I am having a problem getting replicates of one sa mple to have similar peak areas. The peak areas are going from 4800 to 1500 then to 3000. The same sample is being used and the replicates are performed one after another. I am using an autosampler and have cleaned out the syringe, replaced the glass insert in the injector but nothing seems to help. Anyone got any ideas?

These suggestions may seem a bit obvious but they're intended to help.

1) are you operating near the detection limit for the analyte? Generally, at or near the DL you can get 100% variation in response.

2) Have you tried spiking the sample?

3) Do you have lots of gas pressure? If your supplies are getting low, the pressure will fluctuate affecting the chromatogram.

4) Have you done a complete leak check?

I hope this helps.
bwh

I agree with bwh. 4800 area counts is not a lot on any system. You could be seeing the result of suboptimum integration parameters, especially if the peak is not perfectly Gaussian.

Do all the peaks in your sample go up and down in tandem? By that I mean, when one peak is big, are they all big, and when one is small, are they all small? If so, the most likely source of the problem is in sample introduction, perhaps a leak where the column attaches to the injector, or a worn plunger on the autosampler syringe that allows sample to escape upon injection.

If the size of your peak of interest seems to vary independently of other peaks in the chromatogram, you may be experiencing degradation of the component on the column or in the injector. Make sure your injector insert is properly deactivated (silanized). Packed columns are notorious for component degradation, compared to capillary columns. An injector that is either too hot, or perhaps too cold might also cause this.

I agree with bwh and j.g.clark about the size of your peak. It sounds small.

We have a similar problem with our injector. When running our instrument test program, which includes measuring peak areas of acetonitrile (1.0 % in water) in a packed column, the RSD (n=5) is approx 0.5% at 0.2, 1.0 and 2.5 ul injected volume but outside our limits (approx 3%) at 0.5 ul.

We have change septa and syringe. We have checked column connections and graphite ferrules. The service engineer is at a loss. No leaks are detected.

Could the problem arise from the stepping motor changing the piston position in the syringe?

Barb Wheatley's problem sounds like an injection problem to me. Discard the syringe and replace it. Change the syringe at least as often as the septum. Check the inlet seals. Are you injecting too much for the inlet port to accomodate at the flow/temperature/injection mode employed?

Mikael's problem could also be either injector or inlet. The fact that results are better at higher injection volumes suggests the syringe might be at fault, but he has already changed that.

If another injector is available, it might be worth testing that.

There is another possible source of the observed variation: if a syringe filter is being used, it may be causing variable adsorption. This happened to me with a synthetic nucleotide (filtering about 1.5 mL of a 1 mg/mL solution with a 13-mm diameter 0.45 um NYLON syringe filter (rsd > 20%); when I prepared replicates without filtering, the rsd dropped to less than 2%, and this was also the case when I switched to a PTFE filter.

Many, many years ago we had a similar problem with the NYLON filters. I believe the chemical was the pesticide carbaryl; we did the same thing as Gary, with similar results---but only after checking out everything else involved! Wasted a lot of time on that problem.

Since that time, I have avoided using filters unless I have no other choice. For our samples and sample load, it is much more cost effective to change the LC particle frit when the pressure goes up too much, and we have never had a problem with GC samples. I used to inject ethylene dibromide charcoal tube extracts with no filtration. But that was, perhaps, a unique case.