946

# HPLC Method Development and Optimization with Validation in Mind

Authored by: Michael E. Swartz , Ira S. Krull

# Handbook of Analytical Validation

Print publication date:  April  2012
Online publication date:  April  2012

Print ISBN: 9780824706890
eBook ISBN: 9781420014488

10.1201/b12039-4

#### Abstract

Method development and optimization is the foundation of any validated method; a properly developed and optimized method can help to ensure a method’s success upon implementation. Though the focus of this chapter is on HPLC methods, by and large, the conceptual steps outlined here for method development and optimization will also be applicable to many analytical procedures performed in a regulated environment, including, but not limited to, gas chromatography (GC), capillary elec-trophoresis (CE), or mass spectrometry (MS). In this chapter, different approaches to method development and optimization in a regulatory environment are discussed, along with suggested HPLC instrument configurations and software tools.

#### 3.1  Introduction

Method development and optimization is the foundation of any validated method; a properly developed and optimized method can help to ensure a method’s success upon implementation. Though the focus of this chapter is on HPLC methods, by and large, the conceptual steps outlined here for method development and optimization will also be applicable to many analytical procedures performed in a regulated environment, including, but not limited to, gas chromatography (GC), capillary elec-trophoresis (CE), or mass spectrometry (MS). In this chapter, different approaches to method development and optimization in a regulatory environment are discussed, along with suggested HPLC instrument configurations and software tools.

#### 3.2  HPLC Method Development Approaches

An effective analytical method development process involves evaluating and optimizing various method parameters to satisfy the stated goals of the method or procedure. There are many literature reports of the experimental design and approach to method development [1-7], and over the years many different approaches to HPLC method development have evolved. The selection or development of any new or improved method often involves tailoring existing approaches and instrumentation to the current analytes of interest, as well as to the final objectives or requirements of the method. It often also involves robustness (Chapter 5) or “prevalidation” studies, performed to ensure that the resulting method is “validatable.” Perhaps the easiest and most straightforward method development approach is to survey the existing literature to see if methods, either exact or related, already exist. In addition to the scientific literature, many instrument and software vendors offer databases of existing applications, some with methods that can be directly downloaded into a chromatography data system (CDS), for example, the D-Library (Dionex Inc., Sunnyvale, California). Another method development approach involves starting with the structures of the analytes and developing the method based on information determined from these structures, either obtained from reference material, or observed or measured. These physical-chemical properties of the analytes, for example, solubility, pKa or pKb, spectral properties, molecular weight, and polarity, are used to choose rational starting mobile phase and column conditions from which additional fine-tuning or optimization experiments are carried out. Software is available (ACD/ChromGenius or AutoChrom, Advanced Chemistry Development, Toronto, Ontario, Canada) that can utilize structure information to predict retention times and set restrictions on separation conditions. The structure-based physical-chemical property approach is often combined with che-mometric software that can also model chromatographic separations. Several software and instrument vendors offer software for chromatographic modeling, including DryLab (Molnar Institute for Applied Chromatography, Berlin, Germany) and ACD/ LC and GC Simulator (Advanced Chemistry Development, Toronto, Ontario, Canada).

### Table 3.1   Example of Some Commercially Available Method Development Software

SoftwareTitle

Vendor

AutoChrom

ACD Labs, Toronto, ON, Canada

Several different modes

One module uses analyte properties for starting point

Uses MS for peak tracking

ChromSword

Agilent, Wilmington, Delaware

Uses analyte structure

Has embedded column database

Automated or stand-alone

Fusion AE

Waters, Milford, Massachusetts

Quality by design experimental design

Automated with the Waters system

DryLab

Molnar Institute for Applied Chromatography, Berlin, Germany

Theory-based

Software programs, either third party or CDS software itself, can often interact with HPLC instrumentation to automate the entire method development process, some even in a quality-by-design (QbD) framework (Fusion-AE Method Development Software, Waters Corporation, Milford, Massachusetts). Column or method screening approaches take advantage of this interaction to systematically screen different columns and method conditions (mobile phase composition, pH) followed by additional optimization, often in combination with other approaches. Table 3.1 lists some of the common, commercially available software available to assist in method development.

#### 3.3  Method Goals

There are several valid reasons for developing new methods of analysis:

• There may not be a suitable method for a particular analyte in the specific sample matrix.
• Existing methods may be too error-, artifact-, and/or contamination-prone, or they may be unreliable (have poor accuracy or precision).
• Existing methods may be too expensive, time consuming, or energy intensive, or they may not be easily automated.
• Existing methods may not provide adequate sensitivity or analyte selectivity in samples of interest.
• Newer instrumentation and techniques may have evolved that provide opportunities for improved methods, including improved analyte identification or detection limits, greater accuracy or precision, or better return on investment.
• For legal, compliance, or scientific reasons, there may be a need for an alternative or orthogonal method to confirm the analytical data originally obtained from existing methods.
• Existing method may not be stability indicating.

Before undertaking method development, it is important to have a complete understanding of the goals, objectives, and expectations of the method, and then to translate the goals of the method into a method development design and to define the required analytical performance characteristics for validation.

Goals for a new or improved analytical method might include the following:

• Qualitative identification of the specific analytes of interest providing some structural information to confirm “general behavior” (e.g., retention time, color change, pH)
• Quantitative determination (at trace levels when necessary) that is accurate, precise, and reproducible in any laboratory setting when performed according to established procedures (e.g., SOPs)
• Stability indicating
• Ease of use
• Ability to be automated
• High sample throughput
• Rapid sample turnaround time
• Low cost per analysis
• Sample preparation that minimizes cost, time, effort, materials, and volume of sample consumed
• Direct output of qualitative or quantitative data to laboratory computers in a format usable for evaluation, interpretation, printing, and transmission to other locations via a network/laboratory information system (LIMS)

Reference standards that have been well identified and characterized, and whose purity is already known should be used for initial method development and preliminary evaluation of the goals and requirements of the method, and the initial analytical performance characteristics for validation should be identified, according to the type of method/procedure (Chapter 7).

