Troubleshooting LC Column Problems—Will It Ever End?

Troubleshooting LC Column Problems-Will It Ever End?

Liquid chromatography (LC) columns have been a central focus of LC troubleshooting activities during the approximately 50 years of modern pressure-driven LC systems. For practicing chromatographers today, the primary concerns are related to the chemistry of the separation, blockage by particulate matter, and contamination by sample components. This has not always been the case. This article briefly reviews some of the physical and chemical challenges of LC columns that historically challenged users of LC, but are largely limited today by advances in column technology. 

Column Packing Techniques

Before about 1972, most workers packed their own columns. The column packing material comprised naturally occurring silica that had been crushed and sieved. The first reversed-phase columns were silica coated with a nonpolar liquid, such as octadipropionitrile, that created a usable, but meta-stable reversed-phase system. The use of liquid phases quickly disappeared when stationary phases bonded to the silica were introduced. Very similar bonded phases are the most popular reversed phases today and provide a robust system that can be readily reproduced. Sizing the silica was accomplished by sieving the crushed silica through screens. A combination of a 325-mesh screen stacked on a 400-mesh screen allowed the collection of a 37–44 µm diameter (d_p) fraction of silica particles, which would pass through the first, but not the second screen. This material was then dry-packed into columns, commonly 1 m long × 2.1–4.6 mm i.d. using a tap-and-fill technique. A frit was placed on the outlet of the column and a funnel was attached to the inlet with a short length of rubber tubing. Packing was poured in the top and the column was bounced on the laboratory floor and tapped on the side with a stick until the column was full. The top was then smoothed off, an inlet frit was added, and the column inlet was secured with an endfitting. A well-packed column might have a plate number, N, of >6000, but in most users' hands half as many plates indicated success. As one might imagine, these columns had high column-to-column variability in performance and stability. Settling of the packing material because of physical rearrangement during use or fracture of the particles was common, but was quickly mitigated by removing the column inlet fitting and adding a bit more packing.

Columns containing smaller particles were required to get higher plate numbers, but particles with particle diameters less than 30 µm interacted too strongly with each other to allow dry-packing (1). A big breakthrough in packing technology came in 1972, when Ron Majors introduced a technique in which particles were suspended in a solvent mixture that had the same density as silica (2). This technique minimized interactions between particles and avoided problems because of particles of different density settling during packing and thus reducing column quality. Now, 10-µm d_p particles could be mixed in a slurry and pumped into a column blank. Although it was later discovered that particle settling during packing could be ignored if the columns were packed fairly quickly (1), the basic technique introduced by Majors is still the basis of packing LC columns today.

Columns packed with 10-µm d_p particles were the standard in the 1970s, initially based on crushed and size-classified natural silica, and later on synthetic, spherical silica. During the 1980s, columns packed with 5-µm d_p spherical particles quickly displaced 10-µm d_p columns and very few users packed their own columns anymore. The stability of the columns improved, although column settling was not uncommon and was still corrected by adding a little packing material to the top of the damaged column. In the 1990s, 3-µm d_p columns were introduced, but because of increased regulatory pressure that discouraged arbitrary changes from one column to another, these columns were accepted more slowly than when 5-µm d_p particles were introduced. In the early 2000s, sub-2-µm d_p particles came on the market, which helped pave the way for ultrahigh-pressure LC (UHPLC).

Although the general principles of column packing are widely known, the details are closely held secrets of the column manufacturers. One discovery that was made during the improvement of column packing techniques is that columns packed at pressures substantially above normal operation pressures had better physical stability. When the packing pressure is high enough, the column actually stretches a tiny bit (3) and then returns to its original size after the pressure is removed. This in effect allows the column to be over-packed, so if the inlet frit is applied quickly before the column returns to its original dimensions, the column can be thought to be pressure-loaded. Such columns are quite stable, but if the inlet frit is removed, packing tends to ooze out and the column can be ruined, so the column endfittings no longer can be removed to try to repair a faulty column.

