GC Analysis – Part VI. Applied Example of Analysis with Retention Indices
Alexis St-Gelais – Popularization
I wrote some time ago about retention indices in GC-FID and the advantages of using two columns to analyze an essential oil. Since this is a central aspect of the work we do using GC, I wished to detail a real-case analysis, allowing you to understand more clearly our methodology. Throughout the discussion (as in our reports), the abbreviation RI stands for “retention index”.
We will look at a sample from the roots of Anthriscus sylvestris (L.) Hoffm., commonly known as wild chervil. Let us first observe the chromatographic profiles in Figure 1.
The observation of chromatograms already provides us with some information. With experience, the time frames corresponding to different chemical families of essential oils can be recognized. Here, the oil contains mostly non-oxygenated monoterpenes (which are eluted early on), and a major oxygenated monoterpene. Sesquiterpenes are scarce, and I have excluded them from the figure. From these chromatograms, we then extract data tables, of which Table 1 shows an excerpt.
Table 1. Excerpt of Raw Data Extracted from Chromatograms
RT denotes the retention time of the peaks. The Area column indicates the peak integration, and the % column shows the percentage of the total integration that can be attributed to a given peak. At this stage, neither RIs nor identification are available.
Shortly before analyzing this essential oil, in July 2013, we had also injected our standard mixture of alkanes, which allows us to calculate the RIs of the peaks. Consider the peak eluting at 35.86 minutes on the DB-5 column. The two closest alkanes are dodecane (12 carbons) at 30.05 minutes, and tetradecane (14 carbons) at 42.96 minutes. The formula we use to perform the calculation is at the bottom of this post*. We use a spreadsheet to simplify our task, and we find that the peak at 35.86 minutes has a RI of 1290. This is a logical value since, as I explained in a previous post, we assign a RI of 1200 to the dodecane and of 1400 to the tetradecane, and this peak elutes approximately halfway between them.
We proceed in the same way for all our peaks on both columns. We can also ignore the Area column, since in the case of an essential oil, we use the “% of the total area =% w/w of the total oil” convention. To simplify the example, Table 2 shows the results only for the 5 largest peaks on each capillary column.
Table 2. Retention Time, Retention Index and Percentage of
the Five Most Prominent Peaks on Each Capillary Column
Let us once again focus on peak #29, on DB-5. It is time to take a look at our database, shown in Figure 2, to see what compounds could match its RI of 1290. To this value, we add a small margin (1288-1292), knowing that RIs can be slightly offset from analysis to analysis.
Our database gets richer with each of our analyses, allowing us to add our findings when we have to use a GC-MS. As it can be seen here, several compounds may match the peak on the DB-5 column, which represents 9.4% of the oil. From Table 2, we easily spot a peak representing 9.36% of the oil on the Solgel-Wax column, with a RI of 1627 on this second column. Looking again at our database, we can only conclude that the peak is trans-sabinyl acetate, not one of the four other candidates. This compound has already been reported in wild chervil (1), increasing our confidence in our identification. As percentages for this compound are very similar on both columns, our work for this compound is complete: there is no coelution.
However, we well know that one particular coelution occurs in lots of essential oils: that of limonene and β-phellandrene. These two compounds are observed around a RI of 1024 on DB-5, and the main peak of the oil (peak #15, comprising 48.83% of the oil) corresponds to them. From Table 2, we see that peak #16 on Solgel-Wax represents 46.02% of the oil, with a RI of 1188. It therefore is β-phellandrene, according to our databases. However, the gap between the percentages on the two column suggests that limonene is also present. From our database, we know that limonene is usually observed at a RI of 1178 on the Solgel-Wax column. Indeed, a peak on this column eluting at 11.61 minutes (not shown in the table), has exactly this RI value and comprises 1.45% of the oil. In this case, the use of two columns has helped us determine which of the two compounds was the largest, and has documented the presence of the second.
Keeping up with the identification process, we would gradually complete Table 3 for the major constituents of the oil. This is starting to look like one of our reports.
Table 3. Identified Compounds from Wild Chervil Essential Oil
More efforts would allow us to identify the minor components of the oil, which are often just as important for quality control than the few major compounds. In the end, we would be left with some unidentified small peaks, which is quite usual. That is why, in our reports, we indicate the total percentage of the oil that has been characterized. We usually need a few hours to fully analyze a sample.
In short, the use of RIs with two columns allows us to analyze systematically an oil without resorting to a GC-MS, and nevertheless to reach a high level of confidence in our identifications. If this wild chervil oil is pretty simple, RI couples are generally equally useful for more complex cases, such as sesquiterpenes- or esters-rich oils.
(1) Bos, R., Koulman, A., Woerdenbag, H., Quax, WJ, Pras, N. Volatile components from Anthriscus sylvestris (L.) Hoffm. J. Chromatogr. In 2002, 966, 233-238.
*The formula from reference books is adjusted because we use only alkanes with an even number of carbons: