detection of liberated N-terminus phenylthiol-hydantoin amino acids, Sci. Rep. 12 (2022) 10285.

Sample preparation for high-performance liquid chromatography analysis

299

[75] P.K. Held, L. White, M. Pasquali, Quantitative urine amino acid analysis using liquid chromatography tandem mass spectrometry and a TRAQ reagents, J. Chromatogr. B 879 (2011) 2695e2703. [76] M.E. de Jonge, S.M. van Dam, M.J. Hillebrand, H. Rosing, A.D. Huitema, S. Rodenhuis, J.H. Beijnen, Simultaneous quantification of cyclophosphamide, 4-hydroxycyclophosphamide, N,N’,N"-triethylenethiophosphoramide (thiotepa) and N,N’,N"-triethylenephosphoramide (tepa) in human plasma by high-performance liquid chromatography coupled with electrospray ionization tandem mass spectrometry, J. Mass Spectrom. 39 (2004) 262e271. [77] http://www.chem.agilent.com/Library/usermanuals/Public/G1329-90012_StandPrepSamplers_ ebook.pdf. [78] K. Zacharis, P.D. Tzanavaras, Liquid chromatography coupled to on-line post column derivatization for the analysis of organic compounds: a review on instrumentation and chemistries, Anal. Chim. Acta 798 (2013) 1e24. [79] A.M. Cirigliano, G.M. Cabrera, Post-column in-source derivatisation in LC-MS: a tool for natural products characterization and metabolomics, Phytochem. Anal. 31 (2020) 606e615. [80] A.E. Pierce, Silylation of Organic Compounds, Pierce Chem. Co., Rockford, 1982. [81] S.N. Atapattu, J.M. Rosenfeld, Solid phase analytical derivatization, in: J. Reedijk (Ed.), Reference Module in Chemistry, Molecular Sciences and Chemical Engineering, Elsevier, 2018. [82] https://www.richmondscientific.com/product/biotage-rapid-trace-spe-workstation-2. [83] F. Foster, O. Cabrices, J. Whitecavage, J. Stuff, E. Pfannkoch, Automating liquid-liquid extractions using a bench-top workstation, Gerstel AppNote 177. [84] J. Pan, C. Zhang, Z. Zhang, G. Li, Review of online coupling of sample preparation techniques with liquid chromatography, Anal. Chim. Acta 815 (2014) 1e15. [85] Y. Liang, T. Zhou, Recent advances of online coupling of sample preparation techniques with ultra high performance liquid chromatography and supercritical fluid chromatography, J. Separ. Sci. 42 (2019) 226e242.

Chromatographic columns in high-performance liquid chromatography 7.1

7

Construction of an HPLC column

The chromatographic column is a key component of the high-performance liquid chromatography (HPLC) instrumentation. The columns are usually made from a tube (typically made from metal) filled with the stationary phase (see Section 3.2). The stationary phases are usually made from a porous material in the form of relatively uniform particles on which an active phase is bonded, grafted, or coated [1]. A common solid support used for HPLC columns is hydrated silica. On this support, various active phases are attached, such as moieties of long hydrocarbon chains (C8 or C18), or shorter hydrocarbon chains connected to polar groups (e.g., OH, NH2). Other types of groups such as CN, C6F5, CONH2, N(CH3)þ 3 , SO3H, etc., can be attached to the porous support, usually through a short hydrocarbon handle, offering different properties to the active stationary phase. Some columns have the active phase bonded on a porous organic polymer. Besides particles filling the chromatographic columns, monolithic porous rods can be used as stationary phase, but particles are more common. Other physical forms of separation media are also known in HPLC, such as cartridges and microfluidic chips, but their use is far less common than that of columns. Several characteristics of columns and their stationary phases (the solid support and the active phase) are further discussed.

Column physical construction Physical dimensions of the chromatographic column including length L (e.g., 50, 100, 150, 250 mm) and inner diameter d (e.g., 2.1, 3.0, 4.6 mm) play an important role in its properties. Several parameters regarding to peak characteristics and separation in HPLC are influenced by these parameters. The retention time tR is a main such parameter that depends on L and d. This is shown by Eq. 4.1.22 that can be written as follows: i ε pd 2 L h 0 ½1 þ KðXÞJ tR ðXÞ ¼ t0 1 þ k ðXÞ ¼ 4U

(7.1.1)

As indicated in Eq. 7.1.1, the retention time is proportional with column length L and square of the inner column diameter d, but the dependence is more complicated. A bigger column contains a larger amount of stationary phase Wpac and as a result (see Eq. 4.1.23) the volume of stationary phase Vst and therefore the phase ratio J are affected. The longest retention time tR determines the run time trun for a separation and therefore the productivity of the HPLC application. Method Development in Analytical HPLC. https://doi.org/10.1016/B978-0-443-29849-3.00002-6 Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

302

Method Development in Analytical HPLC

Resolution of a column is also dependent on column dimensions (among other factors). The number of theoretical plates of the column N is directly proportional with the column length L (see Eq. 4.1.56) and as a result the column resolution Rs (see Eq. pffiffiffi 4.2.9) is proportional with L . The dependence of Rs on column dimensions is however more complicated since retention factor k' is also affected by column dimensions through the value of J. Another parameter depending on column physical dimensions is the optimum flow rate in the column. As indicated by Eq. 4.1.52, in order to obtain a maximum number of theoretical plates for a column, an optimum flow rate Uopt should be used. The value of Uopt is directly proportional with d 2 of the column, and wider columns afford a larger Uopt . Eq. 4.1.52 also shows that for maintaining the same column performance (the same linear flow rate u) when changing a column with diameter d1 with a column with diameter d2 , the following expression must be utilized: U2 ¼

d22 U1 d12

(7.1.2)

According to Eqs. 4.5.16 and 4.5.17, the maximum volume Vinj and the maximum amount qinj of sample that can be injection in a column are also determined by the values of L and d. As expected, larger columns accommodate larger Vinj and qinj . One more parameter where L and d of the column are important is the backpressure created by the packed material in column. Eq. 4.1.57 can be written in the form (Darcy equation): Dp ¼

4hUfr L 2

ε pd 2 d p

(7.1.3)

As shown by Eq. 7.1.3, the backpressure of a column is directly proportional with its length L and inversely proportional with d2 .

Physical characteristics of the solid support of stationary phase The role of solid support of particles used in HPLC is to offer a highly porous material with specific characteristics on which the active phase resides. The main physical characteristics of the particles making the solid support include: (1) type of solid particles (porous, core-shell, pellicular), (2) shape of the particles, (3) the dimension of particles, (4) the size distribution of the particles, (5) surface area of the particles, (6) pore size of the particles, (7) mechanical rigidity of the support, and (8) other solid support characteristics. 1) The particles in common HPLC columns can be porous, core-shell, or pellicular. Porous particles have a porous structure for the entire particle, and core-shell particles have a solid nonporous core (1.7e3 mm diameter) surrounded by a porous outer shell 0.3e0.5 mm in depth. Pellicular particles which are not commonly used are solid nonporous spheres covered

Chromatographic columns in high-performance liquid chromatography

2)

3)

4)

5)

6)

7)

303

with a thin layer of stationary phase. The core-shell particles offer specific advantages compared to fully porous particles because they generate less eddy diffusion (lower HE contribution to theoretical plate height) [2]. For typical packed columns, the efficiency per meter N for the column can be between 20,000 and 150,000, and for core-shell columns, it can be as high as 300,000. On the other hand, the core-shell particles have less active stationary phase since the actual Wpac for a core shell column is lower compared to an identical column loaded with fully porous particles. Based on Eq. 4.1.23, the result is a lower volume of stationary phase Vst and therefore a lower phase ratio J. This affects in a complex manner the retention, which is typically lower for a core-shell type column compared to the fully porous column. Having less active phase, a column with coreshell particles accommodates smaller Vinj and qinj compared to a column with fully porous particles with the same Wpac . The shape of particles can be irregular or spherical. Effort has been involved in generating particles as close as possible to spherical form because they offer a more homogeneous stationary phase. An irregular particle shape leads to the decrease in the theoretical plate N for the column. A more irregular particle shape leads to a larger parameter L in Eq. 4.1.55, and therefore to a lower N. The dimension of particles (diameter) dp has typical values of 5 mm, 3 mm, 2.1 mm, 1.8 mm, and 1.7 mm. The particle diameter is affecting theoretical plate N with lower dp leading to higher N. This is directly expressed by Eq. 4.1.55 showing that N has a value inversely proportional with dp . As a result, columns with smaller particles have higher resolution Rs. The decrease in particle diameter dp increases however the backpressure of the column (see Eq. 4.1.57). The size distribution of the particles is also an important physical property since the uniformity of the dimensions of the particles leads to smaller differences in the paths of molecules for the same species in the chromatographic column. More uniform particles lead to less peak broadening. Surface area of the particles is an important parameter since the active stationary phase is distributed on the surface of the solid support and a larger surface area is related to a larger amount of active stationary phase for a given phase volume. For silica particles, for example, surface area varies between 100 m2/g for low surface area particles and 300 m2/g for high surface area particles. A larger surface area increases the load of active stationary phase. For a reversed-phase HPLC (RP-HPLC) column, for example, a higher %C can be obtained, and therefore a higher phase ratio J is obtained, which affects retention time tR and resolution Rs. The pore size (diameter) of the porous materials is also an important characteristic of the solid support. There are several classifications of pore size, but usually indicated as small (below 60 Å), medium (in the range 60e150 Å), and large (of about 300 Å or larger). The most common type used in HPLC are medium pore phases used for the separation of nonpolymeric molecules and large for the separation of polymers (such as proteins). The larger pores for the separation of macromolecules are necessary for allowing the contact of compounds from injected sample with the active stationary phase from the pores. Mechanical rigidity of the support refers to the capacity of stationary phase to withstand the high pressure in the chromatographic column without volume modification. Shrinking of the stationary phase under pressure generates void volumes in the column and reduction of N. Inorganic supports such as hydrated silica stand high pressures with values between 9000 and 17,500 psi (600e1200 bar). Specific high strength silica (e.g., HSS developed by Waters) can be used at up to 18,000 psi without deterioration. Rigid organic polymeric particles typically stand pressure up to about 5000 psi (350 bar).

304

Method Development in Analytical HPLC

8) Other solid support characteristics include the proper packing of the material in the body of the chromatographic column, the uniformity of the network of channels, the absence of unintended fine particles, etc.

Different from the stationary phase made of small particles, the monolithic stationary phases are made from a single piece (rod) of a solid porous material. Such rods can be made from similar materials as the particles (e.g., porous silica) and they have specific porosity (mesopores 2e50 nm in diameter, and macropores 4000 to 20,000 nm in diameter). Also, monolithic stationary phase must have specific resistance to pressure, obtained, for example, by a silica skeleton of approximately 1e2 mm thick, having a void volume of almost 80% of the entire column volume. With the hope of further improving column characteristics, various studies were performed regarding the use of other shapes than particles or monoliths for the stationary phase [3,4]. For example, with the use of microlithographic techniques, it was possible to obtain stationary phases with very homogeneous internal channels (such as pillar array) significantly diminishing the eddy dispersion (see Eq. 4.1.48) [5].

Chemical characteristics of the solid support of stationary phase Besides its own chemical structure, the main chemical characteristic of material making the support part of the particles in an HPLC column is its capability to bind the part that represents the active component of the stationary phase. This active component is frequently covalently bound to the solid support; therefore, the reactivity to the binding component is an important chemical characteristic of the support. Among the chemical characteristics of the solid support can be listed: (1) chemical nature of the support, (2) reactivity of the groups used for binding the active phase (e.g., eOH groups in case of silica or other hydrated oxides that have pK a varying between 3.5 and 4.5), (3) chemical resistance of the support to the mobile phase characteristics (e.g., pH), and (4) chemical purity of the solid support. 1) Regarding its chemical nature, the most common support of stationary phase used in HPLC is porous hydrated silica (SiO2 x H2O). This material is obtained from a chemical reaction that generates silicic acids (typically a controlled hydrolysis of an alkoxysilane or of a silicic acid salt) followed by condensations of silicic acid and the formation of silica gel. For the case of hydrolysis of tetraethoxysilane, the process can be described by the chain of reactions indicated in Fig. 7.1.1. In the first stages of this reaction, the alkoxy groups may remain attached to the silicon atom and further participate in the elimination reaction to form siloxane bonds. The resulting polysilicic acid is initially present in solution form (sol form), and a large number of water molecules are bound through hydrogen bonds or mechanically retained in the structure. This sol is further transformed into a gel by losing water molecules. Various conditions are used to obtain the silica gels with the goal of reticulation of the gel [6]. Following the gelling and aging process, the hydrogel (or alcogel) is typically washed and it is dried (converted into a xerogel) for obtaining a solid material. This material contains a large number of reactive silanol groups (hSieOH) which can be further reduced to siloxane groups (SieOeSi) by heating. The resulting hydrated silica remains highly porous

Chromatographic columns in high-performance liquid chromatography

305

Figure 7.1.1 The hydrolysis/condensation reactions of tetraethoxysilane with the formation of hydrated silica.

Figure 7.1.2 The reactions involved in the formation of ethylene-bridged silica. and different classifications are based on their porosity (fine pores, medium pores, large pores). Also, it retains a large number of silanol groups. A relatively newer but becoming common support used for HPLC is ethylene-bridged silica (indicated as BEH technology by Waters or TWIN technology, and Gemini NX or EVO by Phenomenex). This support is prepared in two steps. In the first step, a polyethoxysiloxane oil is prepared from hydrolytic condensation of bis(triethoxysilyl)ethane and tetraethoxysilane using a small amount of water. The resulting material undergoes further condensation under alkaline conditions in an oil-in-water emulsion generating porous hybrid particles. This reaction is schematically indicated in Fig. 7.1.2. Ethylene-bridged silica is in many respects similar to silica, containing a large number of silanol groups but has additional characteristics such as higher chemical stability. Other inorganic materials besides silica and ethylene-bridged silica were evaluated as potential supports for stationary phase. Among these are silica hydride, hydrated zirconium dioxide (hydrated zirconia), hydrated titanium dioxide, hydrated alumina, zeolites, porous graphitic carbon and other carbon-based materials such as carbon nanotubes and fullerenes, etc. Some disadvantages of those materials compared to hydrated silica make their utilization rather limited (e.g., Ref. [7].). For example, zirconia surface may act as a Brönsted base and also as a Lewis acid due to the d electrons of zirconium atoms that can interact with various ligands. Additional interaction may interfere with the main intended type of interaction produced by the active phase, generating modifications in the retention and peak shape. Besides inorganic support materials for the stationary phase in HPLC, certain organic polymers are used for the same purpose. Porous polymers are used as stationary phases mainly for size exclusion chromatography and for ion exchange chromatography (IEC).

306

Method Development in Analytical HPLC

One of the most common polymeric supports used as stationary phase is polystyrene crosslinked with divinylbenzene (PS-DVB), but other polymers are also utilized. For size exclusion, the polymers are generated with a desired pore size, but for uses in other applications such as ion exchange, it is common that polymers are synthesized directly with the desired functional groups attached to the macromolecular frame. 2) The reactivity of the groups used for binding the active phase plays an important role in generating the final stationary phase. These groups are characterized by their nature and acidic/basic character. Most inorganic supports have eOH groups as reactive sites. The density of the active groups versus that of backbone structure (e.g., -O-Si-O- groups) is an important chemical characteristic of the solid support. The proportion of silanol groups on silica surface varies depending on silica preparation, but values in the range 7e8 mmol/m2 are common. Related to the amount of silica, the amount of silanol groups dOH (in mmol/g) is in the range of 3e7 mmol OH groups per g of SiO2. The number of OH groups per unit mass of silica depends linearly on surface area of the material following an expression of the form [8]: aOH ¼ 602:214

dOH Ssurf

(7.1.4)

In Eq. 7.1.4, Ssurf is the surface in m2/g and aOH indicates silanol density or the number of OH groups per unit surface area (nm2) of the stationary phase. The value of aOH was found to vary between 4.1 and 5.6 OH groups per nm2 (with an average of 4.9 OH groups per nm2) for the silica used as support in HPLC. 3) Chemical resistance of the support to the mobile phase is another important characteristic of the solid support. Mobile phases used in HPLC may have different pH values. Stability of the solid support in a wide range of pH values is an important quality. Simple hydrated silica is resistant to acid and bases only in the range of pH between 2 and 7 or 8, and significant effort has been made to obtain supports with stability in a wider pH range. One path to enlarge the pH range of stability is the use of ethylene-bridged silica which is stable at pH between 1 and 11. Other procedures, such as controlled surface charge procedure (charged surface hybrid or CSH technology) which starts with an ethylene-bridged silica and applies a modification of the surface with low level of ionic groups before bonding the active phase, are also used for this purpose [9], extending the resilience to high pH even further than that of ethylene bridge silica. The supports based on organic polymers are resistant to a wider range of pH, but they may have problems with incompatibility with specific solvents and their mechanical rigidity is lower compared to that of inorganic supports. 4) Chemical purity of the solid support is important because the presence of certain impurities such as transitional ions may lead to undesired interactions with the analytes. These interactions will superpose over the intended interactions in the separation process, disturbing it. For example, the presence of Fe3þ and Al3þ in the silica matrix enhances the acidity of silanols and pure silica (indicated as Type B) is highly preferred as support in HPLC. A silica of slightly lower purity than Type B is indicated as Type A. 5) Various organic synthetic polymers such as polystyrene-divinilbenzene, polyacrylates, etc. can also be used as solid support. Such supports are more stable at extreme pH but typically have lower mechanical resilience.

The chemical properties of the monolithic support materials are similar to those of the particles.

