Methods Development
Strong science with a dash of experience

by Angelo DePalma

Chromatography methods development relies principally on sound science, as the hundreds of books and tens of thousands of articles on the subject will attest. Underestimating the value of experience, intuition and personal preference would be a mistake, however.

Solubilizing the active ingredient from a complex mixture can be problematic, since chemical properties often differ tremendously between analyte and matrix. In these situations, where analyte recovery becomes essential, Sutton finds diode array ultraviolet detectors "mandatory" for obtaining accurate analyses. "Diode array UV allows you to check the analyte’s spectrum, as well as the optimal detection wavelength." Because several components may absorb at the same wavelength, that may be different from the analyte’s l max. Diode arrays are usually an expensive option on HPLC systems.

CIA uses UPLC instruments from Waters exclusively because of the instruments’ throughput and reliability. "Most people who develop methods have a backlog of work as it is," Mr. Sutton notes. "Method development takes four to five times as long with conventional LC compared with sub-2-micron systems. We could not have developed as many methods were it not for fast LC."

Life in the fast lane
With so many users acquiring fast LC and GC capability, strategies for transferring methods from traditional to rapid instruments are greatly needed. Alessandro Baldi, product manager for chromatography systems at PerkinElmer (Waltham, MA), likens the process to tuning up a car for a race. "What you do ultimately depends on the type of race," he says. "Everything that is critical with regular LC becomes very critical in fast LC."

The process of porting methods to high-speed LC systems does not begin and end with the column and mobile phase, as one might think. Baldi suggests first examining the volume of injection, how the injection is performed and the nature of the diluent and sample.

Columns with very small particle sizes can operate in high-throughput mode in a conventional LC system, which pumps in the 6 to 7 kpsi range, provided the column is short enough (under about 40 mm). Applications demanding both resolution and throughput require longer columns, and therefore pumps that operate at up to 15 kpsi.

At those pressures, mobile phase viscosity becomes an issue, which in turn makes temperature much more of a factor than with "normal" LC. Dr. Baldi suggests that methods developers "play with the column temperature" and experiment with selectivity versus temperature for a particular column. Also along these lines, one might consider switching to a solvent such as water/methanol, which is less dense than water/acetonitrile and permits higher linear velocities for a given pressure.

Dead volumes are another critical consideration in porting methods from conventional to fast LC. For example, systems with a large solvent mixer volume will provide a clean baseline and very little noise, but will introduce a delay in applying the gradient. Smaller mixers will create gradients much more quickly, but can create a noisy baseline.

Users who require very high-throughput analyses, and who have the luxury of working with concentrated samples, should therefore use the smallest mixer available. Those with very dilute samples, where signal to noise is a known issue, must sacrifice some throughput and employ a higher-volume mixer.

A related issue is flow cell and tubing geometry. Smaller, narrower cells minimize dispersion at the expense of sensitivity. "If you do method development and want to push to the performance edge, you need to select a very small flow cell to minimize dispersion, and short, narrow tubing for the same reason," notes Dr. Baldi. "The smaller the column and internal diameter and flow, the more important tubing size becomes."

For porting slow GC to fast GC instruments, users need first to consider the quantity of analyte being injected. Fast GC squeezes analytes into narrower, taller peaks, so columns require less material, which is achieved by either a smaller injection or a higher split ratio. "The idea is to get a nice, tight band of analyte onto the column quickly without overloading or hanging out in the injection port," notes William Goodman, GC/GC-MS applications specialist at PerkinElmer. "Injection and the injector port are where you can sacrifice peak width if you don’t have the right technique."

Similarly with LC, fast GC methods relying on shorter columns must also reduce column diameters to retain the

Detectors must be able to handle peak widths from fast GC runs, which can be half a second or less compared with 1 to 3 seconds for conventional GC. Fast detectors need to collect data points at 25 per second, preferably 50 per second, to keep up.

Oven temperature, carrier gas flow and mobile phase composition are additional parameters that users should consider when switching from conventional to rapid GC methods. After finding a column that provides adequate separation, users can experiment with temperature programming and increasing the flow rate. Additionally, notes Mr. Goodman, users are switching from helium to hydrogen carrier gas. Hydrogen’s advantages include greater efficiencies across a wide range of linear velocities, and the ability to generate carrier gas inexpensively at the point of use.

Selling point
ESA Biosciences (Chelmsford, MA), a supplier of specialty HPLC detectors and components, also provides customers with methods optimized for specific applications, for example, neurotransmitter analysis. When customers have a specific compound that they need to measure, ESA may develop a quick method to determine the best system configuration, and if the application is suited to electrochemical or charged aerosol detection, the latter an ESA specialty. Finally, the company runs a commercial laboratory that provides fee-based methods development.

