PerkinElmer sells aerospace business to focus on health sciences 

PerkinElmer, Inc. (Shelton, Connecticut) recently announced that it plans to divest the fluid sciences business segment to increase the company's strategic focus on its health sciences and photonic markets. The company has signed an agreement to sell its aerospace business to Eaton Corporation for $333 million.  

The company also reported that it is in discussions to sell its semiconductor and fluid testing businesses. All planned divestitures are expected to be completed by 2006 with aggregate proceeds of $400 million from the sale of the Fluid Sciences business segment.  

Shimadzu's LC–MS-IT-TOF mass spectrometer wins R&D 100 Award 

Shimadzu Scientific Instruments, Inc. (Columbia, Maryland) recently announced that its hybrid Ion Trap-Time-of-Flight mass spectrometer has received the R&D 100 award from R&D Magazine.  

Shimadzu's LCMS-IT-TOF is designed for a wide range of uses including biomarker studies, metabolomics, proteomics, metabolite identification, environmental analyses, forensics use, and homeland security functions. 

Applied Biosystems sells to Centers for Disease Control and Prevention

Applied Biosystems Group (Foster City, California) and MDS Sciex (Concord, Ontario, Canada) recently announced the sale of 21 API 4000 LC–MS-MS systems to the Centers for Disease Control and Prevention (CDC) and several state health laboratories. These systems will be part of the Laboratory Response Network, providing a validated platform for identifying harmful chemical agents nationwide. 

Utilizing the LC–MS-MS system with validated assays, laboratory personnel can screen for and identify chemicals in human blood or urine samples, including mustard agents, nerve agents, and other toxic chemicals.  

Novasep Process nominated to 2005 CPhI Innovation Awards

Novasep Process recently has been selected to participate to the 2005 CPhI innovation awards selection and will present a conference entitled "Latest Advances in Preparative Supercritical Fluid Chromatographic Separation Technologies" in Madrid in November. 

Peaks of Interest

Tim Wehr is staff scientist at Bio-Rad Laboratories, Hercules, California. Direct correspondence about this column to Direct correspondence about this column to "Directions in Discovery," LCGC, Woodbridge Corporate Plaza, 485 Route 1 South, Building F, First Floor, Iselin, NJ 08830, e-mail lcgcedit@lcgcmag.com.

In some biopharmaceutical companies, CE has virtually replaced gels for electrophoretic analyses. But challenges include management acceptance, technical hurdles, and the need for training.

The previous installment of "Directions in Discovery" discussed the growing use of capillary electrophoresis (CE) to replace gel-based analyses in the biopharmaceutical industry, and pointed out the advantages of CE in terms of obtaining quantitative information with good reproducibility, fully automating routine analysis, and eliminating the requirement for large volumes of toxic reagents (1). The focus of the first column was the replacement of sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) by CE using entangled-polymer sieving systems, a technique known in the biopharmaceutical industry as CE-SDS. This second column will explore the remaining techniques that are replacing gel electrophoresis, including capillary isoelectric focusing (cIEF) and capillary zone electrophoresis (CZE) with laser-induced fluorescence (LIF) detection for glycan analysis. Like CE-SDS, these techniques can be applied to a wide range of samples, and so can be developed as generic or platform methods. 

Gel IEF suffers from the same limitations as those described above for SDS-PAGE. For this reason, many labs in the biopharmaceutical industry have turned to performing IEF in the capillary format.

The initial steps in cIEF are analogous to gel IEF (Figure 1). The sample is mixed with carrier ampholytes and the capillary is filled with the sample plus ampholyte solution. One end of the capillary is immersed in an anolyte solution (typically, dilute phosphoric acid), and the other end is immersed in catholyte (typically, dilute sodium hydroxide). Upon application of high voltage, carrier ampholytes migrate to form a stable pH gradient, and proteins migrate to the region of the gradient where they are isoelectric, and become highly focused. The focusing process typically is complete within a few minutes. Because commercial CE systems use on-tube detection at a fixed point along the capillary, cIEF requires some means of mobilizing focused zones across the detection point. This mobilization process is unique to cIEF. Mobilization can be accomplished by pressurizing the capillary inlet, applying a vacuum to the capillary outlet, or by adjusting the liquid levels at the capillary outlet. This latter technique, termed gravity mobilization, is pictured in the schematic in Figure 1. 

Figure 1:  Schematic of the cIEF process. The upper trace represents the completion of focusing, and the lower traces represent mobilization. Gravity mobilization is pictured here. Note the reduced level in the catholyte reservoir during mobilization and the displacement of anolyte into the capillary. 			

Mobilization also can be accomplished by changing the composition of either the anolyte or the catholyte. This causes the pH gradient to shift along the capillary with time so that proteins move electrophoretically across the detection point. This is variously termed electrophoretic mobilization, ion-addition mobilization, or chemical mobilization. The separation pattern in cIEF is subject to a large number of variables, and obtaining good results requires the use of internal standards, preferably at least two standards bracketing the mobilization time of the analyte proteins. 

The major applications of cIEF in the biopharmaceutical industry are the determination of protein isoelectric point (pI) and monitoring protein charge heterogeneity (2). An example of the first application is presented in Figure 2. Three variants of proteins containing zero, one, or two terminal lysines are resolved by cIEF with internal standards bracketing the analyte proteins. A plot of migration time versus pI permits estimation of the isoelectric points of the three variants.

Figure 2:  Determination of apparent pI values of protein variants differing in the number of terminal lysine residues. (a) Electropherograms of a recombinant monoclonal antibody (rhuMab) and pI markers and (b) plot of pI values against migration times. Open squares, pI markers 7.2, 7.7, and 10.1. Closed circles, antibody variants shown in (a). Figure provided by Genentech, Inc.			

Charge variants can arise from the loss of terminal lysines, deamidation of glutamine residues, or the presence of variable numbers of sialic acids at glycosylation sites. The Fc fusion protein analyzed by cIEF in Figure 3 exhibits approximately 28 charge variants (3). Treatment of the protein with carboxypeptidase (to remove terminal lysines) or neuraminidase (to remove sialic acids) simplifies the cIEF pattern. Moreover, a comparison of the pattern following treatment with the two enzymes reveals that the majority of charge heterogeneity in this protein arises from variable numbers of charged sialic acid residues.