Chromatographic columns in high-performance liquid chromatography

307

The active component of the stationary phase The active component of stationary phases is the medium on which the equilibrium Xmo $ Xst takes place as previously described in Section 2.1. This medium is tailored to be adequate for the separation of specific categories of compounds and corresponds to a specific HPLC type. The classification of HPLC types was described in Section 2.3 and each main type of HPLC described in Section 2.3 (e.g., reversed-phase, hydrophilic interaction, ion exchange, size exclusion, chiral) is performed on specific active stationary phase in the columns. An earlier classification of chromatographic columns related to their stationary phase has been proposed by US Pharmacopeia (USP) [10]. This classification uses codes indicated as L1, L2, L3, etc., and is shown in Appendix 1. Further discussion regarding specific types of stationary phases in columns is presented in Sections 7.2e7.7. The active part of stationary phase is present on the surface of the solid support (such as hydrated silica or other materials previously described). For some stationary phases, the active component is already present when the solid support is synthesized (e.g., in the case of some organic polymers), but the most common procedure to generate an active component of the stationary phase is the derivatization of the surface of the solid support. This is achieved by reactions that add to the solid support specific desired groups (ligands) R that modifies the nature of the support surface such that the ligand R can play its key role in the separation. The ligand R changes the nature of the solid support such that to be utilized in a specific type of interactions (hydrophobic, polar, ionic, chiral, etc.). Other procedures to modify the support surface such as grafting or coating the support with presynthesized active phase polymers are less common. Several aspects regarding the bonding of the active component (active phase) on the support include: (1) the type of reagent used for derivatization, (2) the horizontal or vertical type of derivatization, (3) production of unique or more complex stationary phases, (4) the extent of derivatization, (5) end-capping (or endcapping), and (6) other properties of the active part of stationary phase. 1) Typical reagents used for derivatization are chemical compounds containing a reactive functionality and a group that will change the support external layer (e.g., porous silica) into a material with specific properties desired to act as stationary phase. Such properties may include a hydrophobic character (as used in RP-HPLC), a polar character (as used in hydrophilic interaction liquid chromatography (HILIC)), a chiral character, etc. The model of a derivatization reaction attaching R groups to a silica surface is shown in Fig. 7.1.3. The same type of reaction takes place on an ethylene-bridged silica surface.

Figure 7.1.3 A derivatization reaction attaching R groups to a silica surface.

308

Method Development in Analytical HPLC

Many other types of derivatization are possible. One common procedure is the use of reagents that have two or three reactive functionalities having the capability to react with two adjacent (vicinal) silanols, or even with three silanol groups. The use of difunctional and trifunctional derivatization reagents leads to more stable phases with a better coverage of silanol groups. For example, the reaction with a trifunctional derivatization reagent takes place following the scheme shown in Fig. 7.1.4. 2) The horizontal or vertical type of derivatization refers to the number of layers of added substituents on the surface of the solid support. In horizontal derivatization, basically a single layer of substituents is added to the solid support. This type of derivatization is sometimes indicated as oligomeric. In vertical derivatization, the reagent of the type R  SiðCH3 Þ2  X is replaced with a reagent containing hydrolyzable groups such as R  SiðOC2 H5 Þ2  CH3 and during the reaction, some water is added such that the added moiety containing the R groups also contain OH groups that can continue to be derivatized and can grow a thick layer of bonded phase. In vertical derivatization, the support structure is basically covered with a bonded polymeric layer of some thickness that represents the active phase [11,12]. A schematic diagram of a silica surface with a vertical derivatized surface is indicated in Fig. 7.1.5. Polymeric-type bonded phase can also be obtained by bonding a premade polymer on silica. An example is the bonding of polysuccinimide on silica derivatized with amino-propyl groups, as it is illustrated in Fig. 7.1.6 for making a polar-type stationary phase. 3) Derivatization may produce a unique or a more complex type of stationary phase. In some derivatization, a unique type of group R is attached to the solid support, but moieties with different character (e.g., some hydrophobic, and some polar) can also be attached. Also, some derivatization reactions are performed in two stages, generating more complicated covering structures such as containing embedded polar groups in an aliphatic carbon chain. 4) The extent of derivatization indicates how much desired R groups are attached on the solid support. For hydrophobic phases (e.g., C8 or C18), the carbon load (C%) can be used for the characterization of the degree of derivatization. Values of carbon load can vary from about 5% up to 25%. A more precise characterization of derivatization degree is obtained using the number of micromoles of ligands attached to the OH groups. This degree is given by the expression: dligand ¼

106 C% ðCsil  C%ÞMwsil Ssurf

(7.1.5)

In expression 7.1.5, dligand is the amount in mmol of silyl ligands per m2, C% is the carbon load, Csil is the mole % of carbon in the silyl ligand R, Mwsil is the molecular weight of the silyl ligand, and Ssurf is the surface area of the silica. The value of dligand for a ligand such as dimethyloctadecylsilyl is dligand z 5.6 mmol/m2 for C% ¼ 20% and Ssurf ¼ 200 m2/g. The

Figure 7.1.4 A derivatization reaction attaching R groups to a silica surface using a trifunctional type derivatization reagent.

Chromatographic columns in high-performance liquid chromatography

309

Figure 7.1.5 Schematic illustration of a silica surface vertically derivatized with an active phase R.

Figure 7.1.6 Illustration of synthesis of a polar stationary phase using the bonding of a premade polymer (polysuccinimide). amount of silanol groups of 7e8 m mol OH/m2 is larger (in mmol) than that of the ligand amount. In addition, this value of 5.6 mmol/m2 corresponds to a high C% load typical for stationary phases obtained with vertical polymerization. 5) The comparison of ligand R density with that of OH density indicates that even after derivatization, a considerable number of free OH groups (in case of silica gel support) are still present in the stationary phase. Steric hindrance does not allow more addition of large ligands R on the solid support such as silica. The remaining eOH groups with a polar character may interfere in the intended utilization of the stationary phase, such as the separation using a hydrophobic stationary phase. To reduce the number of free silanol groups, an additional derivatization indicated as end-capping (or endcapping) can be performed on a stationary phase. End-capping consists in adding small ligands such as eCH3 to the remaining silanols (other small ligands such as eCH2OH can also be utilized for end-capping). Some stationary phases

310

Method Development in Analytical HPLC

are not subject to this process (and remain not end-capped), but end-capping is rather common. For end-capping, reagents such as trimethylchlorosilane or hexamethyldisilazane can be utilized. By end-capping with eCH3 groups, the carbon load C% of packing material does not significantly change. The end-capping is limiting the types of interactions the stationary phase has with the analytes to the one produced by the ligand R for which the column was designed. For reversed-phase type of columns, for example, end-capping with eCH3 groups is limiting the polar interactions. The advantage of this limitation is that the alternative types of interactions of a chromatographic column may produce peak distortions. End-capping is however not utilized by purpose to allow such alternative interactions to take place. 6) The active part of the stationary phase must also be stable to the mobile phase composition. Strongly acidic or basic mobile phase can also destroy the active phase and not only its support, by hydrolyzing it. The use of reagents with two or three reactive functionalities typically leads to more stable stationary phases. Stable stationary phases also have low bleeding of traces of compounds in the mobile phase. Low bleeding for the columns is very important in particular for detection when traces of impurities leaking into the mobile phase may lead to high background for some detectors in particular for sensitive detectors such as those based on mass spectrometry. One specific characteristic for hydrophobic stationary phases is their capability to work in a mobile phase with a high proportion of water, property indicated as wettability. Some columns, for example, of C8 or C18 type, must be used only with a maximum water content in the mobile phase of about 90%, and a water excess beyond this value may produce dewetting (or phase collapse). Column that suffered dewetting lose their separation capability and may become unutilizable. The process of dewetting is mainly caused by the exclusion of the aqueous mobile phase from the pores covered with a hydrophobic material and the inability of the mobile phase to reenter the pores after dewetting. However, the change in the conformation of the alkyl chains from the silica surface under the influence of water is also likely to occur and to play a role in the undesired modification of phase properties. 7) In some cases an active phases can be bound on organic polymers, but some polymers can be synthesized having an active phase in the structure, or no active phase may be necessary (e.g. for SEC use).

Summary of the main characteristics of columns used in HPLC The main characteristics of a chromatographic column important for its selection in a method development are summarized in Table 7.1.1.

Key points • • •

Column construction regarding its length and inner diameter affects the chromatographic separation. Stationary phase is (usually) made from a porous support and an active phase, each with specific role in column properties. The characteristics of a chromatographic column affect the retention time tR , the resolution Rs, the column backpressure, as well as other attributes for the separation.

7.2

Columns with hydrophobic character

The columns with hydrophobic character are used in RP-HPLC which is the most commonly utilized type of HPLC. By RP-HPLC, a wide range of molecules can be

Chromatographic columns in high-performance liquid chromatography

311

Table 7.1.1 Characteristics important in selecting a column. No.

Property

Common range

1 2 3 4 5

Type of stationary phase Column length Column i.d. Particle size Theoretical plate number (per m)

6 7

Silanol activity and end-capping (for silica) pH range resilience

8 9

Particle type Support for stationary phase

10 11 12 13 14

Purity (for silica), and metal activity Particle shape Surface area Pore size Resilience to backpressure

15

Reagent type for support derivatization (usually on silica) Polymerization type in case of bonded phase (usually on silica) Carbon load on hydrophobic columns Coverage of support for any functionality Phase ratio Lack of bleeding

See Sections 7.2e7.7 e.g., 20, 30, 50, 100, 150, 250 mm e.g., 1.0, 2.0, 3.0, 4.6 mm 1.6e5 mm (older column may have 10 mm) 30,000 to 120,000 for porous particles, higher for core-sell. Silanol activity low to high based on endcapping, polar end-capping Range 2e7 or from 1 to 11 for ethylene bridge type. Even larger range for organic polymers. Porous, core-shell, monolithic Silica, silica/organic (ethyl bridged), organic polymer, other inorganic, porous carbon High (type B) or medium (type A) Irregular or spherical 50e500 m2/g 50, 100, 300 Å, etc. Up to 18,000 psi for silica based, up to 1500 psi for some polymers used in size exclusion Mono-functional, bifunctional, trifunctional

16 17 18 19 20 21 22

23

Wetting characteristics for hydrophobic columns Temperature range of functionality

Life of the column (No. of injection without degradation)

Horizontal, vertical 3% to 25% C Depending on phase, 2.0 to 4.5 mmol/m2 log J between - 0.3 and - 0.9 Very inert columns to columns releasing traces of compounds in the mobile phase Poor to good Common temperature range between 5 and 65 C. For some columns used in SEC up to 150 C. Depending on the type of injected samples (cleanliness, volume) 300e600 injections

analyzed, including nonpolymeric ones with log Kow values in the range 1 to 6, as well as macromolecules such as proteins, other natural and synthetic polymers, etc. Columns with hydrophobic character are also used in ion-pair chromatography (IPC), in hydrophobic interaction chromatography (HIC), and in nonaqueous reversed-phase chromatography (NARP).

312

Method Development in Analytical HPLC

Stationary phases used in RP-HPLC and related techniques Columns with a variety of hydrophobic stationary phases are commercially available. Most columns having a hydrophobic stationary phase are silica based, but some other supports are sometimes used such as organic polymers, zirconia, silica hydride, or graphitized carbon. Several types of hydrophobic phases are indicated in Table 7.2.1.

Table 7.2.1 Several types of hydrophobic stationary phases. No.

Type of phase

Active moiety

Classification

1

Common n-alkyl on silica endcapped n-Alkyl on organic/silica endcapped (BEH type) Core-shell n-alkyl on silica endcapped Special n-alkyl on silica endcapped n-Alkyl on silica not endcapped n-Alkyl on silica polar endcapped Polar embedded on silica endcapped Polar embedded on silica polar endcapped Cyclic alkyl on silica Mixed alkyl on silica Aryl on silica or core-shell silica Mixed alkyl-aryl on silica

C8, C18

L8, L1, L42

C8, C18

L1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Cyano on silica Fluorinated alkyl or aryl phases on silica Other on silica Silicon hydride Monolith silica n-alkyl bonded Graphitized carbon Polymeric block

C8, C18 C2, C4, C12, C14, C20, C22, C27, C30 C8, C12, C14, C18

L16, L26, L30, L62

C8, C18 with CH2OH end-capping C8, C18, etc. with-O-, eCONHe, urea, etc. C8, C18, etc. with-O-, eCONHe, urea, etc. Cyclohexyl, n-C6 linked cyclohexyl C18-short alkyl, C4eC18 Phenyl, diphenyl, C2 linked phenyl, C6 linked phenyl C18-phenyl, C6-phenyl, cyclopropyl-phenyl, C14-phenyl Cyanopropyl, phenylcyanopropyl Pentafluorophenyl, perfluoroalkyl

L10 L43

Fullerene, cholesterol C8, C18 bonded, silicon hydride partially graphitized C8, C18

L3

Graphitized carbon Polystyrene/divinylbenzene, ethylvinylbenzene/ divinylbenzene

L68

L15

L11

L21

Chromatographic columns in high-performance liquid chromatography

313

Table 7.2.1 Several types of hydrophobic stationary phases.dcont’d No.

Type of phase

Active moiety

Classification

20

Phases bonded on organic polymers

21

Phases bonded on zirconia

C18 or phenyl bonded on polymethacrylate, C18 or pentafluorophenyl bonded on divinylbenzene C8, C18, etc.

L49

Characterization of columns with hydrophobic stationary phase The general properties of chromatographic columns as summarized in Table 7.1.1 are fully applicable for the columns with a hydrophobic stationary phase used in RPHPLC and in the related techniques (IPC, HIC, NARP). Specific for these columns is that the retention/elution process is based mainly on hydrophobic character of the stationary phase. This hydrophobic character can be estimated by different parameters (e.g., Ref. [7]). One such parameter is the value of retention factor k' for ethylbenzene. Ethylbenzene as test compound was selected because it is not polar and except for hydrophobic interactions, other interactions with the stationary phase are very likely absent (e.g., interactions with the remaining free silanols or with the embedded polar groups of the column). The values of log k' for ethylbenzene as a test analyte on various RP-HPLC columns for a mobile phase 50/50 (v/v) acetonitrile/aqueous phosphate buffer 60 mM at pH 2.8 are shown in Fig. 7.2.1.

1.40 1.20 1.00 0.80

CN

log k'

0.60

C4 C5

0.40

Phenyl

0.20

C8

0.00

Fluor C18

-0.20 -0.40 -0.60 0

2

4

6

8

10

12

14

16

18

20

No of Carbons

Figure 7.2.1 Values of log k' for ethylbenzene as a test analyte on various RP-HPLC columns for a mobile phase 50/50 (v/v) acetonitrile/aqueous phosphate buffer 60 mM at pH 2.8.

314

Method Development in Analytical HPLC

The most common columns used in RP-HPLC are those having C8 and C18 chains bonded on silica or ethylene-bridged silica, either fully porous or core-shell type. As shown in Fig. 7.2.1, the values for log k' for ethylbenzene for these columns cover a wide range indicating a broad interval of hydrophobicity, depending on other column characteristics not only on the number of carbons in the attached hydrophobic chain (C18). Phenyl phases are also quite hydrophobic, in addition being able to develop special pep interactions. Other hydrophobic phases are those with shorter hydrocarbon chains than the C8 or longer than C18. Those types of columns are used mainly for special applications. Fluorinated hydrocarbons phases do not differ in hydrophobicity from alkyl or aryl type phases, but they offer additional flexibility in some separations. The lowest hydrophobic character can be noticed for cyanopropyl type, which are in fact intermediate phases between hydrophobic and polar one. The cyano phases are used as RP-HPLC type when the mobile phase contains sufficient water to be more polar than the stationary phase. Special types of columns are those with an embedded polar group in a long hydrocarbon chain (e.g., C18). The embedded group can be ether, amide, urea, carbamate, sulfone, etc. The embedded polar groups do not decrease the column hydrophobicity, but offer better wettability and capability to generate additional types of interactions with the analytes besides hydrophobic ones. Some such columns can be used with 100% water mobile phase, which is not recommended for other columns that can develop “dewetting/phase collapse” when the mobile phase is 100% water. A better characterization of hydrophobic type columns is methylene selectivity aðCH2 Þ. This parameter was developed based on the observation that a linear dependence for log aðCH2 Þ on the number of CH2 groups exists for certain homologous series of compounds such as the one from benzene to amylbenzene. Based on this observation, the values for log aðCH2 Þ can be obtained from the slope of the trendline of the graph of log k' versus the number of methylene groups in an alkylbenzene. When a column is selected for an application requiring the separation of hydrophobic compounds, the preferred characteristic should be a high aðCH2 Þ for obtaining a better separation. A comprehensive characterization of hydrophobic type columns can be obtained using hydrophobic subtraction model [13e17]. Because in hydrophobic type columns besides hydrophobic interactions, other types of interaction mechanisms can also take place (e.g., steric, some polar, and even ion exchange type interactions), a parameter indicated as aðX; EBÞ was developed to capture all those possibilities. For a compound X in the separation on a hydrophobic type column, the following expression was developed for log aðX; EBÞ: 0

k log aðX; EBÞ ¼ log 0 X kEB

! 0

0

0

0

¼ h ðXÞH   s ðXÞS þ b ðXÞA þ a ðXÞB 0

þ k ðXÞC (7.2.1)

Chromatographic columns in high-performance liquid chromatography

315 0

The contribution of each term in Eq. 7.2.1 has a specific meaning, and h ðXÞH  accounts for column capability to separate the analyte from ethylbenzene (EB) based 0 0 only on hydrophobic interactions, s ðXÞS accounts for steric interactions, b ðXÞA accounts for hydrogen bonding between a basic solute and the acidic groups of 0 the stationary phase, a ðXÞB accounts for hydrogen bonding between an acidic so0 lute and basic groups of the stationary phase, and k ðXÞC accounts for cation ex0 0 0 0 0 change type and/or ioneion interactions. Parameters h , s , b , a , and k depend    on solute X properties (and mobile phase composition), and H , S , A , B , and C  depend only on column properties. The values for H  , S , A , B , and C can therefore be utilized for comparing different columns and consequently a possibility to establish when two different columns will have similar separation capabilities and can replace each other [18]. 0 0 0 0 0 The values for both h , s , b , a , k and H  , S , A , B , C  were obtained for a specific mobile phase that consisted of 50/50 acetonitrile/aqueous buffer containing 60 mM phosphate at pH ¼ 2.8, or at pH ¼ 7.0 based on a large number of measure0 ments for the retention factors kX on a specific set of compounds X and a number of different columns. The values for the parameters H  , S , A , B , and C are reported for about 753 different alkyl-silica columns (PQRI approach from Product Quality Research Institute) [19]. The hydrophobic subtraction model provides a useful procedure for the column comparison. A similarity parameter F between two columns can be calculated for each column pair by the formula: Fcol:1col:2 ¼

h

  Hcol:1  Hcol:2

2

 2  2 þ Scol:1  Scol:2 þ Acol:1  Acol:2 þ

 2   2 i1=2   Ccol:2 þ Bcol:1  Bcol:2 þ Ccol:1

(7.2.2)

The values for F are calculated in Ref. [19] and the closer to zero is the value of F, the more similar are the columns. For large F values, the columns can be considered “orthogonal,” leading to a different separation. The differences in selectivity between RP-LC columns can be evaluated quantitatively by this model, allowing the selection between different columns that are either equivalent or different in selectivity as the applications require [20].

