Speeding Up Separations

Chromatography is a waiting game. There’s no way around the fact that separating a mixture takes time, sometimes a lot of time. Separation by conventional high-performance liquid chromatography of even straightforward samples can take more than 15 minutes, and complex mixtures can require hours. Ultra-high-pressure liquid chromatography is allowing researchers to slash those times.

“In a UHPLC mode compared to an HPLC mode, we talk about reducing the time by 90%,” says Andrew Altman, vice president and general manager of LC at Thermo Fisher Scientific. “You can’t make that statement true for every application and every analysis, but that would be the best-case scenario.”

How fast is fast, in UHPLC terms? “Fast can be seconds,” says Helmut Schulenberg-Schell of Agilent Technologies. “Fast can also be 10 minutes or 30 minutes.” It depends on how long an analysis requires in the first place.

Fast separation translates into reduced demand for instruments, operators, laboratory space, and ultimately, money, Altman says. In addition, the short columns and injection volumes that UHPLC requires reduce solvent use. UHPLC makers are making these cost savings possible by decreasing the particle size of packing materials, increasing the pressure, and, sometimes, increasing temperature.

Pharmaceutical companies are the biggest users of UHPLC, Altman says. One of the major drivers for UHPLC in the pharmaceutical industry is the high number of samples, says Xiaoli Wang, a chromatographer who until recently worked for AstraZeneca and now works for Agilent. Wang has shortened some analysis times to less than a minute, making it possible to run more samples for simple separations such as active-ingredient assays.

“If you only care about how much active ingredient is in there,” he explains, “then you probably don’t need much separation power.” For more complicated work, such as impurity analyses, he has reduced analysis times from 30 minutes or an hour to about 10 minutes.

With fast separations, analysts not only can run more samples, but they also have the option of running each sample multiple times, says Alessandro Baldi, business manager for mass spectrometry and chromatography data systems at PerkinElmer. Data redundancy from multiple runs of the same sample yields high-quality analyses, he says, allowing effective and robust method development processes as well as complete statistical evaluation of final data.

For some analyses, companies just want to complete them as quickly as possible. Uwe Neue, an R&D scientist at Waters, cites cleaning validation as an example. This analysis must be done when equipment is switched from one process to another. “I’ve seen separations that are done in 15 seconds,” Neue says. “You’re simply interested in a single peak, and you want to get the analysis done as fast as possible so your equipment is running again.”

To optimize separation speed in UHPLC, Wang advocates a theoretical approach. At the Pittsburgh Conference on Analytical Chemistry & Applied Spectroscopy (Pittcon) earlier this year, he presented his strategy for getting the best separation in the shortest time. In his opinion, every aspect of the separation is fair game for adjustment. Once he decides how much time he wants to spend, he goes through a mathematical calculation to optimize particle size, column length, and chromatographic conditions such as temperature and flow rate. Because particles and columns come in only certain sizes, he has to settle for the commercially available products that most closely match his optimized value. The optimization looks complicated, but he hopes that the stepwise approach is simple enough for others to use to improve their separations, he says.

For many scientists, improved resolution—not speed—is the primary objective. John R. Yates III of Scripps Research Institute in La Jolla, Calif., regularly does complicated peptide separations that take nearly a day. “Going faster would certainly be beneficial,” he says, “but we would need to be able to get the same data quality and amount that we get in 24 hours in half or a quarter of that time.”

“If you really picked people’s brains, I think they would say they want to have the same number of theoretical plates at a higher speed,” says Mary J. Wirth, a chemistry professor at Purdue University. In chromatographers’ lingo, the number of theoretical plates is related to the number of components that can be separated.

In fact, the improved speed that UHPL­C offers has come precisely because the method’s extra pressure and tiny particles provide “excess” resolution, some of which can be converted to speed.

“As you go to UHPLC media, you basically get this bolus of increased resolving power,” Altman says. “I can use all the resolving power and get a much better separation in the same amount of time I used to do my classical, mediocre separation, or I can use all the extra resolving power for speed, getting the same resolution I got in my classical case but now getting it 10 times faster.”