Figure 3:  Determination of charge heterogeneity in an Fc fusion protein using cIEF. (a) Intact sample, (b) sample treated with carboxypeptidase, (c) sample treated with neuraminidase, and (d) sample treated with carboxypeptidase and neuraminidase. Reproduced from reference 3 with permission from Vieweg & Sohn Verlagsgesellschaft GmbH. 			

The requirement for mobilization of focused proteins in a conventional CE instrument lengthens analysis time, can compromise resolution by allowing diffusional band broadening during transition from focusing to mobilization, and can increase the risk of protein precipitation. Imaging cIEF, a relatively new approach being commercialized by Convergent Bioscience (Toronto, Ontario, Canada), eliminates the need for mobilization (4). A UV-transparent capillary is used, and an entire capillary length is illuminated with light from a xenon source and imaged by a charge-coupled device (CCD) camera (Figure 4). Because mobilization is not required, analysis time is much shorter, and the risk of precipation is reduced. In addition, focusing can be monitored in real time, which can be an aid in method development and optimization.

Figure 4:  Schematic diagram of imaging cIEF. Figure provided by Convergent BioScience, Ltd.			

The presence of carbohydrates and their structure is known to influence the biological activity of protein therapeutics, so it is of great importance to characterize the carbohydrate content of a therapeutic monoclonal antibody (Mab) and the batch-to-batch consistency of the glycosylation (5). The occupancy of glycosylation sites and the structure of the oligosaccharide is influenced by the cell line, the conformation of the protein, and the culture conditions. Carbohydrate processing during biosynthesis can result in microheterogeneity with regard to oligosaccharide structure. 

Carbohydrates present a challenge for CE analysis because they have poor UV absorbance even at low wavelengths, and neutral sugars exhibit electrophoretic mobility only under extremely basic conditions. However, derivatization with a charged fluorophore provides a solution to both problems. Aminopyrene trisulfonic acid in the presence of sodium cyanoborohydride reacts with the carbonyl of the reducing end of an oligosaccharide to form a derivative that fluoresces strongly when excited by the 488 line of an argon ion laser. In addition, the sulfonic acid groups provide electrophoretic mobility. The derivatives can be separated readily using CZE. This approach can be used to characterize glycans cleaved from the protein and to analyze monosaccharides produced by hydrolysis of the glycans. This is illustrated in the characterization of oligosaccharides cleaved from the Fc region of a chimeric humanized monoclonal antibody by PNGase (Figure 5). In this particular molecule, a single N-linked glycosylation at the Asn-301 site of the Fc domain consists of a biantennary structure with zero, one, or two terminal galactose residues (G0, G1, G2). 

Figure 5:  Structure of carbohydrates located on the Fc domain of rituximab, a genetically engineered chimeric monoclonal antibody. Reproduced from reference 5 with permission from the American Chemical Society.			

Analysis of the cleaved glycans by CZE with LIF detection readily resolves all four species (5), including the positional isomers of G1 (Figure 6). The separation was performed using reversed polarity (anode at detector end) and coated capillaries to reduce electroosmotic flow (EOF). Because all of the glycans have the same net charge, separation is based upon differences in size and shape. Therefore, differences in the number and position of terminal galactose residues alters glycan mobility, with G0 (no terminal galactose residues) migrating faster than G2 (with two terminal galactose residues). The G1 isomers migrate at intermediate mobility; the faster-migrating G1 species is present at relatively higher amounts. Additional experiments involving seqential enzymatic removal of 

-acetylglucosamine and mannose residues suggested that the fast-migrating major G1 peak was the branch linked to the core structure via the 1–6 linkage (refer to Figure 5).

Figure 6:  Separation of glycans cleaved from rituximab with PNGase. The internal standard is maltoheptaose. Reproduced from reference 5 with permission from the American Chemical Society.			

Current and Future CE Usage in the Biopharmaceutical Industry

This and the previous installment of "Directions in Discovery" have explored the use of CE as a replacement for conventional GE assays, with a focus on generic or platform methods which could be applied to many different products. In the course of preparing this report, scientists at several biopharmaceutical companies were interviewed. Of these, Genentech and Amgen emerged as the leaders in this technology.  

Genentech currently has twenty validated CE assays (2). In the QC environment, this includes lot release testing. In the product development environment, CE assays are used for cell culture development, recovery process design, formulation and stability studies, product characterization including comparability studies, and clinical lot release. The company started their CE program almost 10 years ago, and has over 30 CE systems in use. Currently every new molecule has one or two CE assays. Of the electrophoretic methods in use, 100% of the QC methods are CE, and 90% of analytical characterization methods are CE-based. 

Amgen (6) has validated methods for CE-SDS (primarily for lot release of monoclonal antibody products), cIEF, CZE-LIF (for oligosaccharide characterization), and CZE-UV (for isoform determination). The company has 40 CE systems at four sites. At Amgen, CE has replaced gel electrophoresis for QC assays. 

Although CE is routinely used in many biopharmaceutical companies, the technology is not without challenges. Management acceptance, which was certainly an issue when CE programs were initiated in the industry in the mid-to-late 1990s, is still a concern. Questions about analytical performance and vendor commitment still arise. Practitioners desire better chemistries, including more stable capillary coatings, improved derivatization procedures, and more versatile polymer sieving matrices. Training surfaced as a serious problem as well. Personnel turnover is a fact of life in this industry, and academic labs are not providing a pipeline of students with CE expertise. 

In spite of these concerns, CE is now — and is likely to remain — a core analytical technology in the development and manufacture of biotherapeutics. In the future, CE will most likely evolve away from general-purpose capillary-based instruments towards miniaturized chip-based platforms and specialized applications-directed systems. 

"Directions in Discovery" editor Tim Wehr is staff scientist at Bio-Rad Laboratories, Hercules, California. Direct correspondence about this column to "Directions in Discovery," LCGC, Woodbridge Corporate Plaza, 485 Route 1 South, Building F, First Floor, Iselin, NJ 08830, e-mail 

(2) S. Ma, Abstract L05-K4-M, presented at the 18th International Symposium on Microscale Bioseparations, New Orleans, Louisiana, 2005. 