Other properties and parameters characterizing hydrophobic stationary phases End-capping of silica support for the columns with hydrophobic character is another important property. End-capping with eCH3 groups covers the silanol groups of the silica and diminishes the polar interactions still possible with a column having hydrophobic chains such as C18 or C8 that are not end-capped. The better control of the types of interactions provided by the stationary phase by limiting them more toward hydrophobic interactions is in many cases a desired

316

Method Development in Analytical HPLC

property. The peak shape for polar compounds on highly end-capped columns is usually better defined (closer to ideal Gaussian). However, for some columns, such polar interactions in addition to the hydrophobic character are desired. Some RP-columns are intentionally not end-capped. Also, end-capping with e CH2OH groups, which are polar, was performed for some columns (e.g., Synergy Hydro from Phenomenex). This type of column can also be used with 100% aqueous mobile phase, without the risk of dewetting. Carbon load value indicated in Table 7.1.1 is another important characteristic of 0 RP-type columns, the columns with high C% content usually having higher k values for hydrophobic compounds. The correlation is not linear, but among other parameters of the column, this is a factor to consider for column properties. The horizontal or vertical type column characterization is also important for RPtype columns. The horizontal-type derivatization offers usually better peak shape 0 (close to ideal Gaussian), but lower loading capacity and lower k values for hydrophobic compounds. Vertical derivatization leads to columns with higher carbon load C% 0 and larger k values toward hydrophobic compounds. The type of reagent used for support derivatization, difunctional and trifunctional usually provides columns with better coverage of silanol groups and therefore with lower interference of polar interactions, but also more resilient to degradation stationary phases, more stable to a range of pH for stationary phase and usable for a larger number of injections. Another property characterizing all columns, including the hydrophobic ones, is peak asymmetry As (see Eq. 4.1.58). In order to put in evidence the column potential to generate asymmetric peaks, it is necessary to use specific test compounds and mobile phase composition. A column showing significant asymmetry for such compounds may show no asymmetry, for example, for a hydrocarbon. Two common test compounds used to put in evidence asymmetry are quinizarin and amitriptyline. Quinizarin is typically used for the evaluation of metal activity in a column, and amitriptyline, which is a basic compound (pKa ¼ 9.4), is used for the characterization of silanol activity based on its peak shape. Values for As (for amitriptyline and quinizarin in a specific mobile phase) are reported in the literature for a number of columns [21]. Some general properties for a number of common columns utilized in RP-HPLC are given in Table 7.2.2. A similar approach for comparing columns as the PQRI approach based on hydrophobic subtraction model is a different approach indicated as USP approach (U.S. Pharmacopeia) [19]. The comparison of columns is based in USP approach on a different set of parameters that include: Hy the retention factor k' for ethylbenzene, CTF the chelating tailing factor for quinizarin, CFA the retention factor k' for amitriptyline, TFA the tailing factor for amitriptyline, and BD the bonding density of hydrophobic phase in m mol/m2. A similarity parameter also noted with F is developed in this case, based on the formula: Fcol:1col:2 ¼



Hy;col:1  Hy;col:2

2

þ ðCTF col:1  CTFcol:2 Þ2 þ ðCFA col:1

2 2  CFAcol:2 Þ2 þ ðTFA col:1  TFAcol:2 Þ þ ðBD col:1  BDcol:2 Þ

(7.2.3)

No.

Column name

1 2 3 4 5 6 7

AccQ (C18) ACE C18 ACE C18-300 ACE C18-HL ACE SUPERC18 ACE C18-AMIDE Acquity UPLC BEH C18 Acquity UPLC CSH C18 Acquity UPLC HSS C18 Acquity UPLC HSS T3 Acquity UPLC CSH Phenyl-Hexyl Acquity UPLC CSH Fluoro Phenyl Atlantis T3 (C18)

8 9 10 11 12 13 14 15

Cover mM/ m2

Endcapd

pH Stabil.

log k’ Ethylbenzene

185 300 100 400 400 300 185

17 15.5 9 20 14.8 17.0 17

e 111,000 103,500 102,000 e e e

e e e e e e e

1 1 1 2 1 1 1

e e e e 1.5e1.5 2e8 e

e 1.000 0.983 1.045 e e 1.000

0.70

230

15

e

e

1

e

0.968

100

0.70

230

15

e

e

1

1e8

e

Sph. 1.8

100

0.70

230

11

e

e

1

2e8

e

Sph. 1.7

130

0.70

185

14

e

e

1

e

0.764

Sph. 1.7

130

0.70

185

10

e

e

1

e

e

Sph. 3, 5, 10 Sph. 3,5 CeS 3.6

100

1.00

330

12

e

1.6

1

e

0.941

100 e

e e

330 25

12 e

e e

e e

1 3

3e7 1.5e9

0.917 0.934

CeS 3.6

e

e

25

e

e

e

4

1.5e9

0.788

Pore size Å

Pore vol. mL/g

Surface area m2/g

Sph. 1.7 5 5 5 1.7, 2, 3, 5 1.7, 2, 3, 5 Sph. 1.7

135 100 300 90 90 100 130

0.70 e e e e e 0.70

Sph. 1.8

100

Sph. 1.8

Continued

317

16

Atlantis dC18 Aeris widepore XBC18 Aeris widepore XBC8

C%b

Theor. Plate N/ mc

Particle diam. dp mma

Chromatographic columns in high-performance liquid chromatography

0

Table 7.2.2 Common physical and chemical properties of various hydrophobic columns, log k for ethylbenzene in mobile phase 50/50 acetonitrile/ aqueous buffer 60 mM phosphate v/v.

0

No.

Column name

17

Aeris widepore XBC4 Aeris PEPTIDE XB-C18

18

19 20 21 22

31 32

Clarity Oligo-WAX Clarity Oligo-MS

23 24 25 26 27 28 29

Cover mM/ m2

Endcapd

pH Stabil.

log k’ Ethylbenzene

25

e

e

e

2

1.5e9

0.699

e

200

10

e

e

1

1.5e9

e

1.05 1.15

320 215 450 e

15 11 25 e

e e e e

e e 3.7 e

2 4 2 2

2.5e7.5 2.5e7.5 2e8 2e9

0.966 0.869 1.077 1.136

Pore size Å

Pore vol. mL/g

Surface area m2/g

CeS 3.6

e

e

CeS 1.7, 2.6, 3.6, 5 Sph. 3, 5 Sph. 5 Sph. 5 Sph. 5

100

125 200 100 90

e

CeS 2.7 CeS 2.7

90 90

120 120

9 7

200,000 200,000

e e

1 1

2e8.5 2e8.5

e e

CeS 2.7 CeS 2.7 Irreg. 10 Irreg. 10 5

90 90 148 125 120

1.10 e e

120 120 300 330 300

8 5 10 m. 10 15

200,000 200,000 e 36,000 51,000

e e 1.61 e e

1 1 4 3 4

1.5e12 1.5e10 2.5e7.5 2.5e7.5 e

e e 0.824 0.790 1.030

Sph. 3, 5, 10 Sph. 10 CeS 1.3, 1.7, 2.6, 5

110

e

375

14

e

e

2

1e12

e

360 100

e e

e 200

e 12

e e

0.80 e

2 3

1e11 1.5e10

e e

Method Development in Analytical HPLC

30

Aqua C18 Aqua C8 Ascentis C18 Ascentis express C18 CeS Boltimate C18 Boltimate PhenylHexyl Boltimate EXT-C18 Boltimate PFP Bondaclone C18 mBondapak C18 Capcell Pak AG C18 Clarity Oligo-RP

C%b

Theor. Plate N/ mc

Particle diam. dp mma

318

Table 7.2.2 Common physical and chemical properties of various hydrophobic columns, logk for ethylbenzene in mobile phase 50/50 acetonitrile/ aqueous buffer 60 mM phosphate v/v.dcont’d

47 48

Gemini C6-Phenyl Gemini NX-C18

49

HyperClone C8 HyperClone C18 HyperClone (C8) HyperClone (C18) HyperClone (CPS)

36

50 51 52 53

BDS

Sph. 5 Sph. 5 Sph. 1.6, 2.7 Sph. 1.6, 2.7 Sph. 5, 15 Sph. 5, 15 5 5 5 Sph. 5 5 5 5 Sph. 3, 5,10 Sph. 3, 5 Sph. 3, 5,10 Sph. 3, 5

110 110 90

e e 0.26

375 375 100

13 19 5.7

e e e

e e e

1 1 1

2.5e7.5 2.5e7.5 e

e e 1.037

90

0.26

100

6.6

e

e

1

e

1.075

100 100 140 100 140 180 100 100 100 110

1.00 1.00 e e e e e e e 1.10

300 300 300 450 300 200 200 200 200 375

7.3 17 18 15 18 12 11 11 12 14

e e 85,500 66,000 92,000 e 93,000 114,000 82,000 75,500

e e e 3.0 e e e e

1 1 3 1 2 2 2 4 4 1

e e e e e 2e8 e e e 1e12

e 1.028 0.980 0.963 0.996 0.984 0.992 e e 1.010

110 110

e e

375 375

12 14

e e

e e

3 4

1e12 1e12

0.803 0.969

130

0.60

155

7

e

e

3

2.0e7.5

0.847

BDS

Sph. 3, 5

130

0.60

155

11

e

e

4

2.0e7.5

0.988

MOS

Sph. 3, 5

120

0.60

155

6.6

e

e

4

2.0e7.5

0.847

ODS

Sph. 3, 5

120

0.60

155

10

e

e

3

2.0e7.5

1.030

CN

Sph. 3, 5

120

0.60

155

4

e

e

1

2.0e7.5

e Continued

319

37 38 39 40 41 42 43 44 45 46

Columbus C8 Columbus C18 CORTEX C18þ (trifunct.) CORTEX C18 (trifunct.) Delta-Pak C4 Delta-Pak C18 Develosil ODS-HG Develosil ODS-MG Develosil ODS-UG Discovery HS C18 Exsil ODS Exsil ODS1 Exsil ODSB Gemini C18

Chromatographic columns in high-performance liquid chromatography

33 34 35

0

Column name

54 55 56 57 58 59 60 61 62 63 64 65

68 69 70 71 72

Hichrom RPB Hypersil BDS C18 Hypersil GOLD Hypersil ODS Hypersil SAS C1 HyPurity C18 IB-Sil C18 IB-Sil C18 Inertsil ODS-2 Inertsil ODS-P InertSustain C18 InertSustain AQC18 InertSustain C8 InertSustain AXC186 Inertsil ODS-HL Inertsil ODS-3 Inertsil ODS-2 Inertsil C4 Jupiter C4

73 74

Jupiter C5 Jupiter C18

66 67

Cover mM/ m2

Endcapd

pH Stabil.

log k’ Ethylbenzene

97,500 76,500 91,000 94,500 73,000 e e 73,500 e e 120,000

e e e e e e 3.27 4.29 e e e

2 2 4 2 2 1 3 4 4 4 1 1

e e e e 2e8 e 2.5e7.5 2.5e7.5 e e 1e10 1e10

e 0.993 0.881 0.974 0.980 e e 1.007 0.978 1.010 0.939

e 8

e

e

1 1

1e10 1e9

0.868 e

450 450 320 320 170

e 15 18.5 7.5 5

e 60,500 32,000 e e

e e e e 6.30

1 3 3 1 3

1e8 e e 1e10 1.5e10

1.070 0.990 1.007 e 0.698

170 170

5.5 13.34

e e

5.30 5.50

2 2

1.5e10 1.5e10

0.729 0.945

Pore size Å

Pore vol. mL/g

Surface area m2/g

C%b

5 5 5 5 Sph. 5 5 Sph. 3, 5 Sph. 5 5 5 2,3,5 1.9,3,5

110 130 180 120 120 180 125 125 100 100 100 100

e e e e e e 0.75 0.75 e 1.15 e e

340 170 200 170 170 200 165 165 350 450 350 350

14 11 10 10 3 13 11 m. 7.5 m. 14 29 e e

2,3,5 Sph. 3, 5

e 200

e 1.0

350 200

3,5 5 5 5 Sph. 5,10, 15 Sph. 5,10 Sph. 5,10, 15

100 100 150 150 300

e e e e e

300 300

e e

Method Development in Analytical HPLC

No.

Theor. Plate N/ mc

Particle diam. dp mma

320

Table 7.2.2 Common physical and chemical properties of various hydrophobic columns, logk for ethylbenzene in mobile phase 50/50 acetonitrile/ aqueous buffer 60 mM phosphate v/v.dcont’d

Jupiter Proteo Kinetex EVO C18

77

Kinetex C18

78

Kinetex XB-C18

79

Kinetex C8

80

Kinetex biphenyl

81

Kinetex PhenylHexyl Kinetex F5

82 83 84 85 86 87

Kromasil C18 LiChrosorb RP-18 LiChroSph. 100 RP18 Luna PFP(2) Luna Phenyl-Hexyl

88 89 90

Luna C5 Luna C8 Luna C8(2)

91 92

Luna C18 Luna C18(2)-HST

Sph. 4, 10 CeS 1.7, 2.6, 5 CeS 1.3,1.7, 2.6, 5 CeS 1.7, 2.6, 5 CeS 1.7, 2.6, 5 CeS 1.7, 2.6, 5 CeS 1.7, 2.6, 5 CeS 1.7, 2.6 5 10 5 Sph. 3, 5 Sph. 3, 5,10 Sph. 5,10 Sph. 5,10 Sph. 3, 5,10,15 Sph. 5,10 Sph. 2.5

90 100

e e

475 200

15 11

e e

e e

1 2

1.5e10 1e12

1.010

100

e

200

12

e

e

4

1.5e8.5

1.055

100

e

200

10

e

e

3

1.5e8.5

0.975

100

e

200

8

e

e

3

1.5e8.5

0.864

100

e

200

11

e

e

2

1.5e8.5

-

100

e

200

11

e

e

2

1.5e8.5

0.795

100

e

200

9

e

e

2

1.5e8.5

e

100 100 100

e e e

340 300 350

19 17 21.6

99,000 74,000 80,000

e e e

4 1 2

e e e

1.050 0.909 1.006

100 100

1.00 1.00

400 400

11.5 17.5

e e

2.20 4.00

3 4

1.5e9.0 1.5e9.0

0.753 0.782

100 100 100

1.00 1.00 1.00

440 440 440

12.5 14.75 13.5

e e e

7.85 5.50 5.50

4 3 3

1.5e9.0 1.5e9.0 1.5e9.0

0.800 0.875 0.889

100 100

1.00 1.00

440 400

19 17.5

e e

3.00 3.00

3 2

1.5e9.0 1.5e9.0

1.018 e 321

Continued

Chromatographic columns in high-performance liquid chromatography

75 76

0

Column name

93

Luna C18(2)

94

Luna Omega C18

95 96 97

Luna Omega polar C18 Luna C5 Luna CN

98

Luna Phenyl-Hexyl

99 100 101 102 103

Luna PFP Novapak C18 Novapak C8 Novapak CN Novapak Phenyl Hexyl Nucleosil C18 Nucleosil C18AB Nucleoshell RP 18 Nucleodur C18 Gravity Onix C18 Partisil ODS Partisil ODS2 Partisil ODS3

104 105 106 107 108 109 110 111

C%b

Cover mM/ m2

Endcapd

pH Stabil.

log k’ Ethylbenzene

400

17.5

88,000

3.00

2

1.5e9.0

1.002

e

260

e

350,000

e

1

1.5e8.5

0.976

100

e

260

11.0

e

e

1

e

0.860

100 100

e 1.00

440 400

12.5 7

e e

e 3.80

1 1

e 1.5e7.0

0.800 0.452

100

e

400

17.5

e

e

e

e

0.782

100 60 60 60 60

e e e e e

400 120 120 120 120

e 7.3 4 3 5

e 60,000 e e e

e e e e

e 2 1 1 2

e e 2e8 2e8 2e8

0.753 1.049 0.899 0.362 e

5 5 CeS 2.7 1.8, 3, 5

100 100 90 110

e e e e

350 350 e e

15 24 7.5 18

101,000 87,000 250,000 80,000

e e e e

4 3 1 1

e e 1e11 1e11

0.906 e e 1.056

Monolith 10 10 10

130 85 85 85

1.00 e e e

300 350 350 350

18 5 15 10.5

e 47,500 41,000 52,000

3.60 e e e

3 2 2 2

2.0e7.5 e e e

1.012 e e 0.810

Pore size Å

Pore vol. mL/g

Surface area m2/g

Sph. 3, 5,10,15 Sph. 1.6, 3, 5 Sph. 1.6, 3, 5 Sph. 5, 10 Sph. 3, 5,10 Sph. 3, 5,10,15 Sph. 3.5 Sph. 4 Sph. 4 Sph. 4 Sph. 4

100

1.00

100

Method Development in Analytical HPLC

No.

