Peak Area Declining Over Time

[GiantTree]

I have been using an automatic switching valve to inject 1 mL of gaseous sample from a sample loop into a Shimadzu 2014-GC. I have run at the same conditions for several months and routinely conduct "bypass" checks to ensure my feed composition is not changing. Over the past 5 months I have seen the peak area and height of my feed chemical dropping gradually. To iterate my point, the peak heights of the compound over the past few months, for the exact same injection technique, are below:

Jan - Area = 13.1e6 Height = 2.9e6
Feb - Area = 7.1e6 Height = 2.1e6
March - Area = 6.0e6 Height = 1.8e6
April - Area = 6.3e6 Height = 1.3e6
May - Area = 4.6e6 Height = 1.0e6

I typically keep the GC oven and detectors hot all the time. So far in troubleshooting I have replaced the inlet liner and injection septum, cleaned the auxiliary sample loop, and checked all tubing for leaks. My only remaining suspicion for the declining peak over time is the loss of sensitivity of the FID detector.

Any ideas on what the problem could be?

[AICMM]

You don't say anything about how you maintain or control your sample pressure. That could easily cause your issues so information about that would be helpful.

Best regards,

AICMM

[aidnai]

Any reason for you to rule out an *actual* change in your feed composition?

As AICMM states, sample pressure is important, in fact it is directly proportional to response, all other factors being equal. So the 2x drop from January to March (area) would require a 2x drop in loop pressure -- unlikely unless you have pretty high sample flow rates and constricted vent lines.

Another concern is the changing area/height ratio. Are the RTs shifting as well? Do you need to change your column or at least bake it out?

I have encountered issues with FID response changing -- generally I have found it is related to the hydrogen, either off spec or running at a different pressure or something.

This could be completely irrelevant but, 1mL sample injection sounds huge. I think my sample loop is a few microliters.

So there's a few ideas for you.