Retention Cross-over Phenomenon in Gash Chromatography Can the Mystery be Revealed?

Have you ever faced changes in elution order after modifying the column temperature or the heating rate in the temperature program of the GC analysis of complex samples? This so-called cross-over phenomenon, which can lead to problems in peak identification, has been a well-known mystery in GC for decades.1 But, so far, the physico-chemical background is still not well understood.

The cross-over phenomenon is very common when separating compounds with different functional groups on polar stationary phases. For example, we observed a reversal in the elution order for components like linalool and camphor on a polyethylene glycol column (Carbowax 20M) after changing the column temperature programming rate: at 5°C/min. linalool elutes before camphor but at 3°C/min. camphor will elute first. Effects like this are often observed when essential oils are analyzed or, to be more precise, when the GC methods are optimized. The reversal of the elution order is mainly explained as a result of the different temperature-dependencies of the intermolecular interactions, which are responsible for the retention: London-type dispersion forces and induction forces are independent of temperature, whereas the orientation forces and hydrogen bridge bonds depend strongly on the temperature (Figure 1).

Figure 1 Functional groups influence elution order.
Linalool (bp.: 199°C) Camphor (bp.: 209°C)

However, this explanation is only half the truth and we should examine the influence of column temperature on retention in some more detail. It is generally known that the column temperature is one of the two most important variables in GC (the other being of course the nature of the stationary phase). In partition GC, the effect of temperature on the solute partition coefficient K is given by the van?t Hoff relationship ln K = HS/RTC + C (with HS being the molar heat of solution of solute). From this follows the fundamental correlation between column temperature TC and retention factors:

where k? is the retention or capacity factor (k? = t?R/t M) and ? the column phase ratio. This equation indicates that the retention decreases logarithmically as the column temperature increases.

Therefore, the dependence of the retention time upon column temperature is usually expressed graphically as the log of the retention parameter (net retention time t?R or retention factor k? or retention index I) vs. TC or 1/TC, where TC is the absolute column temperature. In many cases, the plots are linear over the temperature range employed and, furthermore, the lines are approximately parallel to each other indicating that there is little change in selectivity by changing the column temperature in isothermal mode. This is valid for chemically similar compounds. But closer inspection reveals that some lines diverge slightly in their slope and even cross each other (Figure 2).2 The practical implication is coelution of the two compounds at the temperature where the lines intersect. By further changing the column temperature the compounds are again separated but in reverse elution order. As mentioned above, this kind of behavior is often experienced when compounds of different chemical nature are analyzed on moderate to highly polar stationary phases. But not only compounds with different functional groups will behave this way!

Figure 2 Retention indices on squalane (IS) as a function of TC for isothermal GC at 27°C, 49°C, 67°C, and 86°C (Data from J. Gas Chromatogr. 6 (1968) 203-217).

It is known to a lesser extent that changes in peak elution order also occur on nonpolar or weakly polar stationary phases for hydrocarbons that differ only in their carbon skeleton, e.g. aliphatic versus cyclic compounds or cyclic compounds differing in their ring number. The terpenes sabinene, ?-pinene and myrcene are given as an example in Figure 3. The cross-over effect was observed on a polydimethylsiloxane phase with 5% phenyl (60m, 0.25mm ID, 1µm film thickness) as well as on a 100% polydimethylsiloxane phase (60m, 0.32mm ID, 0.5µm film thickness). The column temperature was increased from 90°C to 160°C using isothermal mode. The elution order changed from sabinene, ?-pinene, myrcene at 90°C to myrcene, sabinene, ?-pinene at 160°C.What could be the reason for this effect? A closer look at the molecular structure shows that sabinene and ?-pinene are double ring systems whereas myrcene is an aliphatic hydrocarbon.

Figure 3 Elution order on a 100% PDMS column at various temperatures (isothermal GC).
1. ?-Pinene bp.: 155°C compound 2. Sabinene bp.: 162-166°C 3. ?-Pinene bp.: 163-164°C 4. Myrcene bp.: 167°C

Other interesting analyte pairs prone to cross-over on methylsiloxane phases at different column temperatures are o-xylene/n-nonane, naphthalene/dodecane, as well as 1,2,3- trimethylbenzene/n-decane. In the latter case we also observe coelution and cross-over at different temperature programming rates. At a heating rate of 2°C/min., n-decane elutes before 1,2,3-trimethylbenzene, at 5°C/min. coelution occurs, and at 20°C the aromatic hydrocarbon is the first peak (100% PDMS column, 12m, 0.2mm ID, 0.33µm film thickness, starting temperature 35°C). It seems obvious that the geometry of the molecule, e.g. cyclic versus open chain, contributes to the cross-over phenomenon.

Nevertheless, I have this long-standing friendly discussion with a former student of mine, who persistently points out that the examples we have been looking at so far are always pairs of conjugated versus nonconjugated compounds and that ? interactions, specifically with phenyl modified phases, should be taken into account.

Let’s, therefore, go back to the structure of substances presented in Figure 2: they are exclusively saturated aliphatic and alicyclic hydrocarbons. The data in Figure 2 are from Hively and Hinton (1968) and in that paper the relative retention and retention indices of approximately 250 compounds were measured on a squalane stationary phase at four temperatures.1 From these data one can identify numerous reversals in elution order of aliphatic and cyclic hydrocarbons. The solute interactions with a squalane stationary phase, the most nonpolar stationary phase one can use, are largely a result of dispersion interactions. The authors stated that the magnitude of temperature variation is a function of the size of the molecule expressed by the cross-sectional area of the molecules, which should also prove my point in my next discussion over coffee with my former student.

Finally, coming back to our first example, both components not only show different functional groups, they also differ in their carbon skeleton (Figure 1). Linalool is an aliphatic alcohol and camphor is a bi-cyclic ketone, which means that not only the functional groups but also the difference in molecular geometry will contribute to the cross-over phenomenon.

What can we learn from this discussion? Peak overlapping and cross-over in peak elution order caused by variation of column temperature or temperature programming rate can occur not only on polar stationary phases for compounds with different functional groups but also on nonpolar or weak polar stationary phases for compounds that differ in their carbon skeleton. The analyst should, therefore, carefully examine the structure of the compounds to be separated if the information is available. Furthermore, it is recommended to study analyte retention carefully at various temperatures for difficult separations as an important aspect of method optimization.

REFERENCES

1. Mehran M. et al., HRC, 14 (1991) 745 — 750.
2. Hively, R.A. and R.E. Hinton, J.Gas Chromatogr. 6 (1968) 203 — 217.