Putting Packed Column Supercritical Fluid Chromatography into Perspective

Introduction

What Is Packed Column SFC?

Packed column supercritical fluid chromatography (SFC or pSFC) is an analysis technique similar to liquid chromatography (LC) that uses supercritical fluids (SFs), instead of liquids, as the mobile phase (MP) (supercritical fluids are defined in the next section). The MP solvates the solutes. The stationary phase (SP) consists of a bed of very small particles packed in a tube capable of withstanding high pressures. Some SPs are the surfaces of uncoated particles. Some are organic films bonded to the surface of the particles. Solutes are separated by differential attraction to the SP.

Compared with LC, packed column SFC is faster, more efficient, has a wider range of selectivity, and detection options, and produces less toxic waste. Not surprisingly, the fields most likely to be affected by packed column SFC in the future are traditional LC application areas. In particular, pharmaceutical and agricultural chemical development will derive significant benefits. Chiral separations will likely be a major application area for packed column SFC. This list is likely to surprise many readers since the application areas most often associated with capillary SFC have involved less polar but perhaps more complex solute mixtures such as homologous series, of surfactants, polymers, and the like.

In reality, the characteristics of interest in 'SFC' have more to do with intermolecular interactions in the MP than the name of the fluid. Many of the characteristics that make SFs interesting to chromatographers (e.g., high diffusivity, low viscosity) are also available from some fluids defined as gases or liquids. Unfortunately, the name SFC is somewhat misleading. SFC differs from LC in that the MP is a dense compressed fluid which will dramatically expand if external pressure is removed.

The most widely used supercritical fluids (like carbon dioxide, nitrous oxide, or CHF3) are inorganic and do not produce a response in some GC detectors, like the FID. This combination of characteristics allows some LC-like separations with more GC-like figures of merit, such as high speed, high resolution, and multiple detection options.

Several modern packed column SFC chromatograms may help convey the features that make the technique desirable. In Figure 1.1, the 11 carbamates of EPA Method 531.1 are separated in 9 minutes and directly detected using both a UV and an NPD (Nitrogen–Phosphorus Detector).

The methanol/carbon dioxide (MeOH/CO2) MP flow rate is 2.5 ml min-1, producing near optimum chromatographic efficiency. This is approximately 3.5 times the optimum flow rate in LC (on this column) and illustrates the superior diffusion rate in supercritical fluids.

The standard method uses gradient elution LC2 followed by two postcolumn reactions to yield fluorescent products. Although the separation takes ca. 40 minutes, the column must then be re-equilibrated. The whole process requires ca. 1 hour between injections. A representative chromatogram is shown in Figure 1.2.

Cumulatively, the SFC separation and detection options produce a throughput approximately six times that of the LC standard method, and avoid the complexity of the postcolumn reactions.

An alternative example of the unusual characteristics of SFC is shown in Figure 1.3. The separation in Figure 1.3 was developed to suggest the feasibility of using SFC for screening pesticides not amenable to GC analysis. A 10 ml water sample containing 6–22.5 p.p.b. of 31 carbamate, sulfonylurea, phenylurea, and triazine pesticides was injected into a precolumn mounted in place of an external loop on a six port valve. The water was blown off with helium, and then the precolumn was switched into the flowing stream. The solutes were eluted by a gradient of 1–16% MeOH in CO2, 90–140 bar, from a 1.6 m long LC column packed with 5 µm particles. At 2 ml min-1 of 20% MeOH, the pressure drop was 150 bar. After the column, the flow was split, diverting a fraction to an ECD and an NPD while most passed through the UV diode array detector. The detection limits for some solutes were a few tens of parts per trillion (1/10).

One trend in LC is toward the use of smaller diameter packed columns, requiring less MP. Major reasons are a desire to reduce solvent cost and minimize toxic waste generation. In some locations, it is already more expensive to dispose of solvents than purchase them. Unfortunately, smaller columns require more stringent instrumental design. In general, it is more difficult to achieve the same high efficiencies, and high sensitivity on a small column as on a large column. Packed column SFC offers an attractive alternative. Inert CO2 replaces most of the liquid solvent. Modifiers typically represent 2–20% of the mobile phase. An SFC method on a 4.6 mm column creates the same or less liquid waste as a 2 or even 1 mm LC column. By retaining the larger column format, SFC allows relaxed constraints on extra column effects, while often providing higher capacity, better detection, and reproducibility.

What is a Supercritical Fluid?

It is important to understand that 'Supercritical' is only a defined state. Supercritical fluids are not a separate state of matter (there are only gases, liquids, and solids). To be 'supercritical', a fluid must be above BOTH its critical temperature, Tc, and critical pressure, Pc. The combination of Tc and Pc is known as the 'critical point'. Above its critical point, a fluid cannot be liquified, no matter how high the pressure is raised. Note that the definition only deals with T >Tc and P >Pc. The definition ignores what happens at conditions BELOW the critical point.

There has been a great deal of confusion about transitions from subcritical to supercritical conditions. Such transitions are NOT phase transitions. They are only transitions from one DEFINED state to another. This ambiguity is dealt with in depth in Chapter 3.

Supercritical fluids lack adequate intermolecular interactions which would otherwise condense them to liquids. This low intermolecular energy gives the fluids certain advantageous characteristics compared with normal liquids familiar as mobile phases in LC.

With SFs (and some similar fluids), the pressure can be increased until the molecules are as close to each other as the molecules in a condensed liquid. This molecular closeness and resulting high collision frequency between molecules makes the fluids reasonable solvents for many solutes. Simultaneously, less intermolecular interaction results in lower viscosity and high diffusivity of solutes in the fluid (molecules do not 'stick' to each other). Both will be discussed in detail in Chapter 3.

Do Packed and Capillary SFC Compete with Each Other?

Many readers may recall the controversy a decade ago over whether capillary or packed columns were the 'best' column type for SFC. Reopening that controversy is counter productive and has some of the characteristics of an old beer commercial on American television. Two retired athletes argue about WHY their (same) beer is the 'best' (Figure 1.4a). Each sees different aspects of the same product as its most important attribute. The arguments over column type in SFC are much the same (Figure 1.4b). Individual users concentrate on different aspects of the same technique. In reality, the two column types are best suited for different kinds of samples and compound classes, producing different figures of merit using different fluids and detectors.

A few major attributes of packed SFC are: independent dynamic pressure and flow controls, common use of binary and tertiary MPs, composition programming preferred over pressure programming, elution of much more polar solutes, trace vs. major–minor component analysis, and UV, electron capture (ECD), nitrogen–phosphorus (NPD), and sometimes FID (when no modifier is present) detection.

Capillary SFC should be characterized as an extension of GC to larger, low volatility, but mostly thermally stable molecules. However, either packed or capillary SFC can perform many of the separations for which the other is nominally superior. For example, capillary SFC can also produce high speed separations of small polar molecules like agricultural chemicals, as shown in Figure 1.5. The capillary method is fast but lacks easy selectivity adjustment and sensitivity and reproducibility are likely to be poorer than with a packed column.

Are Capillaries Inherently Superior?

The last 20 years has seen dramatic improvements in capillary chromatography. There has, subsequently, been a tendency among chromatographers to assume capillary columns are inherently superior to packed columns in all cases. However, there are situations where either capillary or packed columns produce superior results. It is the analysts job to match the best technique and column type to each application.

In GC, the mix of column applications has shifted from > 70% packed to > 70% capillary since 1979. However, this change was NOT primarily due to speed, resolution, or sensitivity considerations. Packed columns remain faster and allow higher sensitivity even today. Capillary columns were used in GC for nearly 20 years without becoming the columns of choice. These columns had many attributes of modern capillaries. However, they were also relatively active, causing tailing of polar solutes. Fused silica capillaries introduced a new level of inertness to GC, producing symmetrical peaks for even quite polar compounds. This inertness was the primary reason for the switch to capillaries. The overall figures of merit of capillary GC are so dramatically superior that the technique can easily compromise on sample capacity and speed to achieve inertness and efficiency.

In LC, open tubular columns were demonstrated over 15 years age. Some have predicted a switch from packed to capillary columns similar to the switch that occurred in GC. However, capillaries have still made almost no inroads into LC applications. LC has fewer or at least different problems than GC. The trade-off between speed–sensitivity and resolution in LC is much less favorable than in GC. LC on packings is slow compared with GC. Capillaries would be even slower unless dc< 5–10 µm. With such small dc, injection volumes < 1 nl are required, with reproducibility < [+ or -] 1–10 pl. There would be 10-15 grams of a solute in a p.p.b. injection. Extra column effects are extremely difficult to eliminate at the tiny dimensions required.

The liquid solvents and additives used in LC decrease the tailing problems associated with packing activity. Capillaries offer little improvement in LC inertness.