(6) A. Guo, M. Han, P. Jones, and A. Balland, Abstract L05-O12-M, presented at the 18th International Symposium on Microscale Bioseparations, New Orleans, Louisiana, 2005. 

Directions in Discovery

Capillary Electrophoresis in the Biopharmaceutical Industry: Part II

John Dolan has been writing "LC Troubleshooting" for 

 for over 25 years. One of the industry's most respected professionals, John is currently a principal instructor for LC Resources, Walnut Creek, California He is also a member of 

Read no further. Want your liquid chromatography (LC) column to last forever? Never open the box! Not very practical advice, because the fact is that all columns fail sooner or later. However, there are some tricks that will help you extend a column's life. This month's "LC Troubleshooting" discusses some practices that you might find useful to make your column last longer.

One of the most common symptoms of column failure is high column back pressure. Back pressure rises as particulate matter accumulates on the inlet frit of the column. Columns packed with 5-μm packing use 2-μm porosity frits to contain the packing. When 3-μm packings are used, the inlet frit can be either 2 μm or 0.5 μm, depending on the manufacturer. To a certain extent, particulate matter buildup on the frit is unavoidable, but there are several simple practices that can slow this process.

Over the lifetime of the average column, tens, or even hundreds of liters of mobile phase pass through the column. Any particulate matter in the mobile phase usually ends up on the column frit. Filtration of all mobile phase through a 0.5-μm porosity membrane filter should eliminate particulates originating in the mobile phase. However, nearly all high performance liquid chromatography (HPLC)-grade solvents are filtered through a 0.2-μm porosity filter during manufacturing, so additional filtration (especially with a more porous filter) might not be beneficial. HPLC-grade water prepared with a water purification system generally includes a 0.2-μm final filter, so water should not require additional filtration, either. In my laboratory, our standard operating procedure allows us to bypass the filtration step if mobile phase is prepared only from HPLC-grade liquids (acetonitrile, methanol, water, and so forth). If solids, such as buffer salts, are used in the mobile phase, filtration is required. Even though the buffer might be chemically pure, physical contaminants, such as fibers scraped from the bottle cap liner, often are present. When in doubt, filter the mobile phase. And don't forget to cover the mobile phase reservoirs to prevent dust from entering. A loose screw-cap or a piece of aluminum foil works well; do not use one of the polymeric laboratory film products — it might not be compatible with the mobile phase solvents. 

Pump piston seals and injector valve rotor seals are made of polymeric materials that gradually wear over time. Fragments of the seal eventually make their way downstream and end up on the column frit. Seal wear cannot be prevented, but it can be minimized by ensuring that buffers and other salts are not allowed to stand in the LC system when it is not in operation. These salt solutions can evaporate on the piston surface behind the pump seal, creating an abrasive coating of salt crystals that will damage the seal over time. It is best to flush the LC system with nonbuffered solvent (for example, acetonitrile–water) before shutdown so that any buffer residues are removed. If high-salt mobile phases are used, such as buffers > 50 mM in concentration or ion-exchange mobile phases, it is wise to use the seal-flush feature available on most LC pumps. This is simply a liquid path behind the pump seal that can be flushed with water or water–alcohol to remove any salt residues. There is not a similar flushing feature available for most sample injectors, but if you avoid the use of nonvolatile buffers in the autosampler wash, you will reduce the likelihood of rotor abrasion. Piston seals can last for 6–12 months — check your system maintenance records to determine a reasonable replacement cycle. I recommend replacing the pump seals before they fail, or at least once a year during annual maintenance. Injector rotors can last for 20,000 cycles or more. If your autosampler keeps track of the number of rotor cycles, you could include replacement in a preventive maintenance plan, based upon the number of cycles. Rotor seal failure is so rare in my laboratory that we just wait until we have problems before replacing the seal, which can be after several years of use. 

The frit on the tubing connecting the reservoir to the pump is an additional precaution to keep particulate matter from entering the system. This frit should not be considered an alternative to filtering the mobile phase, but rather a safety device to remove any dust that might enter the reservoir or other inadvertent contamination. A 5-or 10-μm porosity frit will serve this purpose well, without generating significant resistance to flow that might result in pump starvation. A simple test for frit blockage is to see if mobile phase siphons freely through the frit and tubing. Just disconnect the inlet tubing at the proportioning manifold (low-pressure-mixing system) or pump inlet (high-pressure mixing). With mobile phase siphoning through this tubing, I like to see delivery of 10 times the pump requirement. For example, if you normally operate the pump at 1–2 mL/min, accept no less than 10–20 mL/min through the siphon. If the flow is less than desired, replace the frit. 

Particulate matter from injected samples is the most common source of column frit blockage. Some workers filter all samples before injection. Individual syringe-type filters can be used or if you are using automated sample preparation, 96-well filter plates are available. I have mixed feelings about the use of sample filters because of the expense, inconvenience, and validation implications. Filtration can add a dollar or more to the cost of sample preparation, not an insignificant cost for a laboratory performing routine analysis on hundreds or thousands of samples per month. 96-well filter plates are easy to automate, but the use of syringe filters is time-consuming. At the top of my list, though, are considerations of sample loss and validation. Every filter has a finite hold-up volume, so sample loss occurs when filters are used. This is not important if you have several milliliters of sample, but if your sample is limited to 100 μL or less, a significant portion of the sample can be lost in the filter. Does anything leach from the filter that might interfere with the chromatographic assay? Does the filter selectively adsorb anything from your sample that might affect the assay results? These concerns require that the filtration process be included in the method validation process and generally preclude selective filtration of some samples and not others. As an alternative, the samples can be centrifuged in a bench-top centrifuge (for example, at 10,000 rpm) for 5 min. Following filtration, the sample is transferred to the injection vial or plate. 

