When your data goes BOING! An electrifying tale of FID troubleshooting!

Recently my colleagues were performing dual inlet/column/detector analyses of 2,4-dinitrophenol (2,4-DNP) using flame ionization detectors (FID) in an Agilent 6890 gas chromatograph.  They noticed something odd.  The results from the front side were considerably lower and less reproducible than those observed from the back, even though they were essentially set up the same way.  Bah!  That has to be an injection port issue, right?  Perhaps not.

Being excellent detectives, they isolated the cause to the detector by simply switching the columns between detectors back and forth while keeping the inlet sides the same and unperturbed.  In doing so, they observed that the results always stayed with the front detector regardless of which inlet was used.  Curious.  Even more curious was the maintenance that seemed to solve the problem.

Having determined the detector was the likely source of the problem, the first thing they did was replace the FID jet.  That did not improve the results.  Crud.  Next they began to wonder if the spring that makes the connection between the collector body and the FID signal board was the problem.  A seasoned vet examined the spring and felt that it looked like it was extending too far into the collector housing and postulated that an electrical connection may be the problem.  Other seasoned vets felt that this spring should not be considered adjustable.  Uh oh.  They decided to try the spring idea and adjusted the length the spring, pushing it farther in the metal tube (away from the collector) a few millimeters.  See Figure 1 for some pictures that illustrate this.  And the result?

They observed a significant improvement in both the response and the reproducibility!  One thing to keep in mind is that they were observing a calculated value called a response factor, which is essentially a normalized measure of the relative response of an analyte to an internal standard.  It was the response factor that got better.  Weird, huh?  If the signal had increased independent of species, the absolute areas for the analyte and the internal standard might have risen proportionally, which would not have resulted in a difference in the relative measurement.  This didn’t happen, though.  They observed a greater increase in response for the analyte, 2,4-DNP, than they observed for the internal standard, which resulted in a higher response factor.  This leads to interesting thoughts about the relative sensitivity of compounds when there are varying voltages applied to the collector.  Table 1 has some representative data illustrating this observation.

What are the conclusions that I took away from this experience?  The first and biggest lesson is that a problem that seems clearly associated with one part of the instrument may actually be caused by something entirely different, which is why methodical troubleshooting helps isolate the source of trouble.  Second, if the problem really did have to do with the spring it taught me to pay more attention to that part of the detector and it made me want to learn more about the fundamentals of the FID.  Finally, I’m not convinced the spring was actually the thing that made the difference, so like lots of investigations, this experience raised at least as many questions as it answered.  But hey, that’s why we do science, right?  Right?  :)