#### 3.4  HPLC Method Development Instrumentation

Since the 1970s, the fundamental components of a basic liquid chromatograph have consisted of the same basic parts: a pump, a means of injecting a sample, a column, a detector, and some type of data recording device. However, during the past decade, and certainly to a greater extent within the past five years or so, the basic components have become much more sophisticated, and specialized systems have emerged for specific applications, including those for method development. There are several key components of any HPLC system, and systems used specifically for method development are really no different. HPLC systems can be modular or integrated, and use either isocratic or gradient solvent delivery. Modular systems consist of separate modules connected in such a way as to function as a single unit, and can provide a degree of flexibility to exchange different components in and out of the system, sometimes necessary for maintenance purposes or experimental requirements. However, in regulated laboratories, this flexibility may not be viewed as an advantage due to compliance issues with analytical instrument qualification.

In general, gradient systems are preferred over isocratic systems for method development because of their multisolvent capability. Gradient multisolvent systems can be used to prepare mobile phases on-the-fly, often referred to as “dial-a-mix” or “auto blend,” providing maximum flexibility for method development (especially for method and column screening approaches), and the mobile phases are often more robust and accurate than premixed mobile phases when methods are in routine use. Figure 3.2 illustrates this capability. Figure 3.2c shows three overlaid chro-matographic results from one system, three different chemists, on three different days, using premixed solvents. The chemist-to-chemist reproducibility is seen to be quite variable. In Figure 3.2b, every tenth injection of 100 runs from an experiment using premixed solvents are overlaid. Figure 3.2b illustrates that even on a single system, with a single chemist, premixing solvents can affect repeatability over time. Variability in this instance most likely arises from selective evaporative loss of the organic solvent, as later runs have longer retention times. Finally, in Figure 3.2a, overlaid results are presented for every tenth injection of 100 runs using auto blend or dial-a-mix; that is, using the system to make the mobile phase. As illustrated, the

Figure 3.1   Diagram illustrating (a) high-pressure and (b) low-pressure mixing system architecture. The high-pressure system uses two separate pumps and a controller; the lowpressure system uses a single pump and a multiport proportioning valve.

Figure 3.2   Comparison of premixing mobile phase solvents to auto blend. Figure 1c: three overlaid chromatographic results from one system, three different chemists, on three different days, using premixed solvents. In Figure 1b, every tenth injection of 100 runs from an experiment using premixed solvents are overlaid. In Figure 1a, overlaid results are presented for every tenth injection of 100 runs using auto blend, or dial-a-mix; that is, using the system to make the mobile phase. (Reprinted from HPLC method development for pharmaceuticals, Volume 8 of Separation Science and Technology, S. Ahuja, Editor, Chapter 6, Contemporary liquid chromatographic systems for method development, p. 148, 2007.)

system is far more accurate in preparing the mobile phases than either a single or multiple analysts premixing the mobile phase. Using auto blend, different organic solvent proportions, buffer strength, and pH can be generated using the solvent manager to proportionally mix the appropriate stock solutions to obtain the final mobile phase conditions. Auto blending of this type can be used to the analyst’s advantage during method development.

The automated blending of solvents might at first seem a trivial matter. However, automated method development systems depend on precise and reproducible blending in scouting experiments designed to study the effects of different mobile phase conditions on selectivity. In the strictest sense, gradient chromatography is essentially auto blending, albeit over time. The kinds of results obtained in Figure 3.2c

Figure 3.3   Six overlaid chromatograms of a method requiring critical resolution of a series of minor components requiring accurate and precise mobile phase delivery. (Reprinted from HPLC method development for pharmaceuticals, Volume 8 of Separation Science and Technology, S. Ahuja, Editor, Chapter 6, Contemporary liquid chromatographic systems for method development, p. 150, 2007.)

are critical to the type of separations illustrated in Figure 3.3. Figure 3.3 shows six overlaid chromatograms of a method requiring critical resolution of a series of minor components. Without accurate and precise mobile phase generation and solvent delivery, this critical resolution could not be maintained. Isocratic conditions, if desired for the final method, can be determined from gradient conditions, and of course still be run on the gradient system.

Figure 3.4   Example of a typical HPLC system configured for method development. (Reprinted from HPLC method development for pharmaceuticals, Volume 8 of Separation Science and Technology, S. Ahuja, Editor, Chapter 6, Contemporary liquid chromatographic systems for method development, p. 151, 2007.)

#### 3.4.1  HPLC Systems for Column and Method Scouting

Method and column scouting is a method development approach commonly used to investigate potential starting conditions for further method optimization. A typical HPLC system used to generate the kind of results obtained in Figures 3.2 and 3.3 and run scouting experiments is shown in Figure 3.4. Most major LC manufacturers’ systems can be similarly configured into a resulting method development workhorse system to generate the kind of results obtained in Figure 3.3 and run column scouting experiments where the mobile phase can be varied over a range of conditions, including organic content and pH. In addition to the basic solvent and sample manager, systems for method development are often configured with solvent and column switching valves, a column oven, and multiple detector capabilities (Section 3.4.6). For the most part, photodiode array (PDA) and single quadrupole mass spectrometry (MS) are the most useful detectors for method development. Other useful detectors include evaporative light scattering (ELSD) or corona charged aerosol (CAD). Systems configured in this manner are capable of delivering mobile phases consisting of different blends of multiple organic solvents, multiple buffers and pHs, and operating four or more columns at different temperatures. This multiparameter mobile phase and multicolumn capability provides access to many of the potential variables the analyst needs to investigate when manipulating selectivity in method development. In addition, using multiple detectors in combination, there is less dependence on individual analyte properties such as UV chromophores or ionization.