One of the least expensive and most effective ways to protect the column from particulate contamination is to use an in-line filter. This 0.5-μm porosity frit is mounted just downstream from the autosampler, ahead of the column (and guard column, if used). The in-line frit will trap particles that would otherwise contaminate the column inlet frit. When the system back pressure rises, change the in-line frit and you should be back in business. These devices are inexpensive and can be purchased from any LC supplies vendor. Some manufacturers design their columns so that the in-line filter frit is built into the analytical column (sometimes called a double frit). Disposable tube fittings that contain a frit also are available. Be sure to keep plenty of replacement frits on hand. I consider the in-line frit a mandatory part of every LC system. 

A guard column is simply a miniature LC column that contains packing material similar to that in the analytical column. It also has a frit at each end, and the frits trap particulate matter. However, the porosity of the guard column frits generally is the same as the analytical column (for example, 5 μm), so it is beneficial to use an in-line 0.5-μm frit even if a guard column is used. 

It might or might not be possible to change the frit at the column inlet. Some columns use pressed-in frits that cannot be replaced. 10–15 years ago, when columns were not packed as well as they are today, many workers routinely replaced the column inlet frit. In fact, many columns came with an extra frit or two, just for this purpose. Today's columns, however, are packed at higher pressure. The resulting column will generally ooze packing if the frit is removed — this can irreversibly damage the column, so frit replacement is not recommended for most analytical columns today. An alternative is to backflush the column. Just remove the column from the system and connect the outlet end to the pump and pump 10–20 mL of solvent through the column to displace any particles from the frit. (Be sure to pump directly from the column to waste and not through the detector to keep contaminants out of the detector.) After reverse-flushing, I prefer to leave the column in the reversed direction — I have seen many cases in which the pressure rise returned when the column was placed back in the normal flow direction. Most 5-μm particle silica-based columns are amenable to reversal without damage. However, some manufacturers of 3-μm particle columns use a 2-μm frit at the inlet end and a 0.5-μm frit at the outlet of the column; reversal of these columns can result in leakage of packing and damage to the column. 

Always consult the column manufacturer's care-and-use instructions before reversing any column. 

There are two types of contamination that will ruin a column — physical and chemical. The previous discussion presented several techniques to minimize problems from physical contamination by particles from various sources. Next, let's look at what can be done to minimize chemical contamination. 

Before considering sample cleanup, it should go without saying that the mobile phase should be compatible with the column. For the reversed-phase, silica-based columns that most of us use on a daily basis, pH is the most important parameter. As a general rule, the bonded phase is hydrolyzed from the column packing at pH < 2 and the silica dissolves at pH > 8, but these are not hard-and-fast rules. Some columns are specifically designed for use under high-or low-pH conditions. However, unless you know otherwise, keep the mobile phase pH within the pH 2–8 region (or more conservatively, pH 2.5–7.5) for maximum column stability. Again, consult the manufacturer's care-and-use instructions for specific directions for your column. 

The whole purpose of the chromatographic process is to separate different chemical compounds from each other. Some are more retained and others are less retained. Poorly retained materials generally are not a problem, in terms of column contamination. These materials wash through at the solvent front or early in the chromatogram, and can cause problems with quantification, but they do not remain on the column. On the other hand, materials that are strongly retained can build up over time and change the separation or ruin the column. Let's consider these strongly retained materials in two classes. One group contains compounds that are strongly retained, but can be washed off the column with a strong solvent. The other group includes materials that cannot be removed from the column. We will consider only reversed-phase separation, in which both groups are less polar (more hydrophobic) than the sample.  

Materials that are strongly retained under analytical conditions often can be removed from the column when a strong solvent, such as 100% acetonitrile or methanol, is used to flush the column. For isocratic methods, this flushing can be done at the end of a batch of samples before the LC system is shut down. Generally, 10–20 mL of 100% strong solvent will be sufficient to remove strongly retained materials. With gradient-elution methods, each run can be concluded with a ramp or step-gradient to strong solvent to remove hydrophobic materials. Sometimes, a different mobile phase condition, such as a change in pH or buffer salt strength will be beneficial. Usually one of these techniques will result in satisfactory removal of late-eluted materials.  

Some samples contain constituents that bind irreversibly to the column or have such large retention characteristics, even under strong solvent conditions, that they might be considered permanently bound to the column. Often these compounds are proteins, lipids, and other high-molecular weight, hydrophobic compounds if the sample is biological in origin. If the sample is chemical in nature, polymers and other high-molecular-weight, hydrophobic compounds also might be present. At other times, sample constituents might not be soluble in the mobile phase and precipitate at the head of the column. Sometimes, reverse-flushing the column with strong solvent will remove some of the strongly bound material, but often this practice is only marginally effective. As chemical junk builds up on the column, packing pores can become blocked, the packing surface can become coated, or active retention sites on the stationary phase can be permanently blocked. All these actions change the retention characteristics of the column. Eventually, the column must be replaced.  

The best way to deal with strongly and irreversibly retained materials is to remove them before injecting the sample. Just as we use physical filtration techniques to remove particulate matter from the sample and mobile phase, we can use chemical cleanup steps to remove unwanted compounds from the sample.  

Perhaps the simplest way to clean up the sample is to use a guard column. A guard column of the same packing as the analytical column will trap strongly retained material in the same manner as the column does. Sometimes, the guard column can be flushed to extend its life (flush the guard column to waste or the contaminants will be flushed onto the analytical column). However, the guard column is designed as a disposable device and should be discarded regularly. You will have to determine for your samples how long the guard column will last, but guard column lifetimes of 50–100 samples or more can be reasonable. Some workers replace the guard column on a time-basis, such as once a week. 

The best way to cleanup the sample is to use some kind of chemical or chromatographic cleanup. The most common techniques are precipitation to remove proteins, liquid–liquid extraction, and solid-phase extraction. A complete discussion of sample cleanup is beyond the space allowances for this installment of "LC Troubleshooting," so we will leave that for another time. Consult Ron Majors' "Sample Preparation Perspectives" column for a regular discussion of sample preparation. 

There are quite a few techniques that can be used to extend the useful life of an LC column. However, the column should be considered a disposable item. This might be difficult to accept, considering that the typical column costs $500 or so. Think of the cost of the column in view of the overall analysis costs. Typical sample charges for an LC–UV assay by a contract research laboratory are in the range of $50 per sample. If the column lasts for 500 injections, this means that the column contributes $1 to this total, or 2% of the cost. Many workers get 1000–2000 samples analyzed before a column dies, so the cost per sample would be even less. So, use common sense — institute reasonable preventive strategies, but do not be mislead by investing more in sample cleanup than you recover in extended column life. After all, you really shouldn't expect the column to last forever.