Peter W. Carr, a chemistry professor at the University of Minnesota, disagrees that such extreme improvements are possible from increased pressure. “At constant efficiency, the best one can do is decrease analysis time by a factor of two if pressure is increased twofold,” which is the approximate increase in pressure with UHPLC compared with HPLC, he says.

Whether the improvement is twofold or 10-fold, the increased resolution results from a combination of high pressure and small particles in the column packing material. The standard UHPLC particle is only 1.7 or 1.8 μm in diameter, compared with 3 or 5 μm for conventional HPLC particles. “The smaller the particle size, the smaller the spaces between the particles,” Neue says.

Because UHPLC involves particles with diameters of less than 2 μm, it must be done at high pressures. The first entry in this field, the Acquity from Waters, introduced in 2004, uses a pressure of 15,000 psi, which is more than double the pressure used in conventional instruments, most of which work at 6,000 psi or less. As other companies entered the arena, the pressures have crept up. The most recent entrant—the Nexera from Shimadzu, introduced at Pittcon earlier this year—operates at 19,000 psi.

But pressure can’t go up forever. “We’re going to bump into a point where you just really shouldn’t try to take pressure any higher,” says James W. Jorgenson, a chemistry professor at the University of North Carolina, Chapel Hill. “Maybe that’s out there at 30,000 or 50,000 or 75,000 psi. As we edge the pressure up, making valves that can have 10,000 to 100,000 operations before failure gets harder and harder. People don’t want to give up the expectation that an HPLC instrument is going to be pretty rugged and you won’t have to have it serviced more than once or twice a year.”

One side effect of the ultrahigh pressures in UHPLC is that “your column is literally heating up toward the end,” says Gert Desmet, a chemical engineering professor at Free University, in Brussels. The problem can be solved by using narrow columns that allow the heat to dissipate.

Desmet proposes combating this so-called viscous heating with connected short segments of column to help dissipate the heat rather than a single long column—for example, three 5-cm columns instead of one 15-cm column. “You can go to 30,000 psi or even more and still keep the temperature low,” he says. In addition, Desmet notes, short columns are easier to pack uniformly and thus offer better efficiency than long columns do.

Because of temperature sensitivity, most systems now come with ovens to control column temperature. Tight temperature control is essential with UHPLC, Baldi says, to avoid variations in compound retention times.

“If you want to have stable temperature control, you’ve got to go 5 °C or higher above ambient,” says Simon Robinson, LC product manager at Shimadzu. “That way you can accurately control it to 0.1 °C.”

“When you’re trying to go super-duper fast, higher temperature helps,” Carr says. That’s because as temperature rises, diffusion increases and solvent viscosity decreases.

But users need to watch out. “As soon as you get around 50 °C and up, you have to worry about sample degradation, column chemistry, and physically destroying the bonds in a column,” Robinson says.

The biggest strike against raising temperature to hasten separation is the unpredictable way that it can affect a separation. More than just speeding up a separation, temperature can actually change the order in which compounds elute from the column, Altman says.

As a result, “if you raise the temperature a lot, you’ve got to refiddle your mobile-phase composition and pH to get a good separation again,” Carr says.

People tend to focus mainly on the pump and column in fast LC, but speeding up the separation strains other parts of the system as well. With reduced sample volumes, injection reproducibility becomes difficult, and because the time required for injections tends to remain constant, the ability to achieve faster analyses becomes limited. In addition, the detector—whether it’s a diode array or a mass spectrometer—needs to be able to scan fast enough to collect a minimum of 20 points across increasingly narrow peaks, Altman says.

Desmet warns that there are limits to how fast a separation needs to be. “Reducing analysis time from 30 minutes to five minutes certainly makes sense. That’s a useful gain, no doubt about that,” he says. However, “if you already have a method that only takes five minutes, I don’t see much point in bringing it down from five minutes to two minutes.” At a certain point, other factors, such as sample preparation, dominate the total time, and further reductions in the chromatographic run time won’t make a significant impact.

