Take this following chromatogram as an example:
The first thing I checked for was an inlet leak with my electronic leak detector, which came up negative. I also verified that there wasn’t a leak at the MSD source nut. A really bad leak can significantly reduce sensitivity by reducing the mean free path (the average distance an ion can travel before a collision with another particle) to shorter than the length of the analyzer. If your instrument has a vacuum gauge, this can easily be checked. Assuming a 200°C flight path (the temperature of the quadrupole in my method) and a particle mass of 28 amu (assuming a big air leak), the mean free path at 2×10-5 torr is 180 cm. At 6×10-5 torr, the mean free path is 60 cm. At 6×10-4 torr, the mean free path is 60 mm.
If your instrument doesn’t have a vacuum gauge, or it is not working (like mine…which will be covered in future blog), you can spray the MSD nut with 1,1-difluoroethane (an ingredient in 3M Dust Remover) while monitoring m/z 51 in manual tune mode. Generating a detailed tune profile report will also perform an air and water check and evaluate MSD performance. Here is the one I ran following the poor performance you see above, indicating virtually no air leak. For the purpose of comparison, here is one showing a large leak.
If we’ve ruled out leaks, the next place to look is at the actual injection. First I verified that the syringe size in the software matched the syringe installed in the tower, and that the split ratio and flows were correct. Last week, Alan posted about Troubleshooting 2 Common GC Syringe Issues. I verified that the septum wasn’t coring, and that the syringe wasn’t clogged or sticking. I replaced the syringe anyways, and ran some test runs. My first run looked better though not good, but the second showed that my problem was back.
At this point, I stopped my run and watched the autosampler perform an injection. I thought maybe the tower wasn’t pulling back the plunger.
This is a video of what I saw:
You probably noticed all sorts of problems with that injection. I wasn’t slow getting my phone’s video recording up and running; there were no pre injection washes with solvent. You may have also noticed that there were no sample washes, and most important, no needle priming sample pumps. The 7683 autosampler specifications sheet describes two benefits of the pre-injection sample pumps:
- Ensures accurate and reproducible sample volume.
- Removes bubbles.
Bubbles would definitely contribute to poor injection precision, and the poor, variable performance I was seeing, and dry syringes rarely draw the proper amount of liquid on the first attempt. I checked the Automated Liquid Sampler (ALS) settings in Chemstation:
Once I corrected this (3 pre injection washes solvent A, 3 post injection washes solvent B, 1 sample wash, and 2 sample pumps), my problems appeared to be solved.
There is still some PAH tailing, but I’m pretty sure cleaning my source will correct that (the transfer line and source are both already set to 330°C).
Once again, we have an example where the problems are software/configuration based rather than consumables/maintenance related. This does beg the question of how this configuration was saved to begin with. I built the method from scratch using the default.m method, and must have overlooked the injector settings when configuring the injection volume. This serves as a good reminder to double check all the software settings before you start replacing components and trimming columns, especially if your instrumentation is shared with other analysts.