<figcaption> Credit: © JERRY MASON / PHOTO RESEARCHERS, INC</figcaption>

For such a simple technique, there's a lot that can go wrong with high-performance liquid chromatography (HPLC) (For a description of how next-generation HPLCs work, see How It Works). Pressure can go haywire, lines can get clogged, and of course, columns don't last forever. We asked Tom Jupille, president of LC Resources, a company that offers courses in HPLC diagnostics, troubleshooting, and preventative maintenance; Uwe Neue, a principal research chemist in the chemistry operations group at Waters Corp.; and Tom Wheat, principal scientist and manager of the life science application laboratory at Waters, for advice in maintaining and troubleshooting your HPLC. Here's what they said.

"People have paid me exorbitant consulting fees, only to find that two labs are interpreting the same recipe differently," says Jupille.

HPLC relies on differential equilibria between your analyte, the stationary phase...

For the record, here's the right way: Mix 10 mM phosphoric acid and 10 mM potassium dihydrogen phosphate in appropriate ratios to get to pH 2.5, and then mix with an equal volume of acetonitrile. Jupille's recommendation: Develop standard operating procedures that explicitly spell out buffer formulations, expiration dates, and so on.

Aqueous mobile phases can become contaminated with bacteria, mold, and debris. "Those things will turn up as an elevated system pressure and sometimes as an elevated background in the detector," says Wheat. "It may be apparent as the bottle of mobile phase being cloudy, but often as not it's a film on the bottom or the walls and it may not be obvious from a casual inspection." Wheat's recommendation: Either prepare running buffers daily, or tweak your chromatography method to include 10% to 15% organic solvent, to minimize contamination.

"If at all possible, keep crud off your column," says Jupille. Philip Gafken, director of the proteomics core at the Fred Hutchinson Cancer Research Center in Seattle, a facility with some 10 HPLCs, says, "We're very careful about the samples we apply. We quiz our investigators at length: Are there salts that can precipitate? Detergents that can coat and damage the column?"

Neue recommends switching from a simple protein-precipitation step (a common way to separate analytes from plasma) to a solid-phase or even liquid-liquid extraction, if the sample is amenable to the change. For instance, if you're looking for organic drug metabolites, mix your sample with an organic solvent to force the desired compounds into that solvent while leaving the bulk of the impurities behind in the aqueous phase.

A guard or precolumn is a short (0.5-2 cm long), relatively inexpensive column placed in front of the main, analytical column. According to Neue, the investment (typically $10-$50) can significantly extend the lifetime of more expensive analytical columns. "We've done our own experiments where under normal circumstances a column lasts for maybe 1,000 to 2,000 injections, and with a guard column we have gotten a column to last for 10,000 injections without even changing the slightest bit."

Adds Jupille: "Oh, and by the way, change the guard cartridge periodically" - about every 200 to 500 injections or so, or more frequently if your samples are dirty.

"Never shut off your HPLC system with a buffer-containing mobile phase left in the system," says Jupille. "The solvent can evaporate and leave behind crystals of buffer, which will ultimately cause damage inside your column. Bugs will grow in it, too." His recommendation: "Either flush the system with buffer-free solvent at the end of the day, or never shut it down."

The same rules apply when setting aside a column for extended periods, says Neue. "Put the column back into the shipping solvent [for instance, pure acetonitrile], because the shipping solvents have been selected to do the least amount of damage to the column."

"All too many HPLC methods call for ambient temperature," says Jupille, "and that always begs the question, does ambient mean Montreal in January or Puerto Rico in July?" In other words, the running temperature can be whatever you need it to be - 4°C, 25°C, or something else - but it should always be specified.

That's because chromatography depends on having analytes in equilibrium with the stationary and mobile phases of the separation, Jupille explains. "Equilibria are of course temperature dependent, so if you change the temperature, the equilibrium distribution changes, retention times change, and peaks can be misidentified."

Though the exact value depends on the method you're running, system pressure should not be significantly higher or lower than it was the day before, nor should it be erratic. 2,000 plus or minus 10 psi is fine, says Wheat; 1,500 plus or minus 700 psi: "That would be bad."

"Always record the pressure," he adds. Most modern HPLC systems have a feature to record pump pressure with your data, and those numbers can be invaluable during troubleshooting. Suppose, for instance, that during a series of runs you suddenly start getting a lot of negative results. That could happen if the peak of interest "had wandered out of the retention time region that was specified for it," says Wheat. "The first thing you might do is to look at your pump pressure trace."

Don't ignore leaks. They tend to get worse with time, and they aren't mere inconveniences. "They will show up as a shift in retention time," says Wheat, with the peak of interest coming later and later. His recommendation: "I would at least glance at all the places where there's fittings that I can see."

If you do find a leak, cut off the fitting and replace it. "Everybody will try to tighten that nut, including me, and that doesn't usually work." Also, be sure to keep a stock of parts and fittings specific to your system.

When things go wrong, resist the urge to change everything at once, says Jupille. "In troubleshooting, you need to be systematic." So, rather than changing the precolumn, solvent, and detector lamp at the same time, change them one at a time. The alternative, he says, is expensive, may not work, "and can sometimes serve to mask an underlying cause of failure."

You can't troubleshoot your HPLC's problems if you don't know how it's supposed to behave, says Jupille. So, he recommends implementing a daily testing regimen to collect that data. Here, in brief, is what you should do (for more detail, see the FDA's published guidelines at www.fda.gov/cder/guidance/cmc3.pdf).

First, establish a reference standard sample; it can be as simple as the compound used to check the system at installation. "I would make, as part of my routine testing for the day's operation, at least three runs of that standard mixture and make sure that I got the same retention times that I had gotten the day before, and that I had gotten the same peak areas that I got the day before," says Wheat.

If the column will be used for multiple sample volumes, Wheat suggests running three trials at each of the different volumes. At five minutes per injection, that still takes less than an hour per day, he says. "If you lose an entire day's runs because the instrument was not tested before starting the series of runs, the economics of that are pretty straightforward."

Maintain a log of your system's performance and review that log periodically to adjust it as needed. That way, when you finally do need to call tech support - or query online communities such as LC Resources' Chromatography Forum (www.chromforum.com) - you can avoid the hand waving, and tell them exactly what's wrong with your system.


Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!