"LC Troubleshooting" Editor John W. Dolan is Vice-President of BASi Northwest Laboratory of McMinnville, Oregon; a Principal Instructor for LC Resources, Walnut Creek, California; and a member of LCGC's editorial advisory board. Direct correspondence about this column to "LC Troubleshooting," LCGC, Woodbridge Corporate Plaza, 485 Route 1 South, Building F, First Floor, Iselin, NJ 08830, e-mail 

LC Troubleshooting

No column lasts forever, but in this month's installment of "LC Troubleshooting," John Dolan offers some tricks to help extend a column's life.

Extending Column Life

Guest author Rick Parmely is the Director of Technical Training at Restek Corp., Bellefonte, Pennsylvania, and travels domestically and internationally presenting seminars to teach and train Restek customers the fundamentals of gas chromatography. He has also taught college level chemistry.

Capillary gas chromatography (GC) is a well-established separation technique. Guest author Rick Parmely reviews some of the capillary column basics by investigating a standard test mixture and observing some of the chromatographic effects that can affect peak response, peak shape, and column bleed. He provides guidelines for obtaining the best performance from a capillary column. He discusses peak tailing, column overload, ghost peaks, and column bleed and speculates on "when to give up." He concludes with a set of fundamental steps to be used in achieving better GC analyses.

Fused-silica capillary columns are used routinely in a wide variety of gas chromatography (GC) applications ranging from environmental analysis to food testing to pharmaceutical product quality assurance. What is a fused-silica capillary column? What problems can develop with a fused-silica column that might affect data quality and analytical results? How can I address and correct problems when they do occur, and as importantly, when should I give up the fight? Finally, what fundamental steps will lead to better analysis when using this technology? These questions will be addressed in this article.  

A fused-silica capillary column is a narrow bore, open tube coated–bonded on the inside with a polymer stationary phase and on the outside with a polyimide polymer to add strength and flexibility. Liken this to a 4-in. sewer drain pipe in your basement. The rigid white plastic outside is like the polyimide-coated fused silica, while the ubiquitous inside layer of slime can be likened to the polymer stationary phase coating applied by the column manufacturer. For the most part, the inside space is wide open, both in your sewer pipe and in a capillary column. A fascinating article by Steve Griffin, founder of InnovaQuartz, Phoenix, Arizona, a premier manufacturer of fused-silica capillaries, details the history, manufacturing process, and inherent quality issues of fused-silica capillaries (1). In that article, Griffin notes, "Some of these problems include variations in capillary bore and outer dimensions, ovality, and random brittleness, and surface activity." The last mentioned issue of surface activity could describe a wide variety of polymer coating issues. These include polymer-component activity, poor coating efficiency, polymer contamination, and others.  

 (2), Gene Barry notes, "An ideal gas chromatographic column is considered to have high resolving power, high speed of operation, and high capacity." How is "resolving power" assessed and why is it important? Columns that resolve well are said to be efficient, and typically, manufacturers will report a number tied to efficiency that is related to the number of resolving plates that can be theoretically measured during testing of the column. Column efficiency in the most fundamental terms is usually related to some measure of the column's effective theoretical plate number 

As the equation indicates, efficiency can be related directly to the adjusted retention time (

) of the peak and is inversely proportional to peak width at its base (

). Thus, as analytes rapidly move through the column as very narrow bands, they are expressed by the detector as narrow peaks, indicating large numbers of theoretical plates and a high-efficiency column. What does this mean in practical terms? The trend these days is to shorter and shorter runs, meaning that peaks must become even narrower. Gains in column efficiency along with good chromatography practice support this trend.  

In Figure 1, chromatographic peaks are narrow and sharp, indicating high efficiency. The experimental conditions used for this chromatogram are shown in Table I. In subsequent figures, we will use this same test sample and chromatographic conditions to indicate potential problems that can be encountered when using capillary GC.  

Efficiency is important but, in practice, chromatographers are interested in resolving their peaks of interest. So, chromatographic resolution, depicted in Equation 2, is most important or better yet, resolution per unit time — the best resolution in the shortest possible time. Table II not only defines the variables affecting chromatographic resolution, but also gives some hints for chromatographers on how to influence efficiency in separations.  

But there are other problems that can be discerned from the size, shape or nature of the chromatographic peak. Next, we will examine a variety of peaks and see what their relationship is to our gas chromatographic separation.  

 Table II:  Variables affecting GC resolution			

Three types of response problems are observed in GC laboratories:									

A higher-than-expected response results when more sample is delivered on-column and subsequently is detected. A broad, tailing solvent peak with a flat top is the extreme but common example of this type of response; but other components in the injected mixture also might demonstrate this problem. No response usually indicates a lack of sample, the sample did not reach the detector due to a leak, total adsorption, or the sample totally broke down into something else in the injection port or on-column. Of these response problems, low response can be the most difficult to diagnose. Low response can be caused by partial loss of sample due to sample preparation issues (volatiles vaporizing or semivolatiles precipitating in solution), poor injection techniques, or poor injector vaporization, to name just a few causes. In short, low response can be difficult to diagnose and fix but can have huge impact on calibration and, thus, on accurate quantification. Figure 2 shows an example of low response observed for our test sample. Note that all peaks are shorter, compared with the chromatogram in Figure 1. In this case, the chromatographer inadvertently injected less sample than desired due to poor attention to the amount of solution present in the autosampler vial. He simply did not inject the desired amount.  

Figure 2: Chromatogram showing less response due to sample injection error.			

At times in GC, perfect Gaussian shapes are not obtained. What causes these peak distortions, and more importantly, what can be done about them? Let us examine three scenarios that might lead to poor peak shapes and see what can be done to make improvements. 

