GC Troubleshooting: Using "No Injection Instrument Blanks"

Let’s cut to the chase.  Here’s the problem I was observing.  I had just performed injection port maintenance, replacing the septum, liner, and liner o-ring.  I thermally conditioned the system before use.  I injected a standard with two compounds, and my earlier eluting compound, endrin, appeared to be suffering from serious co-elutions.  Glommed on to the earlier-eluting side were two significant peaks.

Nuts.

Now, instead of analyzing my sample, I’m troubleshooting.  Why does it seem like I’m always troubleshooting?  Surely it doesn’t have anything to do with me, right?*

So, I enter Troubleshooting Mode (yep, that’s actually a setting of mine…it’s just one ‘click’ away from Storm Out of Lab in Frustration mode, which is one ‘click’ away from Nap mode).  I can be quite strategic when I’m in Troubleshooting Mode, and my favorite strategy is called “Defeat in Detail.”  In this scenario I try to use my understanding of the situation to identify a number of likely causes of my problem, and then I begin to try to isolate and eliminate each cause using one of two possible ranking criteria; either “most probable to least probable,” or “easiest to hardest.”

Considering my problem involved the observation of multiple well-defined peaks that were consistent in retention time and peak shape, I surmised that my problem must be coming from before the column.  I was clearly performing chromatography on the offending compounds, so they must have been introduced into the column.  But where did they come from?

Well, my next question was whether or not they came from inside the instrument or from outside the instrument.  Divide and conquer!  Here’s where the “easiest to hardest” ranking comes into play, and where I finally get to the point of the post.  I decided to perform a “no injection instrument blank.”  This is a very valuable investigative tool, and it’s very easy to do with the Agilent GC Chemstation** software.  Whether you’re starting your analysis from the “Sample Info” screen or the “Sequence Table” screen, if you simply leave the vial position field empty it tells the instrument to acquire data using the specified method but without making an injection.  Now we should be able to determine if my mystery peaks were a product of the instrument or something I was introducing into the instrument.

And what did I observe?  The culprits were still there!  So, it was pretty clear at this point that I had a source of contamination inside my instrument.  The next step I performed was to allow the instrument to sit for awhile with the oven at the initial oven temperature.  When I ran the instrument again, sure enough I observed the same peaks only at a much greater intensity.  This was another clear indication that whatever was causing my problems was a continuous source of the mystery peaks, and I was able to “trap” the compounds in the relatively cool oven.

My experience with modern septa and instruments with septum purge flows is that I don’t often observe these kinds of peaks unless I core the septa and a piece is sitting inside the liner.  I had only punctured the septum a few times, so this seemed unlikely.  The peaks also didn’t seem to follow the evenly spaced, late eluting, multiple-peak-pattern that often indicates septum bleed.

This lead me to suspect my liner o-ring.  I had installed an o-ring from another instrument manufacturer to evaluate it, and once I swapped out the o-ring and briefly thermally conditioned the system, the peak was nowhere to be seen!

Ah ha! Problem solved very deliberately and with very little down time.  What might have happened if I didn’t perform the “no injection instrument blank” first and decided to explore my sample instead?

Why guess when you can divide and conquer, and this is a useful technique when you’re confronted with a contamination problem.

If you have any questions or comments, I’m always happy to hear from you!

* RIGHT?!

** I don’t think this works with Agilent MSD Chemstation or other instrument software platforms, but there are still ways to get it done. Write a comment if you have any questions.