The separation potential of chiral GC phases
In the following, we wish to give some information for the selection of chiral GC phases. For this purpose, we have studied the separation properties of representative test substances from different chemical groups on chiral MN columns. The selectivity of the respective phase towards spiro compounds, terpenes, aliphatic, cyclic and aromatic ketones, cyclic ethers, oxiranes, lactones, aromatic esters, aromatic amides, and cyclic, non-cyclic, and aromatic alcohols was investigated. All separations were run under the same chromatographic conditions:
Column dimension: 25 m length, 0.25 mm ID, 0.25 mm film
Carrier gas: Hydrogen, 0.6 bar, split 50 mL/min
Temperature program: 60 °C (2 °C/min) 200 °C
The degree of separation obtained was evaluated using a points-based system (0–10 points), which considers the resolution and the level of difficulty. If no separation of the enantiomer pair was achieved, this was valued at 0 points. The quality of a partial separation was rated 1–5 points, a baseline separation gave 6–10 points.
In the first step, the influence of the molecular weight of the test substances on the separation properties of HYDRODEX and LIPODEX phases with different cyclodextrin ring sizes was measured.
Fig. 1: Separation behavior on HYDRODEX β-TBDAc and γ-TBDAc as a function of the molecular weight of the analytes
Figure 1 shows the separation quality (0–10 points) of test substances with increasing molecular weight on HYDRODEX β-TBDAc (β-cyclodextrin phase) compared to γ-TBDAc (γ-cyclodextrin phase) with the larger cyclodextrin ring. For most enantiomer pairs the larger the analyte is, the better it is separated on the larger cyclodextrin.
For LIPODEX phases with different cyclodextrin sizes a similar tendency of the separation properties of LIPODEX A (α-cyclodextrin), LIPODEX C (β-cyclodextrin) and LIPODEX E (γ-cyclodextrin) can be observed, however, the relatively large substituents (larger than for HYDRODEX phases) strongly influence selectivity, since the cyclodextrin appears smaller. Thus LIPODEX E shows a good selectivity over the entire molecular weight range (see figure 2).
Fig. 2: Separation behavior on LIPODEX A, LIPODEX C, and LIPODEX E as a function of the molecular weight of the analytes
In addition to the relation between the molecular weight of the analyte and the size of the cyclodextrin ring, the separation properties of the different substituents of the cyclodextrins relative to the functional groups of the analytes have been studied.
Here, too, LIPODEX E shows a broad range of selectivity, especially for esters (e.g., mandelic acid methyl ester, mecoprop methyl), cyclic ketones (e.g., α-phenyl-γ-butyrolactone, α-ionone), oxiranes (e.g., phenyloxirane, trans-stilbene oxide) and terpenes (e.g., α-pinene, limonene) (see figure 3).
Fig. 3: Separation behavior on LIPODEX E in relation to the functional groups of the analytes
LIPODEX A shows good selectivity for small alcohols, LIPODEX B does not show any accumulation in the studied compound groups under the test conditions and LIPODEX C is well suited for cyclic compounds with polar groups such as cyclic ketones and lactones. A broad range of selectivity, however, without accumulation within a group, is seen for LIPODEX D. This means, that, even if one substance of a group is successfully separated, separation of a structurally similar compound cannot be guaranteed. LIPODEX G is successfully applied for the separation of underivatized alcohols, esters, and ketones.
Investigation of the HYDRODEX phases shows good separation results for any kind of alcohols (e.g., 2-methylpentanol, linalool, 1-phenylethanol), ketones (e.g., α-phenyl-γ-butyrolactone), esters (e.g., mandelic acid methyl ester), and many analytes with polar ring systems (e.g., linalool oxide) on HYDRODEX β-TBDAc (see figure 4).
Fig. 4: Separation behavior on HYDRODEX β-TBDAc in relation to the functional groups of the analytes
The HYDRODEX γ-TBDAc with equal modification (2,3-di-O-acetyl-6-O-t-butyldimethlysilyl), but larger cyclodextrin, shows similar good selectivities for esters, ketones and amides, but modest separation results for low molecular alcohols. Alcohols with a ring system close to the chiral C atom can be well separated after esterification. The earlier phases HYDRODEX β-PM, β-3P and β-6TBDM show selected separation success.
Even if the later developed phases (e.g., LIPODEX E, HYDRODEX β-TBDAc) show broader applicability, successful separation cannot be guaranteed for every substance of the studied group. For this reason, new phases, like HYDRODEX γ-DiMOM are very important, since they offer the possibility to enantioselectively separate substances which in the past were difficult or impossible to analyze (e.g., pulegone – constituent of peppermint oil).
Many more applications of LIPODEX and HYDRODEX phases are available in the SORBTECH application data base.
Optimization of chiral GC separations
After the selection of a chiral GC phase with promising separation properties, and first separations or partial separations on the respective column, the method has to be optimized.
The carrier gas hydrogen allows relatively high plate numbers for low flow rates of 40–80 cm/s. Thus hydrogen is to be preferred to helium or nitrogen.
Chiral GC columns rapidly tend to show peak deformations due to overload. The injected sample amount can be minimized by the choice of a high detector sensitivity combined with a large split flow. Chiral separations are very sensitive to temperatures, and the optimum temperature is in most cases a compromise between analysis time and resolution.
Fig. 5: Optimization of the temperature
Figure 5 shows the separation of the test compound 2-phenylhexanone at 180 °C in isothermal run and with different temperature programs. Isothermal conditions should be preferred. If a temperature program has to be used, the heating rate should be as low as possible (2–4 °C/min).
Conclusion
Chiral GC phases based on modified cyclodextrins allow numerous enantiomer separations, but the prediction of the selectivity of a chiral phase is difficult.
Studies with LIPODEX and HYDRODEX columns show, that the molecular size and comparison of separations with structurally similar substances enable an estimation – although without guarantee of success. For this reason, a literature search may be helpful. The Sorbtech application database and Technical Support of Sorbent Technologies offer valuable assistance.
References
[1] Gil-Av, B., Freibush, R., Charles-Sigler, R., Tetrahedr. Lett. (1966), 1009 [2] Frank, H. , Nicholson, G.J., Bayer, E. , J. Chromatogr. Sci. 15 (1977), 174 [3] Schurig, V., Schmalzing, D., Schleimer, M., Angew. Chem. Int. Ed. 30 (1991), 987 [4] Fischer, P., Aichholz, R., Bölz, U., Juza, M., Krimmer, S., Angew. Chem. Int. Ed. 29 (1990), 427 [5] König, W.A., Krebber, R., Mischnik, P., J. High Resolut. Chromatogr. Commun. 12 (1989), 732 [6] König, W.A., Enantioselective gas chromatography with modified cyclodextrins, Hüthig, Heidelberg, Germany (1992) [7] Schurig, V., Nowotny, H.-P., J. Chromatogr. 441 (1988), 155 [8] Zhou, X.C., Yan, H., Chen, Y.Y., Wu, C.Y, Lu, X.R., J. Chromatogr. A 753 (1996), 269 [9] Pfeiffer, J., Schurig, V., J. Chromatogr. A 840 (1999), 14