Tailing peaks generally indicate interaction of a sample component with something along the sample pathway (an active surface), or alternately, tailing peaks might indicate a turbulent flow in some portion of the flow stream as the sample moves from the injector to the detector. Interactive sites, or activity, in the sample pathway could be in the injection port, in the form of a septum particle or a poorly deactivated inlet liner. Problems in the injection port are the sometimes a cause of tailing peaks (as well as other problems), but other sites in the sample pathway — transfer lines, valves, and so forth — also can be interactive. The practical suggestion is to ensure that the sample pathway is as inert as possible. Proper physical installation of the column also is critical. A poor column cut followed by poor installation in either the injector or detector can be disastrous. 

If you are using gas sampling systems or dual column configurations, special care must be taken to minimize dead volumes that can cause nonlaminar flow in the sample stream. Otherwise, peak tailing can result. Late-eluted compounds also might tail when column activity is present due to poor phase coverage, poor initial deactivation of the fused silica surface, or simply incompatibility with the stationary phase. The example shown in Figure 3 is the basic compound dicyclohexylamine, which is the last of the late-eluted probes and tails severely because of a liner activity problem. In this case, septum coring has deposited reactive septum particles inside the inlet liner.  

Figure 3: Chromatogram showing peak tailing of basic test probe.			

Have you ever returned to your shopping cart in the supermarket aisle and found an extra item in it? Similarly, extra material can show up at the detector from a variety of sources. Like the extra can of soup in your shopping cart, we need to backtrack to find out where it came from and how to eliminate it. Ask yourself the following questions: 					

Check your sample preparation procedures. Have these procedures eliminated all possible side contaminations from the matrix? 

What about the injection made by syringe? Is the sample vial, rinse solvent, or syringe itself contaminated?

What about previous chromatographic runs? Were you careful not to inject too much sample that potentially could backflash into split vent and gas supply lines? Remember, the inlet liner has a limited internal volume (fixed by the geometry), and when exceeded by a vaporized sample, material can spill into other parts of the chromatographic system. Materials condensed into these lines eventually will make it back to your column and into later chromatograms.

Figure 4 shows an example of the observation of ghost peaks. A large volume injection of hydrocarbons was run before analyzing our test mixture. Carryover was observed in the form of extra peaks just above the baseline noise in the chromatogram. Rigorous cleaning of all injection port parts, including the connecting gas lines, might be necessary to address this problem.  

Figure 4: Chromatogram resulting from sample carryover.			

				Fronting peaks generally can be attributed to column overload; that is, putting more material on-column than the column, amount of stationary phase, or thermal conditions will allow. The peak shape is "shark-finned" with a gradual rise from baseline to a sharp peak maximum and rapid dropoff to baseline. In Figure 5, the first peak provides an extreme example of this "shark-fin" shape, while the remaining peaks show similar but somewhat reduced behavior. What causes overloading and how can it be eliminated? It is generally believed (3) that the normal stationary–mobile phase equilibrium shifts toward higher concentrations on the stationary phase. Thus, at the detector, small amounts of nonretained solute are detected first and gradually the retained portions "catch up" to develop into a "normal" chromatogram as seen in Figure 5. How can overloading be corrected? The problem lies in the equilibrium within the column. Preventing overload is accomplished by thickening the phase, increasing the column internal diameter, lowering the injected amount of the overloaded component(s), or increasing the column temperature or a combination of these factors. Any of these practical steps can be taken to correct the problem.  

Figure 5: Chromatogram resulting from column overload.			

Much as been written previously about stationary phase bleed or "column bleed" (4,5). In most cases, measurable "bleed" signal at the detector is simply column phase material that has broken loose and is being detected as any sample component might be. A typical column bleed profile is depicted in Figure 6

Figure 6: Temperature-programmed chromatogram showing column bleed.			

How can we address the problem of column bleed? Recognize first that all columns bleed to some degree and you are on your way to understanding the problem. Bleed is like having the outermost layers of your skin slough away as your cells naturally age and die. Aging occurs at different rates in people; some, because of living in intense sunshine, might be said to "accelerate" the aging process. When a properly bonded column is manufactured and tested, this loss of stationary phase is minimal. However, with time and heavy use the amount of stationary phase that sloughs out of the column increases, and an increasing bleed signal (baseline) is observed. What can be done about it? For a typical analysis conducted at temperatures below the column's maximum operating temperature, the amount of bleed might be barely above baseline and not interfere with your analysis. However, if high-temperature analysis is performed with very dilute samples where high sensitivity is needed, a high baseline rise might be a factor in accurate detection and quantification. At this point, bleed could be addressed by further column conditioning, usually accompanied by clipping one-half meter of heavily used column from the inlet end of the column. In this way, any nonvolatile residue retained at the inlet of the column, which could contribute to the bleed signal, also will be removed. Final conditioning should be done only after leak checking with an electronic leak detector and conditioning at twenty degrees above the maximum analysis temperature, or at the column's maximum operating temperature, whichever is lower. Leaks in the GC flow stream can also allow oxygen into the carrier gas which is known to cause stationary phase degradation causing column bleed (4). 

"When should I give up?" you might ask. That depends how effective you have been when attempting to correct the previously mentioned column efficiency, peak shape, or column bleed issues. Any one of these challenges could "sink your smooth sailing chromatography boat" since any of these, if not successfully met, could lead to inaccurate and unacceptable results for your laboratory or your customer. 

There is an upside. Think of the education you and your staff have acquired while trying to identify and troubleshoot these problems. As you troubleshoot problems, isolate the potential causes, and then fix the problem, you have gained immeasurable skill in the science and practice of chromatography. Additionally, the training you have received has equipped you to accept the next challenge, to fix the next problem, bound to occur sometime down the line. 

Finally, what essential things have been learned from these experiences? First, do not assume GC gases from the cylinder or central gas supply are as pure as is required for high-sensitivity capillary GC. Use moisture, oxygen, and hydrocarbon traps to effectively and economically remove contaminants from carrier, makeup, fuel, and oxidizer gas streams. Evaluate these traps often and replace as needed. Second, become fanatical regarding system leaks. An electronic leak detector is the best tool a gas chromatographer can have in the lab but make sure you use it correctly and effectively. Use it prestigiously in all installations and after all maintenance procedures. And, third, remember that it usually is not a unique list of samples or an aggressive solvent system that destroys a column. Rather, fused-silica capillary column longevity also stems from proper installation and maintenance resulting from following simple but effective installation procedures provided by column and instrument manufacturers. See John Hinshaw's "The Top Ten Habits of Successful Gas Chromatographers," which provides an excellent summary of such procedures (6). Table III is a listing of these 10 points.  