Theor. Plate N/ mc

Particle diam. dp mma

322

Table 7.2.2 Common physical and chemical properties of various hydrophobic columns, logk for ethylbenzene in mobile phase 50/50 acetonitrile/ aqueous buffer 60 mM phosphate v/v.dcont’d

PhenoSphere C1

113

PhenoSphere C6

114

PhenoSphere C8

115

PhenoSphere ODS(1) PhenoSphere ODS(2) PhenoSphere CN

116 117 118

121

PhenoSphere NEXT C8 PhenoSphere NEXT C8 PhenoSphere NEXT Phenyl PolymerX RP-1

122

Poroshell EC-C18

123

Poroshell SB-C18

124

Poroshell HPH-C18

119 120

Sph. 3, 5,10 Sph. 3, 5,10 Sph. 3, 5,10 Sph. 3, 5,10 Sph. 3, 5,10 Sph. 3, 5,10 Sph. 3, 5

80

0.50

220

4 m.

e

1.80

4

2.5e7.5

e

80

0.50

220

6 m.

e

2.27

1

2.5e7.5

e

80

0.50

220

6 m.

e

3.54

2

2.5e7.5

e

80

0.50

220

7 m.

e

1.74

3

2.5e7.5

e

80

0.50

220

12 m.

e

2.50

4

2.5e7.5

e

80

0.50

220

4 m.

e

2.50

4

2.5e7.5

e

120

e

380

10

e

e

3

2.5e7.5

e

Sph. 3, 5

120

e

380

14

e

e

3

2.5e7.5

e

Sph. 5

120

e

380

11

e

e

3

2.5e7.5

e

Sph. 3, 5,10,15 CeS Sph 1.9, 2.7, 4 CeS Sph 1.9, 2.7, 4 CeS Sph 1.9, 2.7, 4

100

e

410

-

e

e

2

0e14

e

120

e

130

10

e

e

1

2e8

1.023

120

e

130

8

e

e

4

1e8

0.956

100

e

95

e

e

e

1

3e11

1.029

323

Continued

Chromatographic columns in high-performance liquid chromatography

112

0

Column name

125

Poroshell CS-C18

126

Poroshell SB-Aq

127 128 129

Prodigy ODS (2) Prodigy C8 Prodigy ODS (3)

130

Prodigy Phenyl (PH-3) Resolve C18 Resolve C8 SiliaChrom AQ C18 SiliaChrom AQ C8 SiliaChrom dt C18 SiliaChrom SB C8300 SiliaChrom XDB C18 SiliaChrom XDB C8 SiliaChrom XDB1 C18

131 132 133 134 135 136 137 138 139

C%b

Cover mM/ m2

Endcapd

pH Stabil.

log k’ Ethylbenzene

95

e

e

e

2

1e11

e

e

130

e

e

e

4

1e8

0.581

150 150 100

1.10 1.10 1.00

310 310 450

18.4 m. 12.5 m. 15.5 m.

48,000 48,000 62,000

3.50 5.00 e

2 1 2

2.0e9.0 2.0e9.0 2.0e9.0

0.995 e 1.023

100

e

450

62,000

e

4

2.0e9.0

0.529

5 5 3, 5, 10 3, 5,10 2.5, 3, 5,10 5

90 90 100 100 100 300

0.5 0.5 e e e e

200 200 380 380 410e440 80

10.0 poly. 10.2 5.1 18 14 18 3

45,500 45,500 e e e e

e e e e e e

4 4 1 1 1 1

e e 1.5e9 1.5e9 1.5e9 1.0e7.5

0.968 e e e e e

5

150

e

200

15

e

e

1

1.5e9.0

e

3, 5

150

e

200

8

e

e

1

1.5e9.0

e

3, 5

100

e

380-400

22

45,000

e

1

1.5e10

e

Pore size Å

Pore vol. mL/g

Surface area m2/g

CeS Sph 1.9, 2.7, 4 CeS Sph 1.9, 2.7, 4 Sph. 5 Sph. 5 Sph. 3, 5,10 Sph. 5

100

e

120

Method Development in Analytical HPLC

No.

Theor. Plate N/ mc

Particle diam. dp mma

324

Table 7.2.2 Common physical and chemical properties of various hydrophobic columns, logk for ethylbenzene in mobile phase 50/50 acetonitrile/ aqueous buffer 60 mM phosphate v/v.dcont’d

140

151

Synergy Max-RP

152

Synergy Hydro-RP

153

Synergy Polar-RP

154 155 156 157 158

Symmetry C18 Symmetry C8 Symmetry C4 Symmetry 300 C18 Symmetry Shield C18 Symmetry Shield C8 TSKgel Protein C4300

141 142 143 144

159 160

5, 10

300

e

80

8

e

e

1

1.5e9.0

e

Sph. 5 Sph. 3, 5 Sph. 3, 5

80 80 80

e e e

200 200 200

6 6 7

e e e

e e e

2 4 3

2.5e7.5 2.5e7.5 2.5e7.5

e e e

Sph. 3, 5

80

e

200

12

e

e

3

2.5e7.5

0.975

Sph. 3, 5 5 5 Sph. 2.5, 5 Sph. 2.5, 5 Sph. 2.5, 4, 10 Sph. 2.5, 4, 10 Sph. 2.5, 4, 10 Sph. 2.5, 4, 10 Sph. 3.5, 5 Sph. 3.5, 5 Sph 5 Sph. 3.5, 5 Sph. 5

80 80 80 100 100 80

0.50 e e 0.90 0.90 1.05

220 220 220 340 340 475

6.2 11.5 11.5 16 12 12

e 91,500 92,000 91,500 e e

e e e e e e

4 1 2 1 1 1

e e e e e 1.5e9.0

0.682 0.962 0.975 1.031 0.856 0.879

80

1.05

475

17

e

3.21

2

1.5e9.0

0.989

80

1.05

475

19

e

2.45

3

1.5e7.5

1.022

80

1.05

475

11

e

3.15

4

1.5e7.5

0.654

100 100 300 300 100

0.90 0.90 e 0.80 0.90

335 335 110 110 335

19.1 11.7 2.8 8.5 17

92,000 e e e e

e e e e e

1 1 2 1 1

e e 2e8 e e

1.052 0.893 0.687 0.984 0.850

Sph. 3.5, 5

100

0.90

335

15

e

e

1

e

0.730

3

300

e

e

3

e

e

1

e

e 325

Continued

Chromatographic columns in high-performance liquid chromatography

145 146 147 148 149 150

SiliaChrom XDB1 C18-300 SphereClone C6 SphereClone C8 SphereClone ODS (1) SphereClone ODS (2) Spherysorb ODS(1) Spherisorb ODS(2) Spherisorb ODSB SunFire C18 SunFire C8 Synergy Fusion-RP

0

Column name

161 162 163 164 165 166 167 168

TSKgel ODS140HTP TSKgel ODS-100V TSKgel ODS-120A Ultracarb C8 Ultracarb ODS (20) Ultracarb ODS (30) Vydac 218 TP XBridge C18

169

XBridge C8

170

XBridge Shiled RP18 XBridge BEH C18 XBridge Phenyl (trifunct.) XSelect CSH C18

171 172 173 174 175 176 177 178

XSelect CSH Phenyl-Hexyl XSelect CSH Fluoro-Phenyl XTerra RP18 XTerra Shield RP18 XTerra RP8

C%b

Cover mM/ m2

Endcapd

pH Stabil.

log k’ Ethylbenzene

e

6

e

e

1

e

0.730

e e 0.80 0.75 0.80 e 0.70

e e 550 370 550 70 185

15 m. 22 14 m. 22 m. 31 m. 8 18

e e e e e 63,000 e

e e 2.71 3.53 4.06 e e

1 4 3 3 3 2 1

e e 2.5e7.5 2.5e7.5 2.5e7.5 e 1e12

0.901 0.896 e e 1.114 0.909 1.007

130

0.70

185

13

e

e

1

1e12

0.805

130

0.70

185

17

e

3.3

1

1e12

0.835

130 135

0.70 0.70

185 185

18 15

e e

e 3.0

1 1

1e12 1e12

e 0.835

130

0.70

185

15

e

e

1

1e11

e

130

0.70

185

14

e

e

1

1e11

0.708

130

0.70

185

10

e

e

1

1e8

0.498

125 125 125

0.70 0.70 0.70

175 175 175

15 15 13.5

e e e

e e e

1 1 1

2e12 2e12 2e12

0.757 0.657

Pore size Å

Pore vol. mL/g

Surface area m2/g

2.3

140

e

3, 5 5,10 Sph. 5 Sph. 3, 5 Sph. 5 5 Sph. 2.5, 3.5, 5 Sph. 2.5, 3.5, 5 Sph. 2.5, 3.5, 5 Sph. 3.5, 5 Sph. 2.5, 3.5,5 Sph. 2.5, 3.5, 5 Sph. 2.5, 3.5, 5 Sph. 2.5, 3.5, 5 Sph. 3.5, 5 Sph. 3.5 Sph. 3.5, 5

100 150 60 90 60 300 130

Method Development in Analytical HPLC

No.

Theor. Plate N/ mc

Particle diam. dp mma

326

Table 7.2.2 Common physical and chemical properties of various hydrophobic columns, logk for ethylbenzene in mobile phase 50/50 acetonitrile/ aqueous buffer 60 mM phosphate v/v.dcont’d

XTerra MS C18

180

XTerra MS C8

181 182

XTerra Phenyl YMC J’Sphere ODS H80 YMC J’Sphere ODS M80 YMC ODS a YMC ODS AM YMC Pro C18 Zorbax Extend C18 Zorbax ODS Zorbax Rx-C18 Zorbax SB-C18 Zorbax Eclipse XDB-C18

183 184 185 186 187 188 189 190 191 a

Sph. 2.5, 3.5, 5 Sph. 2.5, 3.5, 5 Sph. 3.5, 5 4 4 5 5 5 5 5 5 3.5, 5 1.8, 3.5, 5

125

0.70

175

15.5

e

e

1

2e12

0.984

125

0.70

175

12

e

e

1

2e12

0.803

125 80

0.70 e

175 510

12 22

e 64,500

e e

1 4

2e12 e

0.683 e

80

e

510

14

58,000

e

3

e

e

120 120 120 80 70 80 80 80

e e e e e e e e

300 300 335 180 330 180 180 180

17 17 16 12.5 20 12 10 10

99,500 83,500 105,000 80,500 85,500 90,500 103,000 96,000

e e e e e e e e

3 2 4 3 3 2 2 2

e e e e e e e 2e9

e e 1.010 1.098 1.089 1.077 0.996 1.077

CeS indicates core-shell, Sph. indicates spherical particles, Irreg. indicates irregular particles, no specification for unknown particle shape. m. after the value of C% indicates monomeric type derivatization for the bonded phase. Plate number N/m refers to 5 mm particles if more than one particle dimension is indicated.  However, many columns are available in more than one particle size formats. For the estimation of N for columns with fully porous particles, the following expression can be used: N z 1000L Ct: dp with Ct. z 2, 2.5 or 3 and dp the diameter of the particle. d 1 indicates very low silanol activity to 4 high silanol activity. b c

Chromatographic columns in high-performance liquid chromatography

179

327

328

Method Development in Analytical HPLC

The two approaches, PQRI and USP are very useful for comparing columns. They facilitate the choice of a replacement column with similar properties when the recommended column for a separation is not available. The parameters listed for various columns in the PQRI ðH  , S , A , B , and C  Þ and in USP data bases (Hy , CTF, CFA, TFA, and BD) allow the estimation of some specific characters of a column (hydrophobicity, hydrogen bonding, ion exchange effects, tailing, etc.). Other parameters and tests also have been developed for hydrophobic columns characterization (e.g., Ref. [7]).

Key points • •

The columns with mainly hydrophobic character are commercially available in a wide variety of constructions and with various properties manifested in the separations. Several parameters can be used for column characterization and for column comparison that are useful in the selection of a column in method development.

7.3

Columns with polar character

The columns with polar character are used in HILIC and also in normal-phase HPLC (normal-phase chromatography (NPC)). These types of chromatography are utilized for the separation of polar compounds including important classes of molecules such as nonpolymeric carbohydrates, and organic acids. HILIC-type HPLC can also be used in the analysis of certain polymeric molecules such as nucleic acids.

Stationary phases used in HILIC and NPC A variety of stationary phases are used in HILIC chromatography. These include bonded phases containing terminal polar groups such as amino, diol, amide, etc., and also bare silica when the silanols act as the active polar groups. Phases containing attached cyano groups already discussed together with hydrophobic phases, having some polarity can also be used in HILIC applications. For HILIC applications, the mobile phase must be less polar than the cyano stationary phase, although it should contain some water. Several types of polar bonded moieties on silica and on polymeric substrates used in HILIC chromatography are indicated in Table 7.3.1. The columns used in HILIC display specific characteristics in addition to polarity (and hydrogen bonding capability). These include some ion exchange character, some hydrophobicity, and also some steric hindrance. In addition, the type of interactions dominating the separation are influenced by the nature of mobile phase, as explained for cyano-type phases. The polarity of stationary phases used in HILIC is its main characteristic. This property can be considered as decreasing in the following order: bare silica z tertiary amine > primary amine > amide > zwitterionic > imide > urea > diol > cyano. However. this order can change significantly depending on the silica base structure, coverage of the silica support, use of other support materials (polymeric or zirconia),

Chromatographic columns in high-performance liquid chromatography

329

Table 7.3.1 Several types of polar stationary phases used in HILIC chromatography. No.

Type of phase

Phase

1 2 3 4 5 6 7 8 9

Silanol Diol on silica Amide on silica Weak polar on silica Weak anionic on silica Weak cationic on silica Zwitterionic on silica Mixed mode on silica Weak anionic polymeric

Bare silica Diol, ether embedded plus diol Amide terminal, polyamide Cyano (also used in RP mode) eC3H6eNH2, diethylamine, triazole, etc. Sulfonylethyl, etc. Amino-sulfonic, amino-carboxylic Both polar and hydrophobic groups Different polar groups on porous polymers

etc. Stationary phases for polar columns can be synthesized by different procedures and the monomeric type or polymeric type (vertical) can be obtained in a similar manner as indicated in Section 7.1. The ion exchange character of HILIC phases is not necessarily in the same order of intensity as polarity. Amine columns, for example, display a stronger ion exchange character compared with bare silica. Zwitterionic columns (Zic-HILIC) display ion exchange properties, but the two groups seem to suppress this character as a whole. An example of a HILIC zwitterionic column structure is indicated in Fig. 7.3.1. The base silica can also be ethylene-bridged type (e.g., Atlantis BEH Z-HILIC from Waters). The variety of polar groups that can be seen in polar-type columns is quite large. They may include in addition to terminal polar groups, also embedded polar groups, polar premade polymers bonded to silica, such as polysuccinimide, polyetheylene glycol, or polyhydroxyethyl- aspartamide as well as several associations of polar moieties like in the structure of a stationary phase indicated in Fig. 7.3.2 [22]. Because the polar groups in polar-type stationary phases are usually attached to the silica support using a spacer (handle) typically consisting of a hydrocarbon chain that may contain 3 to 10 carbon atoms, a certain hydrophobic character is also present in

Figure 7.3.1 Schematic structure of a sulfobetaine Zic-HILIC stationary phase.

Figure 7.3.2 Schematic structure of a polar stationary phase obtained from a reaction of azide functionalized silica and “click” maltose.

330

Method Development in Analytical HPLC

the columns used for HILIC-type chromatography. For this reason, the carbon load C% of the HILIC columns should also be considered for column characterization. NPC is usually practiced on silica, although other polar stationary phases were used in the past, such as alumina. Some other columns used for HILIC separations can also be used in NPC. The typical properties indicated in Table 7.1.1, characteristic for all columns used in HPLC should also be considered for HILIC and NPC columns. For example, silica purity, coverage of support with the bonded phase, preparation procedure using mono-, bi-, or trifunctional reagents, number of theoretical plates, low bleeding are important for HILIC and (when applicable) to NPC columns. The resilience to low or high pH values of the mobile phase is also an important characteristic of HILIC columns that are silica based. The extension of pH range of stability can be achieved for HILIC columns using, for example, ethylene-bridged support instead of silica, similar to the case of RP-type columns [23]. Other characteristics are of lower concern, such as wettability which is always good for polar columns. In some cases, even end-capping is performed on polar columns, in order to avoid interference of silanols interactions with the intended type of column functionality (e.g., TSKgel NH2-100 is end-capped with TMS groups). However, some polar columns are not end-capped. At the same time, the amount of water present on the column surface is important in HILIC and in particular in NPC where the silica surface must be maintained with a certain content of water. Because in NPC the mobile phase does not contain water, the mobile phase slowly removes the water from the stationary phase surface. This affects the separation and leads to lack of reproducibility in NPC separations. To avoid this problem in NPC, either traces of water are added in the mobile phase, or the stationary phase surface is conditioned before use with a solvent containing water.