Scouting or screening systems are also often used to identify the most promising conditions (solvent, column pH, etc.) for further optimization (Section 3.5) or development. By considering the physical characteristics and analyte properties, templates (collections of instrument methods) can be written in the chromatography data system (CDS) to generate the various mobile phase conditions, equilibrate or switch columns, perform all the chromatographic runs, and run wash or shutdown procedures for both the columns and the system. These templates are usually written once and then used repeatedly as new methods need to be developed, allowing the analyst to run the system in a semi-automated manner. A typical screening experiment might generate in excess of sixteen chromatograms; for example, using four different columns, high and low pH, and two different solvents (e.g., methanol and acetonitrile) as outlined in Figure 3.5. Of course, multiple additional conditions generate a great deal more chromatographic data, and once generated, all the data must be scrutinized. However, rather than go through each chromatogram individually, the analyst can use the custom reporting features of the CDS relational database to generate summary plots that at a glance reveal which runs gave the greatest number of peaks, most resolution, etc., as illustrated in Figure 3.6.

Figure 3.5   Example column scouting method development approach. In this example, four different columns, two different buffers (high and low pH), and two different organic solvents (acetonitrile (ACN) and methanol (MeOH)) are methodically scouted to generate a set of conditions that can be chosen for additional optimization if needed.

Figure 3.6   CDS bar chart data mining. Results from the types of experiments run in Figure 3.2 can be summarized in charts such as these using the CDS, making it much easier to pick out optimum conditions as opposed to reviewing each chromatogram individually.

#### 3.4.2  Automated HPLC Method Development Systems

Automation of the optimization is a natural extension of a methodical, planned method development process. Why automate? The desire to automate method development stems from the simple reason that traditional manual HPLC method development is a labor-intensive, time-consuming, and often-imprecise process, resulting in lost time, money, and productivity because it can take weeks to develop a method manually. Automated method development systems provide an alternative to the traditional slow, manual, and unreliable trial-and-error method development approach and can often reduce method development time to as little as a few hours. In addition, automated systems can often evaluate a larger number of conditions, thus improving the robustness of the method.

Systems have been developed that utilize external modeling software (Table 3.1; e.g., DryLab, Molnar-Institute, Berlin, Germany; or LC Simulator or AutoChrom, Advanced Chemistry Development, Toronto, Ontario, Canada) that either partially or completely automate the HPLC method development process (8-14). These theory-based modeling software programs allow analysts to evaluate a much wider range of experimental conditions than would ever be practical by running experiments in the laboratory, significantly decreasing method development and optimization time. With this type of software, the effects of variables, either alone or in combinations—for example, organic concentration, pH, temperature, gradient slope, and buffer concentration—can be easily observed. In addition, analysts can

• Evaluate method robustness to decrease the cost of revalidating methods
• Transfer gradient methods from one instrument to another, eliminating method redevelopment time
• Model two separation variables simultaneously for faster method development
• Shorten run times to increase sample throughput
• Train new chromatographers and establish laboratory method development SOPs

Figure 3.7 shows a screenshot from DryLab software during the development of a separation of some nitroaromatics. Screens such as this in the software can be used to model separations, including different solvent compositions and column configurations. The underlying software algorithms are based on HPLC theory, and are very accurate in their predictions, as summarized in Table 3.2 for a separation of cocaine, methadone, and related substances.

The critical component in a completely automated system is software that bridges the gap between the modeling software and CDS software that runs the system and generates data. In these systems, the process of method development starts with the help of a Windows-based interface between the modeling software (e.g., AutoChrom, LCD Labs) and the CDS. The software interface asks for specific information about separation needs, and using software protocols, suggests actual starting coditions, including pH, solvent, and column. The software can also facilitate the setup of the method in the CDS and complete the analysis. Systems and software are now available from many vendors that incorporate PDA and MS for peak tracking during automation. A PDA or MS spectral-based peak-tracking algorithm allows more accurate identification of sample components during the method development process, identifying peaks as selectivity and therefore elution order changes over the course of a few “chemistry calibration” runs. Once the calibration runs are processed, the chromatography variables are quickly modeled, and an optimized chromatographic method prediction is obtained. Using systems of this type, with intelligent decision-making software, it is not uncommon to optimize a method in as little as four or five chromatographic runs over just a few hours [13,14].

Figure 3.7   Example DryLab resolution map. Chromatogram at point B is a prediction of the choice of experiment dictated by the mouse placement in the software at point A. (Source: Figure courtesy of Molnar-Institute, Berlin, Germany.)

### Table 3.2   Drylab Prediction Accuracy

Peak

DryLab Predicted Tr

Experimental Tr

% Error

Cocaine

3.98

3.94

0.70

Benzoylecgonine

4.31

4.23

1.32

Antipyrine

5.08

5.01

0.98

Phenacetin

9.50

9.48

0.17

Dibucaine

9.94

10.05

0.77

10.20

10.31

0.73

Source: Data courtesy of Molnar-Institute, Berlin, Germany.

