GC Inlet Liner Selection, Part IIB: Split Liners Continued

My colleague, Alan Sensue, asked a couple of great questions in regards to my previous blog post on split liners.  To summarize, he was interested in what happens to responses for the various liners when you change split ratios.  For instance, if you go from a 20:1 split to a 40:1 split, do detected peak responses decrease to one half, as would be expected?  If you go from a 20:1 to a 10:1 split, do areas approximately double?  By the same token, he also asked, if I set a particular split, am I accurately achieving that specified split on each liner?

Fortunately, I already had some data on hand to help answer these questions.  When collecting the split liner data, I collected data at both a 20:1 split and a 100:1 split.  Ideally, peak area responses at a 20:1 split should be approximately 5 times that found at a 100:1 split if there is a linear relationship.  To test this, I took the average responses found on each liner at the 20:1 split and divided it by the average responses found at a 100:1 split.  Table 1 shows these results.  You can see that for the most part the liners did behave as expected, with the responses at a 20:1 split being around 5 times that found at 100:1.  The one exception was the cyclo liner, which showed a disproportionate drop in responses at a higher split, signifying that higher flows potentially lead to column loading issues with this liner.

Table 1: 20:1 split area counts for hydrocarbons divided by 100:1 split area counts for hydrocarbons. Expected results should be close to 5.

So now that I answered the question as to whether changing the split results in a linear change to peak response, how can we answer the second question?  When I planned out these experiments, I purposefully chose to inject an amount in split mode that once split, would result in the same amount of analyte on column compared to the splitless experiments I performed in Part I.  This way, assuming close to 100% recovery on the best splitless liners, I could see if the split results matched expected responses.

For instance, I injected 2.5 ng on column for the splitless liner experiments in Part I.  For the split experiments at 20:1, I injected 50 ng, so that once split, would result in approximately 2.5 ng on column.  I can then compare the split results to the splitless results to see if the split is relatively accurate.  I chose to benchmark the split responses to the single taper liner with wool in splitless mode, since this was one of the best liners for splitless analyses.  Table 2 shows the results when I take the average 20:1 split peak areas divided by the average splitless peak areas on a single taper liner with wool.  Since the on-column concentrations are the same, the ideal result is close to 1.

Table 2: Ratio of on-column responses of same amount for a 20:1 split vs splitless injection.

Keep in mind, this is a rough approximation, as we cannot assume 100% recovery for the splitless experiments.  An on-column injection would be needed for a true benchmark.  Nonetheless, I think this approximation still provides useful data.  It’s clear that the cyclo and straight liner do not provide the expected on column loading compared to the other liners.