Supercritical fluids are intermediate between gases and liquids in terms of passivation and tailing problems. Higher diffusivity allows the use of larger dc or higher speed on traditional LC packings. Modifiers suppress activity on packings and capillaries. Capillaries make it easier to achieve high efficiency. Packed columns inherently win both speed and sensitivity comparisons. While GC is primarily a capillary technique, and LC is a packed column technique, SFC is intermediate. From the author's perspective, packed columns have an edge in SFC since SFs tend to be more like liquids than gases.

Is There a Need for SFC?

SFC has not been an instant success. The technique was first demonstrated more than 30 years ago. In the 1960s, LC was correctly recognized as the more general of the two techniques and most subsequent development effort has been spent on LC instead of SFC. This was undoubtedly the right choice at the time. Today, LC is reaching maturity and its strengths and weaknesses are well understood. To improve on LC, both SFC and electrophoresis are undergoing a renaissance.

Some have questioned the need for an additional separation technique. Many problems can be solved by either GC or LC, yet no one is surprised when one is arbitrarily chosen over the other for a specific application. Yet, it has often been suggested that SFC should be considered ONLY for separation problems that CANNOT possibly be solved by either GC or LC. To limit SFC (or any technique) only to cases where no other technique works dooms it at the start and is not representative of how real laboratories work.

A significant fraction of analytical methods are at the margins of techniques. Such analyses are often 'expensive' in terms of time, uncertainty in the result, or in the level of operator intervention. Alternatives that offer enhanced performance at lower cost should be attractive. There is, in fact, a continual shift in many applications back and forth between GC and LC due to subtle changes in technology favouring one then the other.

Packed column SFC has characteristics which logically make it superior to LC for most molecules that can be solvated by SFs. As will be shown later, SFC should actually be the second technique of choice, after GC and before LC in terms of such performance factors as speed, efficiency, and detection options. Of course, LC is unlikely to be displaced from applications that exist. However, in the future, chromatographers are likely at least to consider SFC before LC, although the two are closely enough related technologically that they might eventually merge. Laboratories will probably eventually have l/5th as many SFCs as LCs.

2 Chromatographic Attributes or Figures of Merit

It is difficult to compare separation techniques in any general way. However, some basis of comparison is needed to give the chromatographer a means of understanding why a new technique is worth considering. In this section common attributes of separation techniques are described. These attributes, such as speed, resolution and sensitivity, are often called figures of merit.

For each figure of merit, the common separation techniques can be ranked. However, care must be exercised in comparing techniques based solely on figures of merit. For any specific separation, one attribute may be absolutely critical. A technique vastly superior to others in all but the critical attribute may be useless in the specific application.

Resolution

Resolution (Rs) is a fundamental measure of separation between two solutes. The universal resolution equation, Rs = constant (N0.5)[(α - 1)/α][k'(1 + k'), indicates three aspects of chromatographic systems that lead to the physical separation of solutes: efficiency (N or plates), selectivity (α = k2/k1), and retention (k' = (tR - t0)/t0, where k' is called a partition ratio, tR is the solute retention time, t0 is the column transit time of an unretained peak).

Efficiency

Column efficiency N = L/H = 5.54(tR/Wh)2, where Wh is the peak width at half height] is based on physical dimensions, like the size of particles (dp) or the diameter of a capillary (dc), the length (L) of the column, plus mobile phase flow rate, and sometimes retention (k'). Generating large N is not, in itself, desirable, since it is extremely expensive in terms of speed and sensitivity. A 10-fold increase in L increases N by 10 times but Rs by only 3.1 times (100.5), while tR increases 10 times, and sensitivity degrades (peaks get broad and more dilute).

Selectivity

Selectivity (α = k'2/k'1) is a measure of relative retention. If retention is very different for two solutes, they are easily separated with low N and modest tR. In GC, α is primarily a function of the SP. If two peaks cannot be resolved by brute force (large N, or large k') then the only option is to change the column.

In LC, both the SP and the composition of the MP are important in determining ITLαITL. Most liquid chromatographers would suggest that LC is more powerful than GC because the MP is not inert and can change α. The universal resolution equation shows that α is more powerful than N in resolving specific pairs of overlapped peaks.

SFC actually provides a wider range of α adjustment than normally available in LC. Besides the identity and composition of the MP and the SP, both temperature and pressure also play a major role in α adjustment in SFC.

Retention

With very little retention, resolution of two solutes requires either large N or large α Resolution increases with k', up to k' ≈ 10. With larger k' no additional resolution is achieved. Increasing k' directly trades speed for resolution. A k' of 10 roughly doubles Rs over k' = 1 but increases tR by 5 times.