#### 3.4.3  UHPLC in Method Development Systems

One of the primary drivers for the growth and continued use of HPLC has been the evolution of packing materials used to affect the separation. The underlying principles of this evolution are governed by the van Deeter equation:

$H = A( d p ) + B/u + C( d p ) 2 u$

which is a formula that describes the relationship among H, plate height (HETP or column efficiency); linear velocity, u, (flow rate); and particle size or diameter, dp.

The “A” term represents eddy diffusion, the “B” term represents longitudinal diffusion, and the “C” term represents resistance to mass transfer in and out of the particle.

According to the van Deeter equation, as the particle size decreases to less than 2 μηι, not only is there a significant gain in efficiency, but the efficiency does not diminish at increased flow rates or linear velocities [15]. By using smaller particles, speed and peak capacity (number of peaks resolved per unit time) can be extended to new limits; this has come to be known as Ultra High Pressure LC (UHPLC). UHPLC takes full advantage of chromatographic principles to run separations using columns packed with smaller particles, and/or higher flow rates for increased speed, with superior resolution and sensitivity [16-18].

An example of the use of UHPLC for rapid method development is illustrated in Figure 3.8. The method development process for this rather complex separation was accomplished in twenty-two preliminary runs, including organic composition scouting, and individual injections for peak identification. Due to the short UHPLC run times, the entire method was developed in less than an hour. Attempting to do the same via HPLC could take days to weeks longer, and in the end, HPLC may not be able to accomplish this result due to its inherent lower efficiency and resolving power.

Figure 3.8   UHPLC separation of coumarin and related compounds illustrating fast method development. Final conditions included a 2.1 by 50 mm 1.7-μπι ACQUITY UPLC BEH C18 (Waters Corporation, Milford, Massachusetts) column at 35°C. A 5-80% B linear gradient over 1.0 minute, at a flow rate of 1.0 mL/min was used. Mobile phase A was 0.1% formic acid, and B was acetonitrile. UV detection at 254 nm and 40 pts/s. Peaks are (1) 7-hydroxycoumarin-gluconoride, (2) 7-hydroxycoumarin, (3) 4-hydroxycoumarin, (4) coumarin, (5) 7-methoxycoumarin, (6) 7-ethoxycoumarin, (7) 4-ethoxycoumarin.

Figure 3.9   Schematic diagrams of a low-pressure mixing system using a single pump and a four-position solvent proportioning valve (bottom), and a high-pressure mixing system using multiple pumps (top).

#### 3.4.4  Solvent Management

LC pumps are sometimes categorized according to the way solvents are blended (Figure 3.1). Low-pressure designs use a single pump to deliver mobile phases generated by an upstream proportioning valve. High-pressure systems use two or more pumps to proportion solvents downstream at high pressure. As illustrated in Figure 3.9, the most significant difference between high-pressure and low-pressure systems is in the system volume. While low-pressure systems usually exhibit less compositional ripple in the chromatographic baseline, high-pressure systems usually have lower volumes. Therefore, if speed or high throughput is desired, high-pressure systems are usually preferred. However, low-pressure systems can usually accommodate a larger number of different mobile phase solvents, and software-configurable solvent-select valves are also frequently used on method development systems to expand capability. Regardless of the type of system used, it is important to remember that a proper determination of the volume is important for any system used in method development. Problems related to method transfer can often be traced to differences in system, dwell, or gradient delay volumes, as no two systems will have exactly the same volume. The volume difference is particularly significant when transferring methods between low-pressure and high-pressure systems. In addition, problems may also result from how the volumes are calculated [19]. Accurate volume determinations for high-pressure systems can be made using a step gradient method because the mobile phase is generated post-pump. For accurate low-pressure system volume determinations, a linear gradient must be used, to take into account the pre-pump volume from the solvent proportioning valve. It is for this reason that many method developers now recommend programming in a short isocratic step at the beginning of every gradient to accommodate transfer to systems with differing volumes [20]. When UHPLC systems were first introduced in 2004, only high-pressure mixing systems were available. Recently (2010), low-pressure mixing systems were introduced that combine all the attributes of working with small particles at high pressures with quaternary solvent mixing for method development [21].

#### 3.4.5  Sample Management

Conventional injection valves, either automated or manual, are not designed and hardened to work at extreme pressure; and to protect the column from experiencing extreme pressure fluctuations, the injection process must be relatively pulsefree. The swept volume of the sample manager also must be minimized to reduce potential band spreading. For UHPLC, a fast injection cycle time is needed to fully capitalize on the speed of the analysis, which in turn requires a high sample capacity. Low volume injections with minimal carryover are also required to realize the increased sensitivity benefits. Temperature control and compatibility with a wide range of sample formats (e.g., vials, microtiter plates) are also desirable features in any sample management device used for method development.

#### 3.4.6  Detection

Detection plays an important role in method development systems, and the most desirable configurations include a variety of complementary detectors to respond to the widest range of analyte attributes. Depending on analyte properties, the most commonly employed detectors in method development systems include UV (PDA), evaporative light scattering (ELSD), corona charged aerosol (CAD), and mass spec-trometry (MS-either single or triple quadrupole). Multiple detectors in a system can be configured in series or parallel; often, the choice of which configuration to use depends on whether or not the detector is destructive. Destructive detectors (e.g., ELSD, MS, CAD) must be placed last in the flow path and require splitting of the flow stream.

Photodiode array (PDA) detection is commonly used during method development to determine peak identity and purity/homogeneity. PDAs extend the utility of UV detection by providing spectra of eluting peaks that can be used to aid in peak identification, and to monitor for co-elutions (peak homogeneity or purity), helpful during method development. They can also serve as a multiwavelength UV/VIS detector. The spectra collected at the chromatographic peak apex can be used to create a library that can in turn be used to compare subsequent spectra for identification purposes, and spectra collected across the peak at each data point can be compared to evaluate peak homogeneity or purity. The added spectral resolution of modern PDA detectors, coupled with chromatography data system (CDS) software algorithms, can quickly compare fine differences in the spectra not clearly visible to the eye. Some comparisons are done by a simple direct point-to-point comparison of spectra, while in others, complex vector analysis in multidimensional space is performed to look at spectral fine structure. In order for PDA spectral comparisons to work, the compounds must have some UV absorbance, and there must be some degree of spectral and chromatographic resolution. Spectra will also be changed if the organic concentration or pH is altered, for example, during method development. The changes in spectra resulting from mobile phase differences often result in a shifting of the spectra, affecting the quality or the “fit” of the match, but not necessarily the information obtained.

Recent improvements in the ability to efficiently nebulize an HPLC column effluent has led to the increased utility and popularity of the evaporative light scattering detector (ELSD). The ELSD works on the principle of evaporation (nebulization) of the mobile phase, followed by measurement of the light scattered by the resulting particles. The column effluent is nebulized in a stream of nitrogen or air carrier gas in a heated drift tube, and any nonvolatile particles are left suspended in the gas stream. Light scattered by the particles is detected by a photocell mounted at an angle to the incident light beam. Carrier gas flow rate and drift tube temperature must be adjusted for whatever mobile phase is used. Detector response is related to the absolute quantity of analyte present; and while decreased sensitivity will be obtained for volatile analytes, unlike the UV detector, no chromophores are required and it has orders of magnitude more response than the refractive index (RI) detector, another common detector in situations where analytes do not have strong chromophores. The ELSD also has the advantage over RI detection in that the response is independent of the solvent, so it can be used with gradients, and is not sensitive to temperature or flow rate fluctuations. Mobile phases, of course, must be volatile, similar to those used for MS detection, as listed in Table 3.3 [6]. Linearity can be limited in some applications, but is certainly quantitative over a wide enough range if properly calibrated. Recent applications of the ELSD have also been extended to UHPLC.

### Table 3.3   Properties of Common Organic Solvents Used in Liquid Chromatography

Solvent

uv cutoff (nm)a

Viscosity (cP)

Boiling Point (°c)

Acetonitrile

190

0.38

82

1-Butanol

215

2.98

118

Dimethylformamide

268

0.92

153

Dimethylsulfoxide

268

2.24

189

Heptane

200

0.40

98

Hexane

195

0.31

69

Methanol

205

0.55

65

n-Propanol

210

2.30

97

Tetrahydrofuran

212

0.55

66

Water

190

1.00

100

a Wavelength at which solvent absorbs 1.0 AU in a 10-mm cell. (Source: Adapted from Snyder, L. R. et al., Introduction to Modern Liquid Chromatography, 3rd edition, John Wiley & Sons, Hoboken, NJ, 2010, p. 882.)

Corona charged aerosol detection (CAD), sometimes referred to as corona discharge detection (CDD), is a unique technology gaining in popularity in which the HPLC column eluent is first nebulized with a nitrogen (or air) carrier gas to form droplets that are then dried to remove mobile phase, producing analyte particles [22,23]. The primary stream of analyte particles is met by a secondary stream of nitrogen (or air) that is positively charged as a result of having passed a high-voltage platinum corona wire. The charge transfers diffusionally to the opposing stream of analyte particles, and is further transferred to a collector, where it is measured by a highly sensitive electrometer, generating a signal in direct proportion to the quantity of analyte present.

Because the entire process involves particles and direct measurement of charge, CAD is highly sensitive, provides a consistent response, and has a broad dynamic range, which offers advantages when analyzing compounds lacking UV chromo-phores, as illustrated in Figure 3.10. Often compared to other universal-type HPLC detectors, such as RI and ELSD, CAD has been shown to be much easier to use, and similar to ELSD but unlike RI, can accommodate gradients. In addition, CAD response is not dependent on the chemical characteristics of the compounds of interest, but on the initial mass concentration of analyte in the droplets formed upon nebulization, providing a much more uniform response as opposed to, for example, UV, where responses can vary dramatically according to the wavelength used and the extinction coefficient.

Figure 3.10   Simultaneous analysis of anions and cations using HILIC/CAD. Conditions: A Sequant ZIC®-pHILIC 5 mm, 4.6 × 150 mm column (The Nest Group, Southborough, Massachusetts) operated at 30°C was used. Gradient conditions: 20 to 70% B over 26 min; mobile phase A: 15% 100 mM ammonium acetate pH 4.68, 5% methanol, 20% IPA, 60% acetonitrile; mobile phase B: 50% 30 mM ammonium acetate pH 4.68, 5% methanol, 20% IPA, 25% acetonitrile, at a flow rate of 0.5 mL/min and a 10-μL injection.

Mass spectrometry is a powerful analytical technique that can be used to confirm, quantify, identify, or characterize compounds of interest. Mass spectrometers measure the mass-to-charge (m/z) ratio of ions in the gas phase, allowing the determination of a compound’s molecular weight (to varying degrees of accuracy). By breaking apart molecules into fragments, MS can also be used to analyze smaller portions of a molecule. Information from this fragmentation assists in the elucidation of the compound’s chemical structure and properties.

Modern mass spectrometers are simple, easy-to-use instruments with a much smaller footprint than their predecessors and can be configured with a chromato-graphic method development system to provide a wealth of useful information. The basic components of an MS system are shown in Figure 3.11. Because a comprehensive treatment of MS is outside the scope of this chapter, the reader is urged to consult the many excellent detailed reviews of the technology that are available [24,25]. But single quadrupole mass spectrometers are becoming increasingly common in the method development laboratory and is covered here in some detail.

Quadrupole MS uses radio frequency (rf) and direct current (dc) voltages for the separation of ions, and are probably the most widespread mass spectrometers because of their relatively low price and ease of operation. In a quadrupole mass spectrometer, the rf and dc potentials are applied to four rods arranged in a square array, as illustrated in Figure 3.12. Ions are scanned or filtered by varying the DC/ RF voltages across the quadrupole rods. Generally speaking, quadrupole analyzers are used to determine the nominal mass of a compound. Nominal mass is often used to confirm the identity of known compounds in method development.

In method development, MS is used in much the same way as the PDA: to identify and track peaks as selectivity changes, and to monitor for co-elution. But unlike PDA, MS provides a positive identity, and can provide deconvoluted total ion chro-matograms specific for a molecular weight when co-elution of partial resolution does occur.

Of course, no detector response is universal. MS response is dependent on the ability to ionize a compound, and not all compounds can be ionized under all conditions. In similar respects, not all compounds have UV chromophores, so PDA detection is, of course, limited. However, it is very rare to have both no ionization and the lack of a UV chromophore; therefore, it is increasingly common to use MS and PDA in tandem during method development.

When it comes to MS detection, the low UHPLC system and dwell volume increases peak concentrations with reduced chromatographic dispersion at lower flow rates (no flow splitting), and the added resolution promotes increased source ionization efficiencies, making UHPLC the ideal technology for an MS inlet in a method development system. Higher UHPLC sensitivity also improves the quality of the spectra obtained.

Figure 3.11   The basic components of an MS system. (Reprinted from HPLC method development for pharmaceuticals, Volume 8 of Separation Science and Technology, S. Ahuja, Editor, Chapter 6, Contemporary liquid chromatographic systems for method development, p. 167, 2007.)

Figure 3.12   Quadrupole MS schematic. (Reprinted from HPLC method development for pharmaceuticals, Volume 8 of Separation Science and Technology, S. Ahuja, Editor, Chapter 6, Contemporary liquid chromatographic systems for method development, p. 167, 2007.)

However, similar to ELSD and CAD, using a mass detector places constraints on mobile phase selection. Proper selection of the mobile phase and any additives is critical to detection viability (Table 3.3). First and foremost, the mobile phase must be suitable for the ionization, and must be selected depending on the ioniza-tion mode (electrospray (ESI) or atmospheric pressure chemical ionization (APCI), positive or negative mode) and the analyte (e.g., pKa). The molecular weight of the mobile phase components should also be considered. It is not always possible to analyze compounds whose molecular weight is lower than the one of the mobile phase or any additives. For routine operation, it is easier to use volatile buffers. Acids such as HCl, H2SO4, or methane sulfonic acid might damage the instrument and should not be used; volatile organic acids (e.g., TFA, formic, acetic) should be used instead.

Some ions (e.g., Na, NH4, acetate) from the mobile phase can form adducts. In the case of phosphate, multiple adducts are observed, which can produce complicated mass spectra. The formation of an adduct is usually not a reason for avoiding a mobile phase as adducts can sometimes be used to advantage. Ion pairing reagents can impact the spray formation, the droplet evaporation, and compete in terms of ion formation, and are generally avoided. Buffer concentration is generally kept as low as possible (millimolar range). If the buffer concentration is too high, ion suppression occurs, thus affecting sensitivity.

Common eluents for LC/MS include methanol/water; acetonitrile/water (methanol usually gives a better sensitivity than acetonitrile); pH modifiers (formic, acetic acids, TFA, NH4, TEA, DEA); and buffers (carbonates, ammonium formate, ammonium acetate, ammonium carbonates, and ammonium phosphate (all nonvolatile)).

The column, while of course providing the separation without using a high concentration of buffers, or ion pairing reagents, must be stable so that the column will not “bleed” or shed interfering compounds. Special low- or no-bleed MS versions of columns are available from most suppliers.

#### 3.4.7  Column Module

Running a column at room temperature, or even “controlled” room temperature, is a thing of the past. Modern requirements for accuracy and precision require that columns be thermostated. Because temperature can also be used as a selectivity tool, modern method development systems typically require a column heater module. It is also an advantage for the column module to accommodate several columns of various geometries that can be randomly accessed from a software-controlled solvent switching valve. For UHPLC, the column module should also have the capability of adequately preheating the mobile phase without adding too much dispersion to prevent band broadening within the column.

#### 3.4.8  Columns for Method Development

The myriad of column stationary phases available may at first make choosing a column for method development seem like a daunting task. There are, however, a few basic guidelines to keep in mind, and some tools available, which make the task much simpler.

During method development, the primary goal is to manipulate selectivity for the analytes of interest. In order to do that, it is important to choose columns that are orthogonal. For column screening, common columns might include a selection of C18, phenyl, an embedded polar group stationary phase, and perhaps a column selected for highly polar compounds or a C8. Many column vendors provide column selectivity charts, which can provide valuable information about columns that are similar, or different, in selectivity. The USP also provides information of this type; as shown in Figure 3.13, the USP maintains a searchable database of column evaluations on their website that can also be used to find both equivalent or orthogonal (different) columns (for validation and development, respectively) [26].

Figure 3.13   USP Column Equivalency Database. Searchable database of column information that can be used to find orthogonal/different (for method development) or similar (for method validation) columns. (See http://usp.org/USPNF/columnsDB.)