Characterization of columns with polar stationary phase The interactions responsible for the separation on HILIC and NPC phases are more complex than, for example, in RP-HPLC columns where hydrophobic interactions were predominant. As a result, different HILIC-type columns may behave very differently one from another type in a separation. For the description of HILIC-type columns characteristics, several dedicated parameters were developed and they are listed below. 1) For the description of the overall polarity of the column, retention factor k' for uridine or k'ðUriÞ can be utilized. The value for k'ðUriÞ is obtained using as mobile phase acetonitrile/aqueous buffer 90/10, the buffer being 20 mM ammonium acetate at pH ¼ 4.7. 2) Overall hydrophobicity of the column is described by methylene selectivity a ðCH2 ÞHILIC defined using the expression: 0

aðCH2 ÞHILIC ¼

k ðUriÞ k ð5MeUriÞ 0

0

(7.3.1)

In Eq. 7.3.1, k ð5MeUriÞ is the retention factor k 0 for 5-methyluridine measured in the same conditions as k 0 ðUriÞ.

Chromatographic columns in high-performance liquid chromatography

331

3) Hydrophilic character related to hydrogen bonding is measured based on parameter aðOHÞ 0 0 obtained as the ratio of retention factors for uridine k ðUriÞ and 2-deoxyuridine k ð2dUriÞ. 4) Isomer selectivity is described by two parameters, one describing the capability of the column to separate cis-trans isomers and the other to separate diastereoisomers. The first param0 eter is expressed by the ratio retention factors for 20 -deoxyguanosine k ð2dGuaÞ and the one 0 for 30 deoxyguanosine k ð3dGuaÞ. The second parameter is expressed by the ratio of retention 0 0 factors for vidarabine k ðVidÞ and the one for adenosine k ðAdeÞ. 5) Molecular shape selectivity is expressed by a parameter equal to the ratio of retention factors for 4-nitrophenyl-a-D-glucopyranoside and the one for 4-nitrophenyl-b-D-glucopyranoside. 6) A parameter describing anion exchange properties aAX is given by the ration of retention factors for sodium p-toluenesulfonate and the one for uridine k 0 ðUriÞ. 7) A parameter describing cation exchange properties aCX is given by the ration of retention factors for N,N,N-trimethylphenylammonium chloride and the one for uridine k'ðUriÞ. 8) A parameter describing the basic, neutral, or acidic character of the column is obtained from 0 0 the ratio of retention factors for theobromine k ðTbÞ and theophylline k ðTfÞ. This parameter indicated as at shows a basic character if at < 1, neutral character if at z 1, and acidic character if at > 1. 9) Peak asymmetry AsðXÞ is another parameter used for HILIC column characterization and it is usually obtained for the peak of uridine.

The measurements for all retention factors of specific compounds are also performed in the same mobile phase as used for measuring k0 ðUriÞ. More parameters are described in the literature for the characterization of HILIC columns (e.g., Ref. [7]). The values for all those parameters allow a comparison of HILIC columns regarding their separatory properties and if reported, must be used when selecting a column.

Key points • •

The columns used in HILIC and NPC have a stationary phase with polar character, but the separation on those columns is also affected by other interactions such as of ion exchange type or for HILIC of hydrophobic type. Several parameters are used for column characterization in HILIC.

7.4

Chiral chromatographic columns

Chiral HPLC is an important type of chromatography since many compounds, in particular, many pharmaceutical drugs have chiral molecules, and the physiological properties of different chiral isomers can be very different. The isomers generated by chiral atoms can be enantiomers (mirror images to each other) and diastereoisomers. Although the diastereoisomers can be separated on achiral stationary phases, the enantiomers can be separated only on chiral stationary phases (except for a few limited cases where the separation can be performed using an achiral stationary phase but a chiral mobile phase). A short description of chiral stationary phases is given in this section. Information on chiral separations is common in the scientific literature (e.g., Refs. [24,25]), and even a journal (Chirality) is dedicated to the subject.

332

Method Development in Analytical HPLC

Chiral stationary phases Stationary phases with chiral properties (having chiral selectors) are available in a variety of structures, and similar to other columns can have silica support for assuring porosity. On the silica support, the chiral active phase can be coated or bonded [26]. The separations on chiral phases are usually based on multipoint interactions between the analyte enantiomers that should be separated, and the active part containing chiral selectors of the stationary phase. Initial theory indicated that at least three points of interaction must exist for being possible to have chiral separation, but in fact the separation process for chiral molecules is more complex. The interactions involved in chiral separations are the common ones such as the dipole-dipole, hydrogen bonding, p-donor- p-acceptor, p-p stacking types, but their simultaneous presence in special structures generates chiral selectors. For allowing multiple interactions, in chiral separations, the mobile phase is frequently similar to the one used in normal phase (NPC) type condition or HILIC, and characterized by the absence of water or with water at low levels. However, because many analytes are soluble only in water plus polar solvents, significant effort has been made to develop chiral stationary phases that can be used with water þ solvents, where in addition to other interactions, the hydrophobic ones are also involved in the separation that become similar to that for RPHPLC. Even utilization of chiral phases applicable to ionizable compounds have been developed. As a result, chiral separations can be performed in normal phase (NP), polar organic (PO), reversed phase (RP), and even polar ionic (PI) mode. In NP-type separations, the stationary phase has specific polar groups and the mobile phase does not contain water. The NP type of separation is very common. For polar organic type separation, the stationary phases also has polar groups, but the mobile phase has some polarity and possibly contains water although the polarity of the mobile phase remains lower than that of the column. For RP mode separation, the stationary phase contains both polar groups and some hydrophobic moieties, while the mobile phase contains water and the separation resembles the RP-type separations. The polar ionic mode (PI or PIM) is used for the separation of ionizable molecules. Such applications use special columns having a stationary phase with polar groups and as a mobile phase methanol with or without water, plus small amounts of acids or bases (typically 0.1 g per 100 mL MeOH), or volatile ammonium salts (acetate, trifluoroacetate, or formate to allow MS detection). The main types of chiral stationary phases depending on the nature of the chiral elements and on the support on which it is coated or bonded are indicated in Table 7.4.1. 1) The Pirkle-type chiral stationary phases use a silica support having attached one or more chiral moieties plus several additional groups capable of interacting with the analytes. Such structures are frequently obtained using an amino acid involved in an amide bond, and one or more aromatic moieties such as the “electron poor” dinitrobenzene which offers p-acceptor interactions or an “electron rich” moiety which offers p-donor interactions. Many Pirkle-type stationary phases have a structure similar to those indicated in Fig. 7.4.1. These structures include a silica support, a propyl handle (the synthesis starting with aminopropyl silica), an amino acid, an amide or an urea linkage, and an electron acceptor group (such as dinitrobenzoyl) or an electron donor group (such as naphthylethyl).

Chromatographic columns in high-performance liquid chromatography

333

Table 7.4.1 The main types of chiral stationary phases. No.

Support

Chiral selector

Examples

1

Silica

Pirkle type

2

Silica

3

Silica

4

Silica

Cellulose and derivatized cellulose Amylose and derivatized amylose Cyclodextrins and cyclofructans

Chirex type, Whelk, Sumichiral, etc. Chiracel, Lux cellulose, etc.

5

Silica

Crown ether

6

Silica

7

Silica

Macrocyclic antibiotics glycopeptides Proteins

8

Organic polymer Organic polymer

9

Ligand exchange

Lux amylose, Chiral AM, Chiral Art Amylose Cyclobond, Frulic, FructoShell Crownpak CR(þ) & CR(), ChiroSil RCA(þ), or () Bioptic, Chirobiotic, certain Chiralpak Chiral BSA, Chirobiotic CSP, Chiralpack HAS, Ultron ES-BSA Chiralpak WH & MA(þ)

Synthetic polymers

Astec P-CAP

Figure 7.4.1 Two structures of Pirkle-type stationary phases showing the handle to silica, the amino acid, the linkage, and the electron acceptor or electron donor groups. Examples of chiral selectors for Pirkle phases include phases bonded to silica containing the following moieties: (1) a propyl handle þ (R)-phenylglycine þ amide linkage þ 3,5dinitrobenzoic acid (shown in Fig. 7.4.1), (2) a propyl handle þ (R)-1-naphtylglycine þ amide

334

2)

3)

4)

5)

6)

Method Development in Analytical HPLC

linkage þ 3,5-dinitrobenzoic acid, (3) a propyl handle þ (S)-valine þ urea linkage þ 3,5dinitroaniline (4) a propyl handle þ (S)-tert-leucine þ urea linkage þ 3,5-dinitroaniline, (5) a propyl handle þ (S)-tert-leucine þ urea linkage þ 1-(a-naphthylethyl) (shown in Fig. 7.4.1), etc. The combination of those moieties is capable of developing both donor and acceptor hydrogen bonds, establishing either p-donor, p-acceptor, or both p-donor and p-acceptor intermolecular bonds. Depending on the attached moieties, Pirkle stationary phases are usually used as NPs (with nonaqueous mobile phase), but in some cases (e.g., the ones having (L)-tartaric acid þ (S)-valine þ (R)-1-(a-naphthyl)ethylamine moieties), they can be used in RP mode. Many chiral phases are made by taking advantage of cellulose chiral properties (e.g., Ref. [27]). Such phases are either coated on a silica surface or bonded to the silica. The columns with coated silica surface are less resilient to some solvents and the utilization of solvents such as tetrahydrofuran (THF), methylene chloride, and chloroform must be used with caution, avoiding their presence in high concentration in the mobile phase. The cellulose OH groups are derivatized with groups such as triacetate, tribenzoate, trisphenyl-carbamate, tris(3,5-dimethylphenyl-carbamate), tris(3-chloro-4-methylpheny lcarbamate), tris(4methylbenzoate), tris(4-chloro-3-methylphenylcarbamate), etc. The utilization of cellulosebased columns requires a normal mobile phase, such as hexane-ethanol, hexane-isopropyl alcohol, etc. The use of these columns with a mobile phase containing some water can also be done, but addition of a salt in the mobile phase, such as a perchlorate, is necessary to prevent the column degradation by the dissolution of the stationary active coating. Chiral phases based on cellulose have a helical structure and can be used for the separation of a variety of analytes. Amylose and derivatized amylose are used similar to cellulose being placed on silica surface for achieving the high surface area of the active phase. Amylose can also assume a helical shape, can form inclusion compounds, and offers chiral types of interactions. Similar with cellulose, it also offers the capability of hydrogen bonding and some hydrophobic interactions. Amylose can be derivatized to generate chiral stationary phases in the form of carbamates such as tris(3,5-dimethylphenyl-carbamate), tris[(S)-a-methylbenzyl carbamate], or tris(chloro-2-methylphenylcarbamate). Moieties of cyclodextrins or of cyclofructans can be attached to silica in a similar way as cellulose. The resulting materials produce cavity-based phases with a large number of chiral centers. Cyclodextrins can be formed from a different number of glucose moieties, with a-cyclodextrin consisting of six glucoses units, b-cyclodextrin consisting of seven glucoses units, and g-cyclodextrin consisting of eight units. Different cyclodextrins can be used for making the active phase of chiral columns, and, for example, Astec Cyclobond I is based on b-cyclodextrin and Astec Cyclobond II is based on g-cyclodextrin. The free OH groups from the cyclodextrin or cyclofructan can be derivatized with acetyl, (S)-hydroxypropyl ether, (S) or (R)-naphthylethylcarbamate, 3,5-dimethyl-phenylcarbamate, p-toluoyl ester, and other groups. A number of commercial suppliers of chiral phases (Astec, Machery-Nagel, Phenomenex, YMC) offer phases containing cyclodextrins or cyclofructans moieties. Crown ethers based chiral stationary phases are also cavity-based phases that can be chemically bound on silica surface. The crown ether achieves asymmetry only after incorporating additional groups such as binaphthyl, biphenanthryl, tartaric acid, several carbohydrates, etc. Macrocyclic-type antibiotics immobilized on silica are also providing chiral separation stationary phase, due to numerous chiral centers present in the molecule and various potential inclusion areas. Among the glycopeptide type antibiotics used for making stationary phases are: Rifamycin(s), Vancomycin, Avoparcin, Ristocetin, glycopeptide A-40,926 (MDL 62,476), and Teicoplanin (e.g., in Chirobiotic columns made by Astec). The structures of

Chromatographic columns in high-performance liquid chromatography

335

Figure 7.4.2 Structures of Teicoplanin A2-1 and that of Vancomycin showing potential interaction areas with the analytes (A, B, C, and D for Teicoplanin and A, B, and C for Vancomycin). Teicoplanin A2-1 and that of Vancomycin are shown in Fig. 7.4.2. The structures indicate areas capable of providing interaction with the analytes. The differences between different macrocyclic glycopeptides can make significant difference in the range of molecules that can be separated by different columns. For example, the column Chirobiotic T which is based on Teicoplanin and the column Chirobiotic V which is based on Vancomycin have basically complementary applications. Chirobiotic T column is mainly used for the separation of amino acids and hydroxy acids, and Chirobiotic V is mainly used for the separation of less polar molecules. Similar to other chiral columns, the macrocyclic antibiotic-type columns can be used in NP but also in polar ionic mode, polar organic mode, and reversed phase. The polar ionic mode (PI or PIM) is also unique for the Chirobiotic phases. 7) Proteins bonded to silica were also used as stationary phases for chiral separations. Proteins contain a large number of chiral centers and can be bonded to silica. Several types of proteins have been used for making such stationary phases including bovine serum albumin (BSA), human serum albumin (HSA), a 1-acid glycoprotein, other glycoproteins, ovomucoid, ovoglycoprotein, avidin, etc. Protein-based stationary phases offer both polar interactions as well as hydrophobic-type interactions. 8) The ligand exchange chiral phases consist of an organic polymer containing chiral groups capable of forming complexes with a transition metal ion (Cu2þ, Ni2þ, Zn2þ). These phases can separate enantiomers of compounds containing functional groups that are capable of forming complexes with the same ions. The separation is performed by adding the metallic ions in the mobile phase carrying the analytes, due to the formation of ternary complexes stationary phase (chiral) þ metal ion þ analyte (chiral). 9) Some chiral organic polymers were also synthesized. An example of such polymer is the one synthesized by FriedeleCrafts alkylation reaction between Boc-3-(4-biphenyl)-L-alanine and 4,40 -bis(chloromethyl)-1,10 -biphenyl. The resulting porous polymer can be used for HPLC separations using hexane-isopropanol as mobile phase under normal-phase mechanism, or methanol-water in case of reversed-phase separations [28]. Other examples of synthetic polymers include phases based on derivatized acrylamide such as poly(trans-1,2cyclohexanediyl-bis-acrylamide). Some organic polymeric chiral phases can be grafted on

336

Method Development in Analytical HPLC

coated on silica in order to achieve a larger surface area (e.g., polymethacrylate diphenyl(pyridyn-2-yl)methyl coated on silica). Also, metal-organic frameworks were synthesized to be used in chiral separations [29], and monoliths [30]. Common particles for packed chiral HPLC columns have 3, 5, 10, and 25 mm particle sizes, although sub-2 mm size particles are also available. Besides that, monolithic silicabased columns are known in enantioseparations, using specific methods of introducing the chiral selector by covalent functionalization or physically surface coating [31].

Characterization of chiral columns The capability of chiral columns to separate enantiomers, indicated as enantioselectivity, is an important parameter for chiral columns. Enantioselectivity is characterized by the same expression as selectivity a (see Eq. 4.2.1), the only difference being that the two compounds X and Y to be separated are the two enantiomers. An important characteristic of chiral chromatographic columns is their efficiency as characterized by the number of theoretical plates number N. Usually, the chiral columns have lower efficiency than similar achiral columns. For example, for a column packed with 5-m m fully porous particles, this efficiency is an order of magnitude lower than for similar achiral columns. An alternative way to improve chiral column efficiency is the use of superficially porous particles [32]. Chiral columns may also show larger asymmetry than achiral columns. Although the commercial chiral columns can produce acceptable peak symmetry, there are situations when the enantioseparations are characterized by a certain peak asymmetry due to peak tailing or fronting, which can be detrimental to quantitative measurements [33]. Another characteristic of chiral columns is related to the mode of their utilization as NP, RP, hydrophilic interaction type (HILIC), or ion exchange type (IE). Many chiral columns are useful only in NP mode (nonaqueous mobile phase), but there are columns which can be used in other modes, or in more than one mode. The capability of using a stationary phase with partially aqueous mobile phase is important for both the solubility of certain analytes in the mobile phase and also related to the detection in MS-ESI mode where some water is usually necessary for obtaining ionization. Other characteristics to consider for chiral columns are similar to those for other column types and are summarized in Table 7.1.1. Such characteristics include column dimensions, particle size of stationary phase, shape of particles, particles mechanical resilience, range of pH stability, chemical stability in different solvents, and lack of bleeding.

Key points • •

Chiral columns contain one or more chiral selectors, and additional functionalities capable of interacting with the analytes. Based on the nature of mobile phase and types of interactions, the stationary phase in chiral columns may act as normal phase (NP), reversed phase (RP), hydrophilic interaction type (HILIC), or ion exchange type (IE).

Chromatographic columns in high-performance liquid chromatography

7.5

337

Columns used in ion exchange chromatography and related techniques

IEC is used mainly for the separation of ions, but some associated techniques such as ion-moderated chromatography are used for the separation of neutral molecules such as sugars or organic acids (not in ionic form). For the columns used in these techniques, some of the general characteristics listed in Table 7.1.1 are also applicable. These include, for example, the column dimensions, theoretical plate number, support for stationary phase, particle shape, etc. Further details regarding the properties of the stationary phases in IEC and related techniques are presented in this Section.

Stationary phases used in ion exchange chromatography and related techniques Stationary phases used in ion chromatography are differentiated in cation exchange type, anion exchange, zwitterionic/amphoteric. Also, several techniques such as ion exclusion chromatography, ligand exchange chromatography, immobilized metal affinity chromatography, and ion-moderated chromatography are practiced on ion exchange type columns (see Section 2.3). A summary of main types of stationary phases used in IEC is given in Table 7.5.1. The support of stationary phases in IEC columns can be silica, ethylene-bridged silica, but the use of organic polymers as a support for the active functionalities is more common in IEC than for other chromatography types. A summary of stationary phases used in IEC is indicated in Table 7.5.2. Some of the stationary phases used in IEC are not common in other types of HPLC. One example is that of columns with an organic polymeric structure where the synthesis of the polymer is performed from monomers already having the functional groups attached. A more special type of column used in IEC is the latex agglomerated one. Table 7.5.1 Types of functionalities in ion exchange phases.

a

No.

Phase character

Strength

Examples of active groups

1 2 3 4 5 6 7

Cation exchange Cation exchange Cation exchange Anion exchange Anion exchange Anion exchange Zwitterionic a

Weak Medium Strong Weak Medium Strong -

8

Amphoteric

eCOO, eC6H4eO, -AsO3H ePO3H, eSO 3 þ eNHþ 3 , e[NH2(CH3)] , [N(C2H5)2(C2H4OH)] e[N(CH3)2(C2H4OH)]þ e[N(CH3)3]þ, e[N(C2H5) (CH3)2]þ  eN(CH3)þ 2 plus e(CH2)neSO3 or e þ eCH(SO3 )e(CH2)neN(CH3)3 Amphoteric groups such as amino acid moiety

a

-

þ

The difference between zwitterionic and amphoteric is that zwitterionic stationary phases contain separated acidic and basic moieties, while amphoteric stationary phases contain unique groups that can act as both acid or base.

338

Method Development in Analytical HPLC

Table 7.5.2 Types of construction for ion exchange stationary phases. No.

Phase support

Type of phase structure

1 2 3 4 5 6 7 8 9

Silica based Silica based Ethylene-bridged silica Synthetic organic polymer Synthetic organic polymer Synthetic organic polymer Natural organic polymer Inorganic silicates Inorganic oxides

Bonded ionic groups Polymer coated Bonded ionic groups Synthesized with ionic functional groups Surface functionalized Latex agglomerated Functionalized (e.g., sepharose, cellulose) Zeolites Alumina, zirconia

This type of phase contains an internal core particle (or support) that has on its surface ionic functionalities. A monolayer of small diameter particles is attached on this support, those particles carrying functional groups consisting of bonded ions that are of the opposite charge from the functionality of the support. The groups of the outer particles have a double role, to attach the small particles to the core, and also to act as the active ion exchanger for the ions in solution. A schematic illustration of an anion exchange latex agglomerated phase is shown in Fig. 7.5.1. The core particles are typically PS-DVB resin of moderate to high cross-linking, with a particle size in the range 5e25 m m. The outer microparticles consist of finely ground resin or monodisperse polymer (latex) with diameters up to 0.1 m m that have very high porosity, and are functionalized to contain an appropriate ion exchange group. This group determines the ion exchange properties of the composite particle. For example, a core exposing SOe 3 groups can be covered with an aminated porous latex generating an anion exchanger. This type of construction allows the active phase to be highly porous with the preservation of overall rigidity of the stationary phase, which would not be possible if the whole particle would be porous. In order to increase the surface containing the active phase, the core may also contain inner channels, although highly crosslinked (e.g., IonPack AS9-HC column from Dionex/Thermo Fisher). In addition to standard bore IEC columns having an internal diameter d around 4 mm, capillary IEC columns are also utilized more frequently than in other types of HPLC. Such columns have a d of 0.4 mm (e.g., Dionex ICS-4000 or ICS-6000 Capillary). For capillary columns, lower flow rates are utilized (U in the range of SO3- +R3N

NR3+

Core SO3- +R3N

NR3+

Figure 7.5.1 Schematic illustration of an anion exchange latex agglomerated phase.

Chromatographic columns in high-performance liquid chromatography

339

0.01e0.03 mL/min) and lower injection volume (e.g., 0.1 m L). Because of the reduced flow rate, such columns will generate backpressures around 5000 psi. The main advantages of capillary columns include the reduced eluent consumption and a faster equilibration time. The sensitivity of detection must be however higher than that for standard columns. Ion-moderated stationary phases typically consist of a cation exchange stationary phase having covalently bonded groups such as eSO 3 , with these groups having as counterion cations such as Kþ, Naþ, Agþ, Ca2þ, Pb2þ (in some cases Hþ). The cation exchange material is usually an organic polymer based on styrene/divinylbenzene, and the specific metal form is maintained attached to the polymer by having the ions at a low concentration in the mobile phase. A special type of ion-moderated stationary phase is utilized in the separation of cis/trans lipids and fatty acids. This stationary phase contains Agþ ions typically in a silica support type column such as ChromSpher 5 Lipids (from Agilent).

Characterization of columns with an ion exchange phase The type of IEC, cationic weak, medium, or strong, anionic weak, medium, or strong, zwitterionic, etc. is a main characteristic that should be considered when selecting an IEC column, cationic phases having groups that dissociate forming positive ions, and anionic phases having groups that dissociate forming negative ions. Zwitterionic phases have both types of ionic groups in their structure. For the strong cationic or anionic type phases, the phase remains in ionized form for a wide pH range, while the medium and in particular the weak type phases are in ionized form only at a specific range of pH of the utilized mobile phase. Regardless of strong, medium, or weak type phase, when ionized the strength of the bond formed with the analytes by the stationary phase is not stronger for strong type phases, or weaker for weak type phases. Zwitterionic phases act as weak type ionic or may act only as HILIC-type phases, in function of the mobile phase composition. Ion exchange phases are also characterized by their ion capacity Icap . The ion capacity of the ion-exchanger is determined by the number of functional groups per unit weight of the stationary phase. Typical Icap values in IEC columns are low compared to other ion exchange materials, being in the range of 10e400 mEq/g, although some phases may have higher values. The columns may have a wide range of Icap •wst (where wst is the weight of stationary phase), between 3 and 600 m Eq/column depending on the stationary phase structure and column dimensions. Based on the ion capacity, the columns are classified as having high, moderate, and low capacity. The resilience of the stationary phase to a specific mobile phase is also an important characteristic. The stationary phases for anion exchange type, for example, are specifically designed to be used with a KOH or NaOH in the mobile phase, or with Na2CO3/ NaHCO3 in the mobile phase. For the columns designed to be used in strong basic media, the resilience to a high pH is an important column characteristic.

340

Method Development in Analytical HPLC

The compatibility of the stationary phase with a specific organic solvent should also be considered when using IEC columns in particular when they are based on organic polymers. The presence of an organic solvent in the mobile phase in IEC may also affect selectivity by changing the retention characteristics of the column. Another property of the IEC columns is related to their potential hydrophobic character, which should be as low as possible. In many IEC separations besides the inorganic ions other molecules are present, and the hydrophobic character of the analyte should not interfere with the separation. However, the ionic groups of the stationary phase are typically bonded to an organic fragment that is either connected to silica or to an organic polymer. Organic polymers can have a polar character such as those based on polyvinyl alcohol (PVA), polymethylmethacrylate, or copolymer of 2hydroxyethylmethacrylate with ethylenedimethacrylate. However, hydrophobic polymeric bases such as polystyrene-divinylbenzene (PS-DVB) and poly(ethylvinylbenzene-divinylbenzene) are also very common. Hydrophobic character of the polymeric base influences considerably the properties of the chromatographic column and this part of the ionic stationary phase may influence the separation of the analytes by undesired hydrophobic interactions [34]. Phases with low or ultralow hydrophobicity were manufactured, and this property is typically indicated for commercially available IEC columns. Besides ionic and hydrophobic interactions, some phases develop polar interactions with the analytes. The polarity of the stationary phase is not usually considered as detrimental. Some examples of polymeric columns are given in Table 7.5.3 for different cation exchange columns available on the market and some examples of polymeric anion exchange-type columns are given in Tables 7.5.4. Another important characteristic of ion exchange columns is the equilibration time to a specific mobile phase composition. Different for example from RPHPLC type columns, it is common that ion exchange columns require longer times for equilibration. The equilibration time necessary for a column is related to its loading capacity and usually the columns with a higher Icap require longer equilibration times. Other common physical characteristics applicable to IEC-type columns include the density of charged functional groups on the stationary phase surface, pore size, surface area, as well as other general properties indicated in Table 7.1.1. The range of temperatures of operation is another characteristic to be considered for the IEC phases. A special type of columns not used for the separation of ions but containing strong cation exchange resin in different metallic cation forms are the ion-moderated ones. This type of columns is represented, for example, by the Rezex type (from Phenomenex) made from 8 mm particles of sulfonated-divinylbenzene resin containing: Agþ (for RAM-Carbohydrate and RSO-Oligosaccharides), Ca2þ (for RCM-Monosaccharide and RCU-USP Sugar Alcohols), Hþ (for RHM-Monosaccharide and RFQ-Fast Acid), Kþ (for RKP-Potassium), Naþ (for RNM-Carbohydrate and RNO- Oligosaccharides), and Pb2þ (for RPM-Monosaccharide).

Table 7.5.3 Some cation exchange polymeric columns and their characteristics. Length x i.d., mm

Particle size/ pore size

Capacity (mEq. gL1)

CM-825 Diaion Diaion

Weak Strong Weak

75 x 8 Various Various

8 mm/5000 Å

0.4

ES-502C 7C IC YS-50 ICT-521 IonPac CS-10 IonPac CS-11 IonPac CS-12, 14, 16, 17, 18, 19 IonPac CS-12A

Weak Weak Strong Strong Strong Weak Medium

100 x 7.5 125 x 4.6 150 x 4.6 250 x 4 250 x 2 250 x 4 or 250 x 2 Various

9 mm/2000 Å

0.55

8.5 mm 8 mm 8 mm/60 Å 5; 8 mm

0.08 0.035 0.7; 2.8 0.7; 0.94; 2.8

Latex Latex Latex Latex

IonPac CS-15

Weak

250 x 4 or x 2

8.5 mm

0.7; 2.8

Latex

LCA K02 MAbPAC SCX-10 PL-SCX ProPac SCX ProPac WCX PRP-X100, X200, X400, etc. Shimpack IC-C1 SP-825 TSKgel BioAssist S TSKgel CM-STAT TSKgel OApak-A TSKgel SP-5PW TSKgel SP-STAT YS-50

Strong Strong Strong Strong Weak Strong Strong Strong Strong Weak Weak Strong Strong Weak

125 x 4.6 Various Various 250 x 4 250 x 4 250 x 4.6 150 x 4 75 x 8 Various Various Various Various Various 125 x 4.6

5 mm 3; 5; 10 mm 10; 30 mm/1000, 4000 Å 10 mm 10 mm 100 Å 10 mm 8 mm/5000 Å 7; 13 mm/1300 Å 7; 10 mm 5 mm 10, 13, 20 mm/1000 Å 7; 10 mm 5 mm

0.4

Technol

Highly porous

PMA, polymethylacrylate; PS-DVB, polystyrene-divinylbenzene, PVA, polyvinyl alcohol.

Latex Latex Latex 0.035e2.5 0.4 0.1 0.1 1.5 >0.1 0.023

Type of phasea Carboxymethyls PS-DVB-SO3H Acrylic acidmethacrylate PVA carboxymethyl PVA PS-DVB-SO3H PS-DVB-SO3H PS-DVB-SO3H PS-DVB-COOH PS-DVB-COOH þ PSDVB-PO3H PS-DVB-COOH, -PO3H/crown ether PMA PS-DVB-SO3H PS-DVB-SO3H PS-DVB-SO3H PS-DVB-COOH PS-DVB-SO3H PS-DVB-SO3H eC3H6eSO3H eC3H6eSO3H Carboxymethyl Methacrylate eC3H6eSO3H PVA-COOH

341

Type

Chromatographic columns in high-performance liquid chromatography

a

Column

Table 7.5.4 Some anion exchange polymeric columns and their characteristics. 342

Length x i.d., mm

Particle size (mm) 7 mm 9 mm

PMA gel- NRþ 3 PS-DVB-amine

2.5 mm; 10 mm 6 mm

PMA gel- NRþ 3

Strong

Various 250 x 4.6 100 x 7.5 35 x 4.6 75 x 7.5 250 x 2.0 250 x 4.0 Various

Various

Strong

75 x 8

5 mm

Strong Strong Strong

75 x 7.5 50 x 4.6 100 x 4.6

10 mm 5 mm 12, 5 mm

Column

Type

Allsep AN1; AN300

Strong Strong

BioSuite DEAE, Q-PEEK

Strong

CarboPac SA10

Strong

CarboPac PA1, PA10, PA20, PA100, PA200, MA1 Cosmogel QA Cosmogel DEAE Diaion Discovery BIO PolyMA-WAX IC I-524A, IC NI-424

Strong Strong

75 x 4.6 various

10 mm 3.5; 5; 9 mm

Weak Medium

12

Ion Swift Max 200 IonPac AS10, AS11, AS11-HC, AS12, AS17

Medium Strong

IonPac 15, 16, 20, etc.

Various

IonPac AS14, 14, 12, 24A

Medium

75 x 8 250 x 1 250 x 0.25 250 x 0.25 250 x 2 250 x 4 250 x 0.4 etc. various

8.5 mm

Latex nano beads Pellicular, nanoporous

Tertiary amine Tertiary amine PMA gel- NRþ 3 PMA gel- NH(C2H5)þ Styrenic/acrylic amine PMA gel- NRþ 3 Polyhydroxy-methacrylate eNRþ 3 PMA gel- NRþ 3 PVA- NRþ 3 PHMA-NRþ 3 PS-DVB-alkanoleNRþ 3

Monolithic Latex

PS-DVB-alkanoleNRþ 3 Tertiary amine PS-DVB-NRþ 3 and alkanol

9 mm 7; 9 mm

Type of phase

Latex

PS-DVB-alkanoleNRþ 3

Method Development in Analytical HPLC

IC-Pak anion IC SI-35 4D; IC SI-50 4E; IC SI-52 4E; IC SI-90 4E; IC SI-91 4C IEC DEAE-825 Ion Swift Max 100 monolithic

Technol.

Strong

various

7.5; 13 mm

IonPac AS5, AS22, AS23 IonPac AS7, AS9 HC

Medium Strong

15 mm 10 mm

Latex Latex

IonPac AS9 SC, AS4A-SC

Medium

9; 13 mm

Latex

PS-DVB-NRþ 3 and alkanol

LCA A01 MetroSep anion Dual 2, 3 Metrosep A Supp 1, 3, 10,15

Strong Strong Strong

Metrosep A Supp 5 7 Nucleogel SAX ProPac SAX Protein-Pak Q 8HR PRP X100

Strong Strong Strong Strong Strong

PRP X500

Strong

250 x 4 250 x 2 250 x 4 250 x 2 250 x 4 200 x 4.0 75 x 4.6 50 x 4.6 250 x 4.6 various various 250 x 4 250 x 4 125 x 4.0 250 x 4 150 x 4.6

PolyethylvinylbenzeneDVB-NRþ 3 PS-DVB-alkanol eNRþ 3 PS-DVB-NRþ 3

RCX-30 Shim-Pack WAX-1; 2 Shodex IEC QA-825 Star-ion A300 A300 HC Super-Sep IC anion TSKgel BioAssist Q, etc. Zodiac IC anion

Strong Strong Strong Strong Strong Strong Strong

150 x 4.6 50 x 4 75 x 8 100 x 4.6 250 x 4.6 50 x 4.6 various

8; 6 mm 7; 9 mm 5 mm

10 mm 5 mm 7 mm 3; 5 mm 12 mm 7 mm 9 mm 5; 10; 13 mm 5; 10 mm

PS-DVB-amine PMA gel- NRþ 3 PS-DVB-N(CH3)þ 3 PVA gel- NRþ 3 PMA gel- NRþ 3 -NRþ 3 PMA gel- NRþ 3 PS-DVB-N(CH3)þ 3 Poly(methacrylamidopropyl)-N(CH3)þ 3 PS-DVB-N(CH3)þ 3 -NRþ 3 PMA gel- NRþ 3 PS-DVB-NRþ 3 PVA gel- NRþ 3 PMA gel- NRþ 3 PMA gel-NRþ 3

Chromatographic columns in high-performance liquid chromatography

IonPac AS18

343

344

Method Development in Analytical HPLC

Key points • • •

The columns used in IEC have the functionalities from the active phase in ionic form (anionic, cationic, etc.). Weak-type ion exchange columns have the functional group ionized only for a specific pH range of the mobile phase, while the strong-type ion exchange phases have the group in ionized form for a broad pH range. Both silica and polymeric supports are common for IEC columns, but polymeric supports are more frequently utilized compared to other types of HPLC.

7.6

Chromatographic columns used in size exclusion

Stationary phases in size exclusion HPLC (SEC) are designed for the separation of polymers and for the estimation of their molecular weight (or more correctly of their hydrodynamic volume). Depending on the analyte polymer solubility, size exclusion phases are designed to be used in gel permeation chromatography type (GPC) when the mobile phase is an organic solvent, or gel filtration chromatography type (GFC) when the mobile phase is an aqueous solvent or a polar solvent that may also contain water. The nature of stationary phases for the two types of SEC separations are different.