Packing materials have been central to increasingly faster analyses and improved resolution. And researchers continue to develop new materials that push performance even further.

One such type of material is “superficially porous” particles. They can achieve UHPLC levels of speed and efficiency with only a fraction of the pressure required by their totally porous counterparts. Because the particles have a solid core, analytes cannot penetrate into them deeply. With less particle interior to diffuse through, analytes tend to have sharper chromatographic peaks.

Most commercial superficially porous materials, which are available from several companies, have been optimized for small-molecule separations. “You need to tweak them to get them to work optimally for peptides,” says Barry E. Boyes, R&D director of biosciences at Advanced Materials Technology, in Wilmington, Del. The particles optimized for peptides have larger pores than particles for small molecules. At the American Society for Mass Spectrometry meeting last month, Boyes reported that he could separate peptide mixtures at speeds and efficiencies comparable to those of UHPLC by using 2.7-μm superficially porous particles and pressures achievable with a standard HPLC pump. “You can get the resolution you want at back pressure that’s reasonable and is available to the vast majority of people,” Boyes says.

Jorgenson is trying to push superficially porous particles even further. His superficially porous particles consist of a 1-μm nonporous silica core with a thin layer of porous silica on top. The total particle diameter is 1.2 μm. The particles require high pressure, and the reward should be vastly improved efficiency compared with that achieved with 1.7-μm-diameter totally porous particles. So far, however, the results have been “mediocre,” Jorgenson says. The problem, he suspects, lies in the packing of the columns rather than in the synthesis of the smaller particles.

Boyes is skeptical of trying to make superficially porous particles that are too small. “Past a certain limit, the view’s not worth the climb,” he says, “not only because of the cost of the equipment but because you really don’t gain much.”

Trying to make columns with ever-smaller particles has its complications. In Neue’s opinion, it is easy to make smaller particles, but “the pressure requirement increases by at least an order of magnitude when going from 1.7-μm particles to 0.5-μm particles,” Neue says. “Instead of 15,000 psi, you’re talking 150,000 psi.”

Pressure requirements notwithstanding, Wirth is showing that it is possible to make chromatographic materials with nanoscale particles. The micrometer-scale particles used in commercial packing materials can be thought of as aggregates of tinier spheres, Wirth says. Instead of casting tiny spheres into bigger spheres, her group makes tiny particles as uniform face-centered cubic crystals inside capillaries. The 330-nm spheres are made of silica, as are conventional packing materials, so many of the same chemistries that are used in current commercial packing materials would be available with the nanoscale technology.

Wirth borrowed the technology from the world of photonics, she says. “People discovered that if you start with spheres that are similar in size, they spontaneously pack into face-centered cubic crystals,” she says. Similar materials occur naturally in the form of opals.

“When we started working with these, we realized they do pack wonderfully,” she says. However, if the crystals develop cracks and other defects, they perform like conventional materials, so it’s important to prevent that from happening.

At first, Wirth worried that the nanoscale packing materials would require pressures beyond the reach of current systems, just as Neue suggested. However, “we started from the standpoint that pressures are going to be what they are today,” she says. By shortening nanoscale-particle columns to about a centimeter, she has reduced the need to increase pressure. Capillaries packed with the material can withstand pressures of at least 12,400 psi and have achieved approximately 50,000 theoretical plates per centimeter (Anal. Chem. 2010, 82, 2175), compared with 3,000 plates per centimeter for commercial UHPLC.

A short column, Wirth says, “not only reduces the need to have a higher pressure, but just having a shorter racetrack gets you to the end faster.”

Wirth is most interested in applying these materials in proteomics and biomarker discovery, where separations currently take hours. She has started a company called bioVidria, based in Tucson, to commercialize the material.

“High-speed LC is giving you more compounds quantified or identified per day and per dollar, so it’s a huge productivity tool,” Schulenberg-Schell says. “This is like going from a dial phone to an iPhone. It’s a completely different world.” In that new world, maybe the waiting game won’t take as long.