Column lot or batch reproducibility should also be evaluated late in method development or prevalidation; generally, method development is performed on one column lot, and then verified both on another new column from the same lot and a new column from a different lot. These column lot evaluations are often performed either as a part of intermediate precision or robustness (Chapter 4). Column lot reproduc-ibility is less of an issue now compared to earlier, as many column manufacturers now manufacture their columns from scratch instead of buying the base silica, which itself can differ from lot to lot in trace impurities, which can affect chromatography.

One final note on the column front: it is also important that method development be performed using only new HPLC columns as columns can have a memory effect from previous conditions/methods that have been used, resulting in reproducibility issues when the column is eventually replaced.

### Table 3.4   Properties of Common Mobile Phase Buffers and Additives

pKa*

Buffer Range

Ms Compatibility

Acetic acid (glacial)

4.8

Ammonium acetate pKa 1

4.76

3.8–5.8

Yes

Ammonium acetate pKa 2

9.2

8.2–10.2

Yes

Ammonium bicarbonate

9.2, 10.3

8.2–11.3

Yes

Ammonium formate pKa 1

9.2

2.8–4.8

Yes

Ammonium formate pKa 2

9.2

8.2–10.2

Yes

Ammonium hydroxide

9.2

Yes

Ammonium phosphate, dibasic

7.2, 9.2

6.2–10.2

No

Formic acid

3.8

Yes

Phosphoric acid

2.1

No

Potassium phosphate, monobasic

2.1

1.1–3.1

No

Potassium phosphate, dibasic

7.2

6.2–8.2

No

Potassium phosphate, tribasic

12.7

11.7–13.7

No

Sodium citrate, tribasic

3.1, 4.8, 6.4

2.1–7.4

No

Triethylamine

11.0

Yes

Triethylammonium acetate (TEEA) pKa 1

4.76

3.8–5.8

Yes

Triethylammonium acetate (TEEA) pKa 1

11.0

10–12

Yes

Trifluoroacetic acid

0.3

Yes

#### 3.4.9  Mobile Phase Considerations

A rule of thumb for mobile phases used in method validation: the simpler the mobile phase, the more robust it will be. Some of the common organic solvents and their properties used in method development are listed in Table 3.4. Acetonitrile is generally preferred over methanol for method development because it has a lower UV cutoff, resulting in better PDA spectral interpretations. Methanol may be preferred for some MS applications; however, the actual selectivity obtained is more of a driving force for solvent choice. Use of high-temperature (reduced viscosity) and high-pressure technology such as UHPLC (or both) has opened up the range of possible solvents used, to include solvents such as isopropanol that are too viscous to use in conventional LC analyses. Many different buffer types and additives are used in HPLC mobile phases (Table 3.3); however, given the propensity to develop methods compatible with MS, or to use other evaporative-type detectors (ELSD, CAD), volatile components are most often used.

#### 3.5  Method Optimization

During method optimization, the initial set of conditions that has evolved from the first stages of development can be improved or maximized in terms of selectivity, resolution, peak shape, efficiency, and run or inject to inject cycle time. When optimizing any method, an attempt should be made to provide analytical figures of merit or specifications that are required to meet the assay requirements defined at the initial stages of method development. Results obtained during method development can then be measured against the desired specifications to determine how optimization should proceed. A target must be established; without adequate and definitive requirements or specifications, a method cannot be truly optimized. Evaluating the method against the predetermined specifications at this early stage should reveal the direction additional optimization experiments need to take to meet the method specifications.

If the initial analytical data derived from method development appears promising, it is time to evaluate its performance quantitatively. Initially, most work on method development and optimization is performed with analytical standards. In general, the analytical figures of merit generated to evaluate the method are also derived using standards. The scope of the method evaluation should be broad enough to include generation of information that is immediately usable for confirmation or identification of the analyte in any sample, for example, UV or mass spectra. Method optimization goals include increased sensitivity, peak symmetry and resolution, and a lack of analyte co-elutions.

As with method development, optimization of the method can follow either of two general approaches—manual or computer software driven—and the types of systems and software discussed in Section 3.1 for method development can also be used for method optimization. The manual approach commonly involves varying one experimental variable at a time, while holding all others constant, and recording changes in response. This univariate approach to system optimization is slow, time consuming, potentially expensive, and may miss the effects between variables (e.g., the effects of heat on pH). In the second approach, optimization using computer-driven software, higher efficiency/throughput can be obtained while experimental input is minimized. Automated software approaches can be applied to many applications. In addition, they are capable of significantly reducing the time, energy, and cost of virtually all instrumental method development, and can be useful to verify that the optimized method satisfies the stated goals of the method.

Certain general criteria are often considered a part of a “prevalidation” study:

• Chromatographic resolution is adequate.
• Limits of detection or quantitation that provide an adequate signal-to-noise response.
• Calibration plots are linear over several orders of magnitude, beginning with the quantitation limit.
• Suitable accuracy is obtained (perhaps performed in conjunction with linearity).
• Method- or procedure-appropriate precision is obtained (again, perhaps performed in conjunction with linearity).
• Demonstration of peak homogeneity (e.g., no co-elutions, or a demonstration that the method is stability indicating).

System optimization is one of the most time- and energy-consuming parts of the overall method development procedure. It requires an iterative procedure, constant replication, and the acquisition of a large amount of quantitative data. Too often, optimization results in a method that meets the immediate requirements of the analyst but ignores possible future needs. Ideally, the analyst should optimize each new method to the fullest practical extent in the time available, in order to ensure a broad utility of the method and obviate the repetition of experiments for future method development.