Stationary phases and columns for SEC Two main materials were basically used for the construction of stationary phases for SEC. These materials are silica, and organic synthetic polymers. Organic polymeric stationary phases are more common than silica-based columns, because they can be made with a wider range of pore sizes (between 60  A to 1,000,000  A) compared to the silica-based ones. Porous glass phases are also used as separation media in SEC. The phases used for GPC have a hydrophobic character, while those used for GFC have a hydrophilic character. Silica-based phases include bare silica and also derivatized silica that are similar to the construction of other bonded phases on silica. The main advantage of silica and silica-based phases in SEC is their resilience to higher backpressures in the chromatographic column, and the lack of swelling. Besides bare silica, bonded phase silicabased materials with large and controlled pore dimensions can be made to have on silica surface groups such as diol, diol on Zr-clad silica, glycol ether, polyether, amide, etc. These phases have frequently a propyl anchor group to the silica material. The main reason for using derivatized silica for SEC separations is to diminish as much as possible enthalpic interactions of the phase with the analytes and in some cases to make the column adequate for GPC utilization by bonding, for example, C18 groups on the silica surface. Depending on the nature of the functionalities attached to the silica, specific properties can be obtained, for example, the use of diol phases offers controlled polarity and less enthalpic interactions in the separation of proteins. Many polymer-based SEC phases used in GPC are made from PS-DVB. This polymer is typically prepared by suspension polymerization and depending on the proportion of cross-linking reagent (DVB), soft gels are obtained when they contain

Chromatographic columns in high-performance liquid chromatography

345

2%e12% DVB, or rigid polymers when they have more than 20% DVB. PS-DVB polymers are mainly used for GPC-type separations. Phases used in GFC are usually based on dextran or agarose (soft gels) and more rigid polymers are based on hydroxylated poly(methyl methacrylate) (HPMMA), hydroxylated poly(methacrylate) (HPMA), or PVA copolymers.

Characterization of columns and stationary phases used in SEC Besides the hydrophobic character (necessary for GPC separations) or hydrophilic character (necessary for GFC separation), the main characteristic of stationary phases used in SEC is their pore size (or more correctly the range of their pore size) that determines the range of molecular weights of the analytes that can be separated on the columns. For silica-based columns, a desired porosity is typically obtained using controlled hydrolysis of an alkoxysilane. The porosity of the organic polymeric materials is controlled by performing the polymerization in the presence of a nonpolymerizable compound that is subsequently eliminated (e.g., with a solvent) from the porous polymer. The pore size can vary from around 60  A to 1,000,000  A, which corresponds to the separation of polymers in different ranges of Mw. Some phases are designed to separate oligomers with Mw in the range of 1000e2000 Da, and other with larger pores to separate macromolecules as large as having 20,10þ6 Da. The range of porosity of the stationary phase can cover a larger range of pore sizes, or a narrow range of pore sizes. The narrow range of pore sizes phases have a separation capacity concentrated in a limited Mw range of the analytes but a better resolution. In practice, SEC columns of different pore sizes are connected in series of two to four columns to provide a wider molecular weight separation range and keep good resolution [35]. This procedure of multiple columns is not very convenient mainly because it creates high backpressure in the first column of a series. For this reason, unique columns in which the pore size distribution of the stationary phase is broadened by blending together two or more phases were manufactured. Mixed pore packings obtained by blending together several selected pore size materials can be made such that the column exhibits a linear calibration for analytes in a wider range of Mw. The evaluation of pore size of stationary phases in SEC is usually performed using calibrants such as sets of polystyrenes with known Mw. In addition to the pore size of the stationary phase for SEC columns, other properties are also important. Among these can be indicated the resilience of the stationary phase to backpressure characterized by the maximum operating pressure for the SEC column. Organic polymers can collapse if the backpressure in the column exceeds a specific limit (usually around 650 psi or 45 bar). Other properties of importance include the dimension of particles of stationary phase dp with lower dp offering better resolution, particle shape, and column efficiency, the stability of the phase toward different solvents, the stability to extreme pH of the mobile phase which is better for the organic polymers than for the silica-based ones, the stability to increased temperature, etc. The stability to the increased temperature is important, for example, when the analytes to be separated (usually polymeric compounds) are not soluble in the mobile phase except at a higher temperature (e.g., up to 120 C), etc. The lack of

346

Method Development in Analytical HPLC

Table 7.6.1 Characteristics of SEC columns/packing materials.

Property

Importance

Common values

Range

Pore size, and pore volume Column efficiency ðNÞ Particle size

Range of Mw

A 50e105 

50e105 Å

Separation efficiency ðNÞ, resolution Rs

18,000

7000 e25,000

Separation efficiency, resolution Rs, max. backpressure, flow rate Separation efficiency, N resolution Rs

5, 7 mm

4e20 mm

Spherical 500e1000 psi 95%

Irregular sphere 600e2000 psi Variable

Variable

Variable

Up to130 C 2e4 mg 3e20 min

20e160 C

Particle shape Working pressure Column activity Support type and phase type Support type Load capacity Elution time

Column maximum back pressure Inertness, utility for Mw evaluation, recovery of the analyte Maximum working pressure, range of pH, chemical stability, use of solvents Thermal stability, max. temperature of utilization Amount of sample to be loaded on column Time necessary to elute a Mw z 10þ5 Da

2e4 mg Variable

enthalpic interactions of the phase with different classes of analytes is also an important characteristic. In some applications, the SEC columns are used with the purpose of separation and collection of a specific molecular weight fraction from the sample (e.g., in protein separations). The capability of recovering of a certain sample fraction and preservation of its biological activity is another useful property of a column in SEC [36]. Some of these characteristics are summarized in Table 7.6.1.

Key points • • •

The columns in size exclusion can be dedicated to gel permeation (CPC) or gel filtration (GFC). Organic synthetic polymers are frequently used as stationary phase in SEC. Phase porosity must be selected to fit a specific molecular weight range of the polymers to be separated from the sample.

7.7

Other types of chromatographic columns

Several other types of stationary phases can be placed in HPLC columns. These include, for example, phases of immunoaffinity or biomimetic type (e.g., Ref. [7])

Chromatographic columns in high-performance liquid chromatography

347

that have applications in bioanalysis and for the characterization of behavior of various solutes toward complex structures that mimic natural systems like membranes. Biomimetic phospholipid membrane chromatography (BPMC), for example, offers a simple approach of evaluating drug-membrane interactions applied for drug pharmacokinetics, pharmacodynamics, and toxicity studies [37]. For these columns, solid supports including natural polysaccharides (agarose, crosslinked agarose or sepharose, cellulose, cross-linked dextran or sephacryl, methacrylates, polyacrylamide and copolymers of polyacrylamide with other polyvinyl polymers, as well as silica gel) can be utilized. These supports usually have pore sizes in the range between 300 and 500  A (larger than supports typically used in RP-HPLC or HILIC that have pore size around 100  A). The active phase for columns used in immunoaffinity can be very diverse. They may consist of antibody materials or antibody-related moieties, enzymes, or compounds such as heparin, lectins, nucleotides, etc. Also, phospholipids, intact liposomes or fragments of cell membrane can be immobilized on the surface of support by physical adsorption, covalent binding, steric trapping, and avidin-biotin binding. The attachment to the solid support by immobilization is taking advantage of the presence of specific groups on the material used as active phase (e.g., amino, carbonyl, carboxyl) and after the solid support is activated using reagents such as N,N0 -carbonyldiimidazole, cyanogen bromide, N-hydroxysuccinimide, tresyl chloride (2,2,2trifluoroethane-1-sulfonyl chloride), or tosyl chloride (p-toluenesulfonyl chloride). In case of silica surface, this can be activated, for example, using initially a derivatization that adds a propyl-amino group on its surface followed by reaction with one aldehyde group of glutaraldehyde. The second aldehyde group of glutaraldehyde can react with the component generating the active phase. Columns of another type that started to be more frequently utilized in the recent years are those containing a mixed-mode type stationary phase. The stationary phase in mixed-mode HPLC contains different moieties each acting in the separation by a different mechanism. To a certain extent, the stationary phases for virtually all HPLC types contain more than one functionality, but except one of them, they can be considered only as having secondary importance compared to a major column characteristic. For example, in RP-HPLC, the separation is mainly based on hydrophobic interactions, but polar interactions are also produced by the remaining silanol groups of silica on which the hydrophobic chains such as C18 or C8 are bonded. Also, in RPHPLC, the polar embedded groups used for certain hydrophobic type columns are responsible for polar interactions, although the overall property of the stationary phase remains of hydrophobic type. In other HPLC types such as in HILIC, the polar functional groups are usually connected to the solid support (e.g., silica) through hydrocarbon “handles” that provide sites for hydrophobic interactions. Similar situation is encountered in IEC where the phase contains hydrophobic moieties in addition to the ionic groups characteristic for IEC. Truly mixed-mode stationary phases should be considered only those where each of the different functionalities have major effect on the separation. For example, the column Acclaim Mixed-Mode WCX-1 contains a stationary phase that incorporates both hydrophobic (RP) and weak cation-exchange

348

Method Development in Analytical HPLC

(WCX) groups and can be considered of mixed-mode type. In such columns, the active component of the phases has intentional mixed functionalities in which the main interactions consist of two (or more) types (hydrophobic, polar, or ionic). The mixed-mode columns can be reversed-phase/hydrophilic (RP-HILIC), reversed-phase/ion-exchange (RP-IE), hydrophilic/ion-exchange (HILIC-IE), etc. (e.g., Ref. [38]). The support for the stationary phases can be silica-based, polymer-based, silica-ethylene bridged (such as Atlantis Premier BEH C18 AX column), and monolithic. The stationary phase can be obtained using several procedures including: (1) mixed particles of two types of phases (e.g., hydrophobic and ion exchange), (2) mixed ligands attached on a solid support, each having different types of interaction with the analyte, (3) active phase containing specific embedded functionalities in a hydrophobic chain, (4) active phase containing a hydrophobic handle and terminal functional groups, (5) active phase containing a hydrophobic handle, embedded functionalities in a hydrophobic chain, and terminal functional groups (e.g., Refs. [39,40]). From these categories, the presence of an embedded polar group in a long hydrocarbon chain does not have enough contribution to polarity to generate a mixed-mode phase as long as the stationary phase is acting as a usual RP-type phase. However, an ionic embedded group can change the properties of the stationary phase such that it can act through both ionic interactions and hydrophobic interactions. Also, polar groups at the end of a hydrophobic chain can lead to a mixed-mode type phase. One such example is a stationary phase made using poly-Llysine grafted on aminopropyl silica [41] shown in Fig. 7.7.1. Many other types of mixed-mode phases were synthesized based on ionic liquids moieties bound, for example, on silica such as N,N-diallyl-N-methyl-d-glucaminium bromide bonded on 3-mercaptopropyl-modified silica generating a HILIC-anion exchange type phase [42]. Besides the double mode mechanism, there is the possibility of combining three different mechanisms into one HPLC column, the result of such combinations being, for example, an RP HILIC-IE column, or an RP with two IC mechanisms, one for anion exchange and another for cation exchange. However, the three-mode mechanism requires a much more complex procedure for the synthesis of stationary phase [43]. In addition to typical mixed-mode phases that combine common chromatographic techniques such as RP, HILIC, IEC, stationary phases producing specific interactions

Figure 7.7.1 Structures of two mixed-mode stationary phases, poly-L-lysine grafted on aminopropyl silica of RP-HILIC type, and glucaminium bromide bonded on 3-mercaptopropyl silica of HILIC-IE type.

Chromatographic columns in high-performance liquid chromatography

349

different from each other and with major contribution to the separation are also utilized for specific applications. In this category could be included the chiral phases which are not usually treated as mixed-mode and were presented in Section 7.4. Special utilizations also have phases involving separation that combine inclusion complexation (e.g., cryptands formation) and hydrophobic or hydrophilic interactions, etc. (e.g., Ref. [44]).

Key points • •

Special types of columns based on immunoaffinity or having mixed-mode active phases are also used in certain HPLC separations. Immunoaffinity chromatography has applications mainly in bioanalysis.

7.8

Care for the chromatographic column

The repeated use of a chromatographic column may lead to the deterioration of its performance (e.g., Ref. [45]). The lifetime of the column is influenced by several factors related to the injected samples, and experimental conditions such as mobile phase composition. Complex samples containing, for example, proteins, phospholipids, and particulate matter diminish the lifetime, and certain experimental conditions described below have also effects on column’s performances in time. The column deterioration is usually noticed through two effects, one being the reduction of the separation capability showing reduction in the retention times and peak broadening (increase in the theoretical plate height H) and by the increase in the column backpressure. An illustration of the reduction of the increase in the H and its dependence of the linear flow rate u is shown in Fig. 7.8.1 for a Zorbax Eclipse XDB-C8 column (3 mm particle size, 150 mm length, 4.6 mm i.d) after being used for 800 injections of biological samples [7].

Figure 7.8.1 Change in theoretical plate height H for a Zorbax Eclipse XDB-C8 column after 800 injections of biological samples [7].

22 21 after 800 injections

H (µm)

20 19 18 17

new column

16 15 2

4

6

8

10

12

u (cm/min)

14

16

18

20

350

Method Development in Analytical HPLC

Column protection Column protection should consider some characteristics of the mobile phase, and some of the injected sample. Regarding the mobile phase utilized with a specific column, the recommendation of the column manufacturer must be considered. The column should not be used at pH values of the mobile phase outside the recommended range of stability. The use of more basic or more acidic mobile phase that recommended produces deterioration of the stationary phase support and hydrolysis of the active phase. This is important in particular for silica-based stationary phases that may be stable only in the range of pH between 2.5 and 7.5. Injection of aggressive sample diluents (strong acids, or bases in high concentrations) should be also avoided. The solvents used in the mobile phase also must be those recommended, the mixtures of water and methanol or water and acetonitrile being common for most columns. However, for some chiraltype columns intended to be used in NP-type separation, water must be avoided. Also, for some columns used in ion-moderated chromatography, the mobile phase must always contain the ion recommended for column utilization. In case of hydrophobic columns, the use of 100% water as mobile phase can produce phase dewetting which ruins the column. For this reason, only columns indicated as capable of using 100% water can be used with such mobile phase. When columns are used for gradient separations, it must be assured that buffers and/or additives do not precipitate when an organic component in the mobile phase increases in concentration. Solubility of inorganic salts may decrease when the water proportion is diminished and salts precipitation in the column can make it unusable. Another aspect regarding the mobile phase is related to the accepted backpressure for the column. The use of higher backpressures than those recommended by the manufacturer may lead to phase collapse, and a void space is formed at the head of the column, leading to significant loss of column efficiency. Silica-based columns are usually resilient to high pressures, but for organic polymer-type columns, the maximum accepted backpressure is usually lower (400e500 psi) and this value should not be exceeded. The recommended maximum temperature for the column also should not be exceeded. The direction of the flow in the column indicated by the manufacturer should be used. Regarding the injected sample, this should be free of particles, and special filters (frits) and guard columns and cartridges are used to protect the analytical columns. Also, some sample constituents can be accumulated at the head of the analytical column not being eluted even when using very strong eluents. For the retention of such compounds, it is recommended to use the guard columns which can be replaced periodically and are much cheaper than the column. Guard columns or cartridges are very short columns, packed with a stationary phase having similar composition as the analytical column or slightly less retentive bonded-phase as the analytical column. The guard column should have a very small dead volume such that it should not affect the column performance. The guard column retains the noneluting contaminants. Also, in time, a slight solubilization of the stationary phase of the column may take place and the guard column provides saturation of the mobile phase with silica

Chromatographic columns in high-performance liquid chromatography

351

which will be “bleeding” into the mobile phase from the guard column instead of being from the analytical column. Combination of high separation temperature and extreme pH values of the mobile phase accelerate the bleeding process which may also affect detection [46].

Column cleaning and storage After a certain period of use, it may be useful to clean the column from any remaining sample components which were not eluted during the column use. For example, for RP columns, a set of solvents covering a wide range of polarity can be used for cleaning, starting with 95:5 water/acetonitrile v/v, followed by THF followed by 95:5 acetonitrile/water v/v. Cleaning of columns used for protein separations may be washed with 0.1% trifluoroacetic acid (TFA) in water followed by 0.1% TFA in acetonitrile/isopropanol 1/2 v/v, followed by rinsing with a common mobile phase (columns should not be stored in THF). Other specific cleaning procedures may be recommended by the column manufacturer. In general, it is not recommended to use for cleaning a flow in the opposite direction than recommended. Storing of the chromatographic column is recommended to be performed by having the column loaded with the solvent used in the new purchased column. For RP-HPLC columns, these can be stored in methanol, methanol/water mixtures, or acetonitrile/water mixtures.

Key points • •

Columns must be protected by using appropriate mobile phase, and by adding a guard column. Columns must be stored loaded with the same or similar solvents as present when the new column was purchased.

References [1] D. Neue, HPLC Columns: Theory, Technology, and Practice, John Wiley & Sons, Hoboken, New Jersey, 1997. [2] F. Gritti, I. Leonardis, J. Abia, G. Guiochon, Physical properties and structure of fine coreeshell particles used as packing materials for chromatography: relationships between particle characteristics and column performance, J. Chromatogr. A 1217 (2010) 3819e3843. [3] G. Rozing, Micropillar array columns for advancing nanoflow HPLC, Microchem. J. 170 (2021) 106629. [4] L.M. Blumberg, The best structures of LC columnse a theoretical perspective, J. Chromatogr. A 1721 (2024) 464848. [5] W. Malsche, H. Eghbali, D. Clicq, J. Vangelooven, H. Gardeniers, G. Desmet, Pressuredriven reverse-phase liquid chromatography separations in ordered nonporous pillar array columns, Anal. Chem. 79 (2007) 5915e5926. [6] A. Berthod, Silica: backbone material of liquid chromatographic column packings, J. Chromatogr. A 549 (1991) 1e28. [7] S. Moldoveanu, V. David, Essentials in Modern HPLC Separations, second ed., Elsevier, Amsterdam, 2022.