Table III:  The top ten habits of successful gas chromatographers (6)			

Column Watch

guest Columnist Rick Parmely takes a look at some capillary column basics of a well-established separation technique. He discusses peak tailing, column overload, ghost peaks, and column bleed, and speculates on "when to give up."

Successful Gas Chromatography Using Fused-Silica Capillary Columns

Dr. Agnes M. Rimando received the Ph.D. degree in pharmacognosy from the University of Illinois at Chicago in 1993. She is currently a Research Chemist with the USDA ARS, Natural Products Utilization Research Unit, Oxford, Mississippi. She is serving as the 2005 Vice-Chair of the American Chemical Society Agricultural and Food Chemistry Division.

Dr. Robert Cody received the Ph.D. degree in analytical chemistry from Purdue University, West Lafayette, Indiana, in 1982. He is currently a Product Manager of mass spectrometry products at JEOL USA, Inc., Peabody, Massachusetts, a member of the American Chemical Society (Analytical Division) and the American Society for Mass Spectrometry.

There is growing evidence that blueberries have powerful disease-fighting properties. The diminutive, dusky blue fruit recently has caught the attention of researchers and is now attributed with such human health benefits as high antioxidant activity (1), antiaging properties (2), the ability to improve eyesight (3) and brain function (4), and lipid-lowering properties (5). 

Recent studies have correlated the high antioxidant activity of blueberries with their total phenolic content. Phenolics include a group of compounds called stilbenes. This article discusses a study to determine the presence of the stilbenes resveratrol, pterostilbene, and piceatannol (Figure 1) in blueberries by gas chromatography–mass spectrometry (GC–MS) in a selected ion monitoring (SIM) mode. 

 berries growing wild were collected at various sites, while other samples were grown commercially. Samples (approximately 50-g fresh weight) were collected randomly and placed in chilled containers pending transport to the United States Department of Agriculture (USDA), Agricultural Research Service (ARS), Small Fruits Research Station (SFRS), Poplarville, Mississippi. The berries were freeze-dried and sent to the Natural Products Utilization Research Unit, Oxford, Mississippi. The samples from Canada were sent as extracts that had been passed through a C18 column. All samples were stored for about six months in a cold room maintained at 4 °C until extraction and analysis. 

Extraction of berries from U.S. locations:

 Berries were extracted using an ASE apparatus (Dionex Corporation, Sunnyvale, California). A 1-g amount of lyophilized berry was mixed with about 10 g of purified sand and loaded in the extraction cartridge. Purified sand was further added to pack the extraction cartridge fully. Extraction was performed with the following parameters: heat, 5 min; static, 10 min; flush volume, 100 mL; purge, 90 s; pressure, 1000 psi; temperature, 40 °C; extraction solvent, 40:40:20:0.1 (v/v/v/v) methanol–acetone–water–acetic acid (four cycles). The extract was concentrated under vacuum using a rotary evaporator to remove the organic solvents. The resulting aqueous solution was extracted with 1 mL of ethyl acetate (three times). The combined ethyl acetate extract was evaporated to dryness under a stream of nitrogen. 

 A 50-g amount of frozen berries was mixed with three volumes of 40:40:20:0.1 (v/v/v/v) methanol–acetone–water–formic acid, kept for 30 min, then ground in a Virtis homogenizer (The Virtis Co., Gardiner, New York) in a stainless steel flask for 2 min. The extract was filtered through an 11-cm diameter G6 glass fiber filter, and the residue was extracted a second time with 60 mL of extraction solvent, kept for 30 min, then ground for 2 min and filtered. The filtrates were combined into a tared round-bottom flask, and the organic solvents were removed under vacuum. The remaining aqueous portion was freeze-dried, suspended in approximately 10 mL of water, and loaded on a 10-g C18 column (Waters Scientific, Mississauga, Ontario, Canada). The retained components were eluted with about 50 mL of 40:40:20:0.1 (v/v/v/v) methanol–acetone–water–formic acid. The organic components of the solvent were removed under vacuum, and the aqueous portion was freeze-dried. 

Analysis of resveratrol, pterostilbene, and piceatannol:

 A 1-mg amount of sample was derivatized with 100 μL of a mixture of 3.5 : 1 : 0.5 (v/v/v) 

-(trimethylsilyl)trifluoroacetamide–dimethyl formamide–methanol in a 2-mL GC vial. The vial was capped and heated at 70 °C for 1 h. After the vial cooled to room temperature, 2 μL was injected and analyzed by GC–MS on a JEOL GCMate II benchtop magnetic sector MS system (JEOL USA, Inc., Peabody, Massachusetts). The GC temperature program was as follows: initial temperature, 150 °C; increased to 260 °C at a rate of 25 °C/min; then increased to 270 °C at a rate of 1 °C/min; increased to 320 °C at a rate of 60 °C/min; and held at this temperature for 2 min. The GC capillary column was a 3 m × 0.25 mm, 0.25-μm, ZB-50 (Phenomenex, Torrance, California). The carrier gas was ultrahigh purity helium at a 1-mL/min flow rate. The inlet (splitless), GC interface, and ion chamber temperatures were 250, 250, and 230 °C, respectively. 

Figure 2:  Mass chromatograms of: (a) resveratrol, (b) piceatannol, and (c) pterostilbene showing major fragmentations. Reprinted with permission from the Journal of Agricultural and Food Chemistry. Copyright 2004 American Chemical Society..			