#### 3.6  Summary

Methods can be developed from scratch through scouting approaches, or adapted from existing methods found in the literature or other sources. But one thing is certain: method development is a complex, time-consuming process. Any effort to streamline, automate, and methodically and logically approach the process can pay great dividends in terms of throughput, efficiency, and reducing time to market, as well as producing a method that is easily validated.

#### References

Development and Validation of Analytical Methods, C. M. Riley and T. W. Rosanske, Editors, Pergamon Press, Elsevier Science, Amsterdam, Holland, 1996.
Pharmaceutical and Biomedical Applications of Liquid Chromatography, C. M. Riley, W. J. Lough, and I. W. Wainer, Editors, Pergamon Press, Elsevier Science, Amsterdam, Holland, 1994.
Snyder, L. R. , Glajch, J. L. , and Kirkland, J. J. , Practical HPLC Method Development, John Wiley & Sons, New York, 1988.
Snyder, L. R. , Glajch, J. L. , and Kirkland, J. J. , Practical HPLC Method Development, 2nd edition, Wiley-Interscience, New York, 1997.
Neue, U. , HPLC Columns, Theory, Technology, and Practice, Wiley-VCH Publishers, New York, 1997.
Snyder, L. R. , Kirkland, J. J. , and Dolan, J. W. , Introduction to Modern Liquid Chromatography, 3rd edition, John Wiley & Sons, Hoboken, NJ, 2010.
Swartz, M. E., Chapter 6: Contemporary liquid chromatographic systems for method development, pp. 145-186, in HPLC Method Development for Pharmaceuticals, Ahuja, S., Editor, Elsevier, Burlington, MA, 2007.
Dolan, J. W. , Snyder, L. R. , Djordjevic, N. M. , Hill, D. W. , Saunders, D. L. Van Heukelemn, L. , and Waeghe, T. J. Simultaneous variation of temperature and gradient steepness for reversed-phase high-performance liquid chromatography method development. I. Application to 14 different samples using computer simulation. J. Chromatogr., A, 803, 1, 1998.
Dolan, J. W. , Snyder, L. R. , Saunders, D. L. , and Van Heukelem, L. , Simultaneous variation of temperature and gradient steepness for reversed-phase high-performance liquid chromatography method development. II. The use of further changes in conditions. J. Chromatogr., A, 803, 33, 1998.
Dolan, J. W. , Snyder, L. R. , Djordjevic, N. M. , Hill, D. W. , and Waeghe, T. J. , Reversed- phase liquid chromatographic separation of complex samples by optimizing tempera- ture and gradient time. I. Peak capacity limitations. J. Chromatogr., 857, 1, 1999.
Dolan, J. W. , Snyder, L. R. , Wolcott, R. G. , Haber, P. , Baczek, T. , Kaliszan, R. , and Sander L. C. , Reversed-phase liquid chromatographic separation of complex samples by optimizing temperature and gradient time. III. Improving the accuracy of computer simulation. J Chromatogr., A, 857, 41, 1999.
Dolan, J. W. , Snyder, L. R. , Djordjevic, N. M. , Hill, D. W. , and Waeghe, T. J. Reversed-phase liquid chromatographic separation of complex samples by optimizing temperature and gradient time. II. Two-run assay procedures. J. Chromatogr., A, 857, 21, 1999.
Molnar. I. Computerized design of separation strategies by reversed-phase liquid chro-matography: Development of DryLab software. J. Chromatogr., A, 965, 175, 2002.
Swartz, M. E. , Jones, M. , Fowler, P. , and Andrews, M. , Automated HPLC Method Development and Transfer. LCGC, 20(6), 43, June 2002.
Jerkovitch, A. D. , Mellors, J. S. , and Jorgenson, J. W. The Use of Micron-Sized Particles in Ultrahigh-Pressure Liquid Chromatography. LCGC North America, 21(7), 2003.
The Quest for Ultra Performance in Liquid Chromatography—Origins of UPLC Technology, Waters Corporation. See http://www.waters.com/waters/partDetail.htm7ci d=511539&id=38628&ev=10120661&locale=en_US.
Beginners Guide to UPLC—Ultra-Performance Liquid Chromatography. Waters Corporation. See http://www.waters.com/waters/partDetail.htm?cid=511539&id=3862 9&ev=10120684&locale=en_US.
Swartz, M.E. and Murphy, B. , Uplc: An introduction and review. J. Liq. Chrom. Related Tech., 28, 1253, 2005.
Swartz, M. E. and Plumb, R. , Waters Corporation, Milford, MA, unpublished data.
Lukaly, P. , Pfizer, Ann Arbor MI, personal communication.
Swartz, M. E. , HPLC Systems and Components Introduced at Pittcon 2010: A Brief Review. LCGC, 28(5), 376, May 2010.
Corona Charged Aerosol Detector, brochure 70-6454 Rev A, ESA/Dionex, Chelmsford, MA, 2004.
Swartz, M. , Emanuele, M. , Awad, A. , and Hartley, D. , Supplement to LC/GC North America, April 2009. See http://chromatographyonline.findanalytichem.com/lcgc/article/articleDetail.jsp?id=592237&sk=&date=&pageID=2.
Niessen, W. M. A. , Liquid Chromatography-Mass Spectrometry, Chromatographic Science Series, Vol. 97, 3rd ed., Taylor & Francis, London, 2006.
McMaster, M. C. , LC-MS: A Practical Users Guide, Wiley, Hoboken, NJ, 2005.

## Use of cookies on this website

We are using cookies to provide statistics that help us give you the best experience of our site. You can find out more in our Privacy Policy. By continuing to use the site you are agreeing to our use of cookies.