Essential oils (EOs) can be defined as a condensate of the volatile organic compounds (VOC) of a plant. As a consequence, they should be completely volatile and gas chromatography (GC) with direct liquid injection (DLI) is the primary technique to assess their properties like composition, quality, origin,… However, in-house GC experiments on commercially available EOs resulted in severe injector liner and column fouling, as well as signals from decomposing remnants in the chromatogram. Although these issues are generally counteracted by retention gaps and increased maintenance, pure gas phase sample techniques like headspace (HS) and thermal desorption (TD) could also be explored. When applying these, main challenges will be to match the sensitivity and resolution of DLI.

Gas phase sampling techniques can be divided in two groups: static and dynamic ones. Static approaches, like static HS, are well established and easy to use. In dynamic sampling, a gas stream transports the volatiles to a collection device (usually a cold trap) where an enrichment and refocussing take place. Subsequently, the collector is heated and the volatiles are released to the GC. Dynamic techniques, like for example TD, promise higher sensitivity compared to static ones at the cost of more tedious optimization. Attention must also be paid to possible instability of the analytes at higher temperatures. However, both approaches ensure that only volatiles are presented to the GC.

Although systems exist where the TD is built into the GC and the column is directly linked to the cold trap [1,2], most implementations of gas phase samplers have been realised as an add-on to the GC using a heated transfer line to connect both. In contrast with other optimization aspects, the influence of the transfer line is hardly documented. A TD in direct desorption (DD) mode [[3], [4], [5], [6], [7]] should provide sufficient sensitivity to be comparable with DLI and allow to examine the influence of the transfer line parameters.

The purpose of a transfer line, in this context, is to transport a gaseous plug of VOC from the sampler to the column without alteration in composition, concentration or distribution of its components. This requirement is hard to reach as slip boundary conditions will deform the shape of the plug and wall/compound interactions can cause compound discrimination and changes in compound distribution [8,9]. The classical approach to achieve the ideal is to reduce the transit time as much as possible. So, a high temperature is applied to avoid condensation and to increase mass transfer between the wall and the mobile phase, but this may lead to degradation of thermally unstable compounds. A transfer line with a small diameter (capillary) allows a local high gas velocity, reducing the transit time, while the gas flow requirements of the mobile phase for the cold trap (of the TD) and the column can be met. Unfortunately, even after optimization, the VOC plug arriving on the column is wider compared to DLI, resulting in loss of resolution between the peaks in the chromatogram. To partially overcome this, a cold trap outlet split assembly and wide bore columns can be used.

For clarity of the discussion further in this paper, the term ‘transfer line’ (TL) refers to the actual capillary, while the heated tubular construction supporting this capillary will be denominated as ‘transfer segment’ (TS). Routing the column instead of the TL through the TS, creates the situation where the column is directly connected to the cold trap of the TD. In fact, this represents two GC systems in series: one with a part of the column in the TS on a constant temperature and another with the majority of the column in the normal programmable GC oven. This raises the question for the optimal TS temperature. Refocussing effects may benefit from a low temperature, but less volatile compounds could be retained. On the other hand, a high temperature might jeopardise the separation of more volatile, early eluting components.

In this work the effect of different transfer regimes has been studied. Besides the conventional setup with the TL at a fixed high temperature, different constant and varying TS temperatures were evaluated. As examples, some EOs were taken, representing complex mixtures of volatile compounds. Finally, a comparison with DLI was made.