The analysis of resveratrol, piceatannol, and pterostilbene was performed using SIM mode (retention times 8.7, 9.5, and 10.2 min, respectively (Figure 2). Resveratrol was monitored at 

 444 (428, 370, and 354 as qualifier ions). Pterostilbene was monitored at 

 328 (312, 296, and 280 as qualifier ions). The qualifier ions were selected as being the most abundant ions in the mass chromatogram of the respective standard samples. Piceatannol was monitored at 

 532 (516, 443, and 428 as qualifier ions) (Figure 3). Quantitation was done using external standards of commercial samples of resveratrol and piceatannol and a synthetic sample of pterostilbene. The limits of detection for resveratrol, piceatannol, and pterostilbene were 1, 21, and 18 ng/g dry sample, respectively, and limits of quantitation were 23, 69, and 61 ng/g dry sample, respectively. 

Figure 3:  Gas chromatogram of (a) mixture of standard samples of resveratrol (1), piceatannol (3), and pterostilbene (2) with retention times of 8.7, 9.5, and 10.2 min, respectively; (b) extract of V. corymbosum L. (highbush blueberry) cv. Bluecrop (sustainable) from Corvallis, Oregon; and (c) extract of V. ashei Reade (rabbiteye blueberry) cv. Tifblue from Lamar Co., Mississippi. Reprinted with permission from the Journal of Agricultural and Food Chemistry. Copyright 2004 American Chemical Society.			

Samples from four locations in North America, representing selections and cultivars of 10 species of 

 fruits were analyzed using a GC–MS SIM technique. By monitoring only masses that are characteristic of the target compounds, this method allowed for a more selective detection and more accurate determination of the stilbenes in the samples. Variable amounts of resveratrol, pterostilbene, and piceatannol were found in extracts of whole berries of the 

 species randomly collected from Canada and the states of Mississippi, North Carolina, and Oregon. 

Table I:  Resveratrol in Vaccinium and Vitis samples			

Resveratrol was found in all species of the 

 berries that were analyzed in this study, varying in levels from 7 ng/g to as high as 5884 ng/g dry sample (Tables I and II). The highest level compares with content found in grapes, which ranged between 2475 and 6471 ng/g dry sample. Higher levels of resveratrol were found in the berries from Canada, as compared with those from North Carolina, Mississippi, and Oregon. Lingonberry, obtained from Nova Scotia, was found to have the highest content of resveratrol, almost the same as that found in grapes. Although a comparative extraction study was not conducted, not being the objective of this study, differences in extraction procedures can account for the higher resveratrol content in samples from Canada. 

Table II:  Resveratrol in Vaccinium and Vitis samples from Canada			

 berries is of no surprise, as resveratrol seems to be widespread in the plant kingdom. The variability of the resveratrol content of 

 berries in this study could be attributed to biotic and abiotic stresses such as UV light and 

 infection (6) and with time of harvest, environmental and climatic conditions, and plant developmental stage (7–10). It is unlikely that storage conditions played a factor in the variation of levels of resveratrol found in this study. It has been reported that resveratrol content in grape pomace and skins stored over five months in the dark at ambient temperatures varying between 5 and 18 °C showed no change (11).  

Pterostilbene was found in three samples: two cultivars of 

 (rabbiteye blueberry from Mississippi) and in 

 (deerberry from North Carolina) at levels of 99–520 ng/g dry sample. Piceatannol was found in deerberry, highbush blueberry, and blackberry at levels of 138–422 ng/g dry sample (Table III).  

Table III:  Pterostilbene and Piceatannol in Vaccinium samplesa

In grapes, pterostilbene and piceatannol generally are found in very minute quantities and might be completely absent. These two compounds were not found in the grape samples analyzed in this study, even with the sensitive GC–MS SIM method used. Pterostilbene has been detected in fungus-infected grapes at levels of 0.2–4.7 μg/g fresh weight of skin (12) and also in healthy grape berries of the Gamay and Pinot Noir variety at levels of 14–74 ng/g and 120–530 ng/g fresh berries, respectively (13). Pterostilbene is the only known stilbene in the genus  

 and was found to be one of two antidiabetic compounds isolated from the heartwood of 

 (14). Piceatannol has been found in significant amounts (0.052 μg/g fresh wt) in 

 cv. Cabernet Sauvignon (15). Like resveratrol, the piceatannol content in grapes is increased by UV-C irradiation. Piceatannol appears to be present in diverse plant groups: 

It is of great value to have shown that some small fruits of 

 such as blueberries contain these stilbenes that are known to possess health beneficial (such as cancer chemopreventive [20,21], antioxidant [22], and hypolipidemic [23]) properties. It has been hypothesized that additive and synergistic effects of complex mixtures of phytochemicals, instead of a single component, provide the health benefits derived from fruits and vegetables (24). The stilbenes in 

 berries could add to or act synergistically with other phytochemicals in the berries to produce a more effective complement of protection against coronary heart disease and provide cancer chemoprevention. 

The sensitive GC–MS SIM method employed in this study enabled the accurate determination of resveratrol, pterostilbene, and piceatannol that showed the presence in varying amounts of these compounds in 

 berries. This finding adds value to these small fruits. 

 Mention of trade names or company commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA. 

Dr. Agnes M. Rimando received the Ph.D. degree in pharmacognosy from the University of Illinois at Chicago in 1993. She is currently a Research Chemist with the USDA ARS, Natural Products Utilization Research Unit, Oxford, Mississippi. She is serving as the 2005 Vice-Chair of the American Chemical Society Agricultural and Food Chemistry Division. 

Dr. Robert Cody received the Ph.D. degree in analytical chemistry from Purdue University, West Lafayette, Indiana, in 1982. He is currently a Product Manager of mass spectrometry products at JEOL USA, Inc., Peabody, Massachusetts, a member of the American Chemical Society (Analytical Division) and the American Society for Mass Spectrometry. 

(2) J.A. Joseph and B. Shukitt-Hale, NATO Science Series, Series I: Life and Behavioral Sciences 

(14) M. Manickam, M. Ramanathan, M.A.F. Jahromi, J.P.N Chansouria, and A.B. Ray. 

Proc. 228th Nat. Meeting Amer. Chemical Soc.

, Philadelphia, Pennsylvania. AGFD 085 (2004). 

Peer-Reviewed Article

There is growing evidence that blueberries have powerful disease-fighting properties. Here the authors explore the research being done in this area, describing the use of a magnetic seector GC-MS SIM method to determine the presence of phenolic antioxidants in blueberries.

LCGC North America-11-01-2005