Another benefit of higher-density column packing techniques is the stable packing bed at the column inlet does not shift or settle, which is a major cause of peak fronting. Thus, peak fronting is much less common with today's well-packed columns.

Column Endfittings

As mentioned above, the packing material is kept in the column by porous frits at each end, which are held in place with endfittings. The fittings used up through the late 1970s were simply reducing unions, typically 1/4-in. i.d. fittings on the column side and 1/16-in. i.d. fittings on the connecting tubing side. The through-hole between the two fitting ports typically was 0.5 mm i.d. The problem created by this fitting design was that one could install a 1/4-in. o.d. frit on top of a 1/4-in. o.d × 4.6 mm i.d. column, but the effective surface area of the inlet frit was only 0.5 mm in diameter, the size of the connecting passage pressed tightly against the frit. One consequence was that the velocity of the solvent through the 0.5-mm passage was extremely high-greater than 50 times that in the column. This resulted in what was sometimes termed "the firehose effect," where high-velocity solvent tended to disrupt the packing at the column inlet if the column was poorly packed. Peak fronting, as mentioned earlier, often was the symptom of this type of column failure. Another problem with the inlet fitting was that any particulate matter reaching the column quickly blocked the 0.5-mm diameter effective frit area. This was not a huge problem, because most column suppliers provided several extra frits with a new column and replacing the frit was a simple exercise. Several groups independently realized that a slight modification at the column inlet fitting could mitigate this problem. The most popular design was to slightly taper the inside of the 4.6-mm end of the fitting to create a very shallow cone 4.6 mm in diameter. This design created an insignificant additional extracolumn volume, but allowed the sample to be spread across the full area of the frit. For example, a 0.5-mm tall cone and 4.6-mm diameter frit would add <3 µL in volume, but would increase the available frit area by 85-fold, so frit blockage was much less likely.

Silica Purity

A third development in column technology helped to diminish the nemesis of early users of LC-peak tailing. Peak tailing, especially for basic analytes, was a way of life in the 1970s and continued to a lesser degree until the mid-1990s. The problem resulted from the use of naturally occurring silica, which was contaminated with metals, especially Al[III] and Fe[III]. These contaminants enhanced the ion-exchange characteristics of lower purity silica (often referred to as type-A silica), such that cation exchange played a significant role in the retention of basic solutes under reversed-phase conditions, especially at 5 < pH < 8. The result was strongly tailing peaks, where a portion of the analyte molecules would be retained by reversed phase and others by ion exchange. The early solution to this problem was to add ≥25 mM triethylamine to the mobile phase. Triethylamine interacted strongly with the cation-exchange sites, competing with basic solutes, so tailing was reduced. By the mid-1990s, "metal free" silica particles (often referred to as type-B silica) were synthesized and reduced or eliminated tailing of basic solutes. Type-B silica is the standard today, so significant tailing is seldom a problem with these columns and the inconvenience of using triethylamine as a mobile-phase additive is a thing of the past.

Conclusions

The three improvements to LC column technology discussed here-improved packing techniques, better column hardware, and higher-purity silica-have served to largely eliminate problems of short column lifetimes and asymmetric peaks that were major problems for chromatographers practicing in the 1970s and 1980s. Additionally, these changes also improved column-to-column reproducibility, so workers today can depend on being able to purchase a new column that is nearly identical in performance as the prior one.

These improvements do not mean that column problems are eliminated today. We still have to deal with chemical contamination by sample components and column blockage by particulate matter that was not removed from the sample. However, without the roadblocks of the past, chromatographers can concentrate on developing the best separation of their sample without constant worry whether the column will survive until tomorrow or if a replacement column will require significant adjustment of the method.

References

(1) U.D. Neue, HPLC Columns: Theory, Technology, and Practice (Wiley-VCH, New York, 1997).

(2) R.E. Majors, Anal. Chem. 44, 1722 (1972).

(3) J. DeStefano, personal communication.

John W. Dolan

John W. Dolan is with LC Resources in McMinnville, Oregon.