352

Method Development in Analytical HPLC

[8] L.T. Zhuravlev, V.V. Potapov, Density of silanol groups on the surface of silica precipitated from a hydrothermal solution, Russian J. Phys. Chem. 80 (2006) 1119e1128. [9] P.C. Iraneta, H.B. Hewitson, D. Morrison, K.J. Fountain, Practical Applications of Charged Surface Hybrid (CSH) Technology, Waters Application Note, 2010. https://www. gimitec.com/file/720003720en.pdf. [10] B. Bidlingmeyer, C.C. Chan, P. Fastino, R. Henry, P. Koerner, A.T. Maule, M.R.C. Marques, U. Neue, L. Ng, H. Pappa, L. Sander, C. Santasania, L. Snyder, T. Woznyak, HPLC column classification, Pharmacopeial Forum 31 (2005) 637e645. [11] J.J. Kirkland, Development of some stationary phases for reversed-phase HPLC, J. Chromatogr. A 1060 (2004) 9e21. [12] S.C. Moldoveanu, V. David, Progress in technology of the chromatographic columns in HPLC, in: Analytical Liquid Chromatography e New Perspectives, Intechopen, London, 2022. [13] N.S. Wilson, M.D. Nelson, J.W. Dolan, L.R. Snyder, R.G. Wolcott, P.W. Carr, Column selectivity in reversed-phase liquid chromatography: I. A general quantitative relationship, J. Chromatogr. A 961 (2002) 171e193. [14] N.S. Wilson, M.D. Nelson, J.W. Dolan, L.R. Snyder, P.W. Carr, Column selectivity in reversed phase liquid chromatography: II. Effects of a change in conditions, J. Chromatogr. A 961 (2002) 195e215. [15] N.S. Wilson, M.D. Nelson, J.W. Dolan, L.R. Snyder, P.W. Carr, L.C. Sander, Column selectivity in reversed-phase liquid chromatography: III. The physico-chemical basis of selectivity, J. Chromatogr. A 961 (2002) 217e236. [16] J.J. Gilroy, J.W. Dolan, L.R. Snyder, Column selectivity in reversed-phase liquid chromatography: IV. Type-B alkyl-silica columns, J. Chromatogr. A 1000 (2003) 757e778. [17] J.W. Dolan, L.R. Snyder, The Hydrophobic-subtraction model for reversed phase liquid chromatography: a reprise, LCGC North Am. 34 (9) (2016) 730e741. [18] J.W. Dolan, A. Maule, L. Wrisley, C.C. Chan, M. Angod, C. Lunte, R. Krisko, J. Winston, B. Homeierand, D.M. McCalley, L.R. Snyder, Choosing an equivalent replacement column for a reversed-phase liquid chromatographic assay procedure, J. Chromatogr. A 1057 (2004) 59e74. [19] http://apps.usp.org/app/USPNF/columnsDB.html. [20] D.R. Stoll, T.A. Dahlseid, S.C. Rutan, T. Taylor, J.M. Serret, Improvements in the predictive accuracy of the hydrophobic subtraction model of reversed-phase selectivity, J. Chromatogr. A 1636 (2021) 461682. [21] https://www.yumpu.com/en/document/read/40575699/comparison-guide-to-c18-reversedphase-hplc-columns. [22] L. Yu, X. Li, Z. Guo, X. Zhang, X. Liang, Hydrophilic interaction chromatography based enrichment of glycopeptides by using click maltose: a matrix with high selectivity and glycosylation heterogeneity coverage, Chemistry 15 (2009) 12618e12626. [23] T.H. Walter, C. Boissel, J.A. Field, N.L. Lawrence, Further evaluation of the base stability of hydrophilic interaction chromatography columns packed with silica or ethylene-bridged hybrid particles, Separations 10 (2023) 175. [24] T.E. Beesley, Review of chiral stationary phase development and chiral applications, LCGC Eur. 25 (2011) 270e276. [25] T.E. Beesley, R.P.W. Scott, Chiral Chromatography, Wiley, Hoboken, 1999. [26] A.E. Ibrahim, N.A. El Gohary, D. Aboushady, L. Samir, S.E.A. Karim, M. Herz, B.I. Salman, A. Al-Harrasi, R. Hanafi, S. El Deeb, Recent advances in chiral selectors immobilization and chiral mobile phase additives in liquid chromatographic enantioseparations: a review, J. Chromatogr. A 1706 (2023) 464214.

Chromatographic columns in high-performance liquid chromatography

353

[27] S.C. Moldoveanu, Interconversion of nicotine enantiomers during heating and implications for smoke from combustible cigarettes, heated tobacco products, and electronic cigarettes, Chirality 34 (2022) 667e677. [28] Y.-P. Yang, J.-K. Chen, P. Guo, Y.-R. Lu, C.-F. Liu, B.-J. Wang, J.-H. Zhang, S.-M. Xie, L.-M. Yuan, A chiral porous organic polymer COP-1 used as stationary phase for HPLC enantioseparation under normal-phase and reversed-phase conditions, Mikrochim. Acta 189 (2022). [29] C. Liu, K. Quan, J. Chen, x. Shi, H. Qiu, Chiral metal-organic frameworks and their composites as stationary phases for liquid chromatography chiral separation: a minireview, J. Chromatogr. A 1700 (2023) 464032. [30] S. Bayındır, C. Aydogan, A. Denizli, Preparation of chiral monoliths with new modulation of the monolith surface chemistry for the enantioseparation of chiral drugs by nano-liquid chromatography, J. Chromatogr. A 1713 (2024) 464573. [31] M. Asmari, X. Wang, N. Casado, M. Piponski, S. Kovalenko, L. Logoyda, R.S. Hanafi, S. El Deeb, Chiral monolithic silica-based HPLC columns for enantiomeric separation and determination: functionalization of chiral selector and recognition of selector - selectand interaction, Molecules 26 (2021) 5241. [32] Z.S. Breitbac, High efficiency chiral separations in HPLC and SFC, LCGC North Am. 36 (2) (2018) 137e139. [33] T.T. Handlovic, M. Farooq Wahab, D.W. Armstrong, Symmetrization of peaks in chiral chromatography with an area-invariant resolution enhancement method, Anal. Chem. 94 (2022) 16638e16646. [34] P.N. Nesterenko, E.P. Nesterenko, Hydrophobicity of polymer-based anion-exchange columns for ion chromatography, Helyion 7 (2021) e07290. [35] F. Gritti, Theoretical performance of multiple size-exclusion chromatography columns connected in series, J. Chromatogr. A 1634 (2020) 461673. [36] G.D. Saunders, H.G. Barth, Fundamentals and properties of size-exclusion chromatography packings and columns, LCGC Suppl. 30 (4) (2012) 46e53. [37] J. Wang, J. Guo, D. Xu, L. He, J.-H. Qu, Q. Wang, J. Crommen, Z. Jiang, Development of biomimetic phospholipid membrane chromatography for drug discovery: a comprehensive review, TrAC - Trends Anal. Chem. 171 (2024) 117512. [38] L. Zhang, Q. Dai, X. Qiao, C. Yu, X. Qin, H. Yan, Mixed-mode chromatographic stationary phases: recent advancements and its applications for high-performance liquid chromatography, TrAC - Trends Anal. Chem. 82 (2016) 143e163. [39] N.V.T. Nguyen, Perspective chapter: mixed-mode chromatography, in: S.C. Moldoveanu, V. David (Eds.), Analytical Liquid Chromatography -New Perspectives, IntechOpen, London, 2022. [40] T. Taylor, Mixed-mode HPLC separations: what, why and how, LCGC North Am. 32 (3) (2014) 226. [41] Y. Li, Z. Xu, Y. Feng, X. Liu, T. Chen, H. Zhang, Preparation and evaluation of poly-Llysine stationary phase for hydrophilic interaction/reversed-phase mixed-mode chromatography, Chromatographia 74 (2011) 523e530. [42] L. Qiao, S. Wang, H. Li, Y. Shan, A. Dou, X. Shi, G. Xu, A novel surface-confined glucaminium-based ionic liquid stationary phase for hydrophilic interaction/anionexchange mixed-mode chromatography, J. Chromatogr. A 1360 (2014) 240e247. [43] X. Liu, C.A. Pohl, Comparison of reversed-phase/cation-exchange/anion-exchange trimodal stationary phases and their use in active pharmaceutical ingredient and counterion determinations, J. Chromatogr. A 1232 (2012) 190e195.

354

Method Development in Analytical HPLC

[44] Q.-H. Wan, Mixed-Mode Chromatography, Principles, Methods, and Applications, Springer, Singapore, 2021. [45] K.K. Unger, S. Lamotte, E. Machtejevas, Column technology in liquid chromatography, in: S. Fanali, P.R. Haddad, C.F. Poole, M.-L. Riekkola (Eds.), Liquid Chromatography, 1, second ed., Elsevier, Amsterdam, 2017. [46] T. Teutenberg, J. Tuerk, M. Holzhauser, T.K. Kiffmeyer, Evaluation of column bleed by using an ultraviolet and a charged aerosol detector coupled to a high-temperature liquid chromatographic system, J. Chromatogr. A 1119 (2006) 197e201.

Mobile phase and the role of solvents in high-performance liquid chromatography 8.1

8

General aspects regarding mobile phase

The mobile phase in high-performance liquid chromatography (HPLC) is made from one or more solvents in which, in many cases, some additives are dissolved. The role of mobile phase is to carry the sample components through the chromatographic column where the sample components are engaged in the separation process, and then to flow the separated sample components through the detector. Selection of mobile phase composition is a major step in developing an analytical method because mobile phase has important role in both separation and detection. Several aspects regarding mobile phase components are presented in this chapter.

Solvents used in HPLC A range of solvents can be used in HPLC, although depending on the type of HPLC, specific constraints are imposed in the solvent selection for being used in the mobile phase. For example, in most HPLC techniques, water must be one of the components of the mobile phase. In nonaqueous reversed phase (NARP) or in normal phase HPLC (NPC), nonpolar solvents must be used. Besides water, the most common solvents used in different types of chromatography are methanol and acetonitrile. Solvents such as ethanol, tetrahydrofuran, isopropanol, hexane, cyclohexane, dioxane, etc., are also relatively common. However, depending on the type of chromatography and special separation needs, many other solvents can be utilized as mobile phase. A special type of solvents used in HPLC is that of “green solvents.” In an effort to reduce the use of chemicals harmful to the environment, solvents such as pure ethanol or water mixed with compounds like propylene carbonate, ethyl lactate, ethyl acetate, or some ionic liquids were utilized. The use of different solvents for each particular type of HPLC is further discussed in Chapter 9. Besides the use as mobile phase components, the solvents are also used for sample dissolution, sample injection in the flow of mobile phase, for the needle wash of the autosamplers (autoinjectors), and as solvents for the reagents used in postcolumn derivatizations. The nature of the solvents used as mobile phase do not affect only the separation but also the detection. Several utilizations of solvents in HPLC are further presented in this chapter. Solvents used in HPLC must be very pure unless a known impurity that is present in the solvent does not affect in any ways the HPLC analysis. Solvent impurities may affect the HPLC analysis in various ways. The impurity may generate: (1) interaction with the analytes, (2) problems with the separation, (3) problems with the detection, Method Development in Analytical HPLC. https://doi.org/10.1016/B978-0-443-29849-3.00005-1 Copyright © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

356

Method Development in Analytical HPLC

and (4) deterioration of the HPLC equipment. Solvent impurities may come from additives, solvent decomposition (e.g., in case of peroxide formation in ethers, or hydrolysis of esters or in case of chlorinated compounds), or can be present from the solvent synthesis. As an example, chlorinated compounds may hydrolyze to form HCl, which can produce decomposition of the analytes and deterioration of HPLC equipment. Besides the use of high-purity solvents, it is also important to properly maintain the HPLC system such that to prevent any contamination or changes of the solvent composition from trace compounds contaminating the instrumentation. Careful handling and storage of HPLC solvents are also necessary to maintain their high purity and avoid their degradation over time. A more detailed presentation of solvent properties in given in Sections 8.2 and 8.3.

The use of isocratic or of gradient elution As previously indicated, the mobile phase in HPLC separations can be used at a constant composition and flow rate, indicated as isocratic conditions, or they can be modified during the chromatographic run indicated as gradient conditions. Gradient is commonly used for improving separation and for shortening chromatographic runs. The advantages of changing during the chromatographic run of mobile phase composition and possibly flow rate are further discussed in some detail in Section 8.6.

Flow rate and temperature of the mobile phase The flow rate in analytical HPLC is set at a specific value with several purposes: (1) the optimization of separation, (2) maintaining an acceptable backpressure in the column, and (3) achieving the detector requirements. In most analytical separations, the flow rate is kept constant during the chromatographic run, but a change in it can be part of a gradient program. 1) In theory, the volumetric flow rate U in a chromatographic column should be selected close to the optimum for efficiency as indicated based on van Deemter equation (Eq. 4.1.51). Typical flow rates for columns with particle sizes between 3 and 5 mm are in the range of 0.5 and 1.0 mL/min, and for smaller particles (between 1.6 and 2.7 mm) in the range of 0.1 and 0.6 mL/min. However, the flow rate can be selected somewhat different from the ideal value if by trial and error a better separation is obtained by this change. In some instances, the flow rate is increased to shorten the chromatographic run, or is decreased to obtain a better sensitivity (see Eq. 4.4.16). 2) The backpressure in a chromatographic column is related to the flow rate by Darcy equation (Eq. 4.1.57), which written in function of U has the form: Dp ¼

4hUfr L ε pd2 d2p

(8.1.1)

The backpressure must be kept below a specific value during the chromatographic run. For the silica-type stationary phases, the value of Dp can be rather high, and the limitation

Mobile phase and the role of solvents in high-performance liquid chromatography

357

may come from the pumping characteristics of the HPLC system (see Section 3.2). For the use of columns with polymer stationary phase, where the phase collapse may occur at higher pressure, the maximum backpressure is usually indicated by the manufacturer and is dictated by the column construction. 3) Several detectors may have specific requirements regarding an acceptable flow rate. The flow rate should not exceed these recommendations characteristic for the detector used for the analysis. Mobile phase temperature is another parameter that influences both separation and mobile phase viscosity with implication in the variation of column backpressure. The role of temperature in the separation in HPLC was presented in Section 2.4, and the variation of mobile phase viscosity with the temperature will be discussed in Section 8.2. The viscosity variation with temperature may affect the column backpressure (see Eq. 8.1.1) and therefore the selection of the flow rate that should not exceed a specific value.

Buffers and additives used in HPLC The pH of the mobile phase in HPLC has in many instances a crucial role in the separation process. Based on the mobile phase pH, the components of injected sample in the HPLC may have different structures which determine the separation and even the properties of the stationary phase may be affected in some cases by the mobile phase flowing through the column. Also, various additives can be used in the mobile phase having roles in both separation and detection. A more detailed presentation of the role of pH and of additives in the mobile phase will be given in Section 8.5.

Key points • •

Mobile phase composition is a crucial element for an HPLC analysis. Solvents are key components of the mobile phase, but buffers and other additives also have their role in making an appropriate mobile phase.

8.2

Properties of solvents relevant for the HPLC separation

Solvents are the main components of the mobile phase, and their nature is essential in the separation process. For example, solvents have an essential role in the equilibrium Xmo $ Xst but also in the solvation of the molecules of sample components and in the solvation of the functionalities from the stationary phase. For example, regarding the interaction with the stationary phase, in reversed-phase HPLC (RPHPLC), the organic solvents from the mobile phase are involved in adsorption/dissolution in the hydrophobic layer of the stationary phase, and in HILIC, water molecules have intense interactions with polar functionalities of stationary phase. Such interactions significantly influence the separation which is modified as compared to that offered by the stationary active phase alone. Some information and discussions on the main properties of solvents and their influence in the

358

Method Development in Analytical HPLC

separation process are provided in this section, and more details can be found in dedicated publications, such as Refs. [1e3].

General properties of solvents A number of general properties of chemical compounds, as discussed in Section 5.1, are also used for the description of the properties of solvents. Chemical composition of the solvent determines a number of physical parameters important for the separation. 0 Among these are the boiling point B.p., density r, the superficial tension g , molecular volume Vm , dielectric constant ε, diffusion coefficient for a compound X in a solvent when the solvent also affects the value of DX;solv , etc. Such properties should be considered when selecting a solvent as a component in the mobile phase, but they are not directly affecting the mobile phase eluting properties. One general property of solvents that is in particular important in HPLC is solvent viscosity h. The dynamic viscosity, which is usually of interest, is the measure of fluid resistance to shearing flow and it is measured in poise P (or centipoise cP). Data on viscosity for a large number of solvents can be found in the literature [4]. As indicated by Eq. 4.1.57 (Darcy equation), the backpressure Dp in a chromatographic column is directly proportional with mobile phase viscosity (which is made with specific solvents). Two aspects regarding viscosity directly affect the use of solvents in HPLC: (a) the variation of viscosity with temperature, and (b) the variation of viscosity of solvent mixtures depending on the proportion of the solvents in the mixture. Several models were developed for describing the variation of viscosity with the temperature. In Arrhenius-like equation, for example, the variation with temperature of viscosity is given by: hðTÞ ¼ h0 expðE = RTÞ

(8.2.1)

where T is temperature (in Kelvin deg), h is viscosity (in cP), and h0 (preexponential factor)