According to John Waraska, director of strategic marketing for life sciences at ESA, the skill level of chromatographers has reached a point where, at larger companies, particularly pharmaceutical firms, few analytical scientists require help with methods. "The drug industry has chromatographers who are every bit as skilled as those at instrument companies," he says.

Larger vendors, Waraska notes, have moved from applications and methods support, which are sometimes provided gratis, in favor of paid instrument support. "Companies like ours, which specialize in the application of HPLC detection technology, typically provide more in the way of methods."

Having embraced methods support as a selling point, ESA goes as far as to release kits for specific analytes. It recently developed one, for vitamin D, comprised of an HPLC system, training, sample preparation tools, standards, column, mobile phase and software specially configured for that assay.

Developing methods is a scientific exercise, but intuition and experience play important roles as well. Mr. Waraska relates a story from a previous employer, where he worked side by side with another analytical scientist. No matter what came in, the colleague would first try a C18 column and UV detection, whereas Waraska preferred a cyano column and refractive index detector. The moral: "Start with something you’re familiar with, and you will know very quickly where to go next."

The message from Michael Frank, Ph.D., who supervises new methods at Agilent’s Waldbronn, Germany facility, is to beware of perfect-looking peaks during development. "You can get beautiful peak shapes, perfectly Gaussian, and think you’re done with development," he says. Meanwhile, co-eluting materials can easily lurk beneath those apparently lovely traces.

One way to reduce this possibility of a nasty surprise is to experiment with orthogonal components, for example, detectors and/or columns. Similarly, it might be a good idea to check a method on a sub-2-micron instrument, to substantiate data obtained with larger particle sizes.

Employing a second detector, even if it will not be used during normal runs, can provide peace of mind after the method becomes established. Dr. Frank recommends mass detectors, or "something completely different, like nuclear magnetic resonance." These, he says, will provide a picture of the method that is complimentary to, and quite different from, ultraviolet.

Columns are another area where a bit of experimentation can pay off. Obtaining the same number of peaks using two or three standard chemistry columns, particularly when peaks are analyzed with MS, is a good indication that what you see is what you have. For chiral analytes, consider using a chiral column, even if the product is enantiomerically pure.

Agilent has adapted its popular 1200 series LC for methods development to facilitate the strategy of orthogonality. The methods development instrument supports up to eight columns and fifteen solvents, for a maximum of 288 separation conditions (for binary gradients).

What you see is what you have?
Method validation, either during or right after development, can avoid many problems later on. It assures, notes Mr. Baldi of PerkinElmer, not only the validity of the method for its intended purpose, but that the method will be applicable moving forward.

"Rigorous validation is mostly associated with the pharmaceutical industry," he says, "but many more chromatographers should follow this practice." Users may shy away from validation because it requires some knowledge of statistics, but as Baldi points out, software can handle most of the math.

Validation settles four major issues that haunt poorly developed methods:

• The importance or unimportance of sample stability.

• That the method is repeatable and reproducible on a single system and on different systems, run by different users, at different times. "The goal is to better understand where the variance is coming from," Baldi says.

• Method robustness, for example: How sensitive is a GC method to ambient temperature? To what degree does an LC method depend on pH or buffer? "Methods that lack robustness will not provide reproducible data over time, and will be difficult to export outside the group."

• How the instrument performs over time. Consider instituting a list of performance issues that can be checked every day, or before beginning a critical experiment, to assure that the instrument is behaving as intended.

Conclusion
Chromatography methods development has a reputation, somewhat deserved, for being tedious and time consuming. Automation, systems with multiple columns and detectors, method-specific software, and "fast" GC or LC can help lighten the load somewhat. Published methods on related analytes, trade literature and vendor support also help. But in the end, a chromatographer's experience becomes the critical factor in successful deployment of a new method. The good news is that once a method is established and validated, it can serve chromatographers and the groups they support for many years.

Angelo DePalma is a chemist-turned-freelance writer based in Newton, NJ. He may be contacted at ChromatographyTechniques@advantagemedia.com.

• Agilent 1100 Series diode-array detector SL for optimization of speed, resolution, sensitivity, and spectral performance in Ultra-fast LC
• Alternate Selectivity for Bases in UPLC using a Non-Endcapped Silica Stationary Phase
• Analysis of Sarin Nerve Agent Metabolites by Tandem GC Mass Spectrometry and Positive Ion Chemical Ionization
• CycloGraph Centrifugal Chromatography Device