Manual Injection of Gas sample into GC needed

[daffonce]

So here is the run down of the equipment that is being used

GC: Varian GC3800
Injector: Varian 1177 split/splitless
Split Ratio: 8:1
Column: CP-PoraBOND q fused silica
Detector: FID

We are using this set up to measure ethylene concentration of approximately 10 PPM and below. The first testing and application used a small sampling pump to draw samples from the system to the sample inlet on the side of the GC. The pump was left on at all time and the sample was then returned to the system. The flow rate was approximately 2 LPM. This set up worked beautifully and gave very reproducible results and the accuracy needed.

Going forward it is desired to manually inject samples into the GC. I.E. we wish to draw samples into a syringe at a remote or semi remote location and inject the said samples into the GC. I have tried injecting 500 uL of gas through the septa, results looked promising but reproducibility stinks. I have tried upping the volume to 5 mL again results looked promising but the reproducibility stinks. I have tried using both of these syringes to also inject said samples (500 uL and 5 mL) into the sample inlet port on the side of the GC. These were all ideas presented to me via the guy that trained us, I have very little faith in some of the things he says. So does anyone have any suggestions on a method of manual sample injection that might provide reproducible results? How much volume should be injected, where should the volume be injected, at what rate should it be injected, when should the start of the run be triggered (before, during, or after manual injection begins. Any help would be greatly appreciated.

D

[AICMM]

Daffonce,

My experience with a customer of mine is that when they tried to inject large volumes of headspace for hydrogen analysis into a standard injector the peak shape for the air peak was terrible. When they installed a manual injection valve of 2 mL and did the same injection, peak shape for air was much better and hydrogen was much sharper. So, that is how they now do it. Pull a 5 cc headspace sample, load a 2 mL loop and inject. The valve is downstream of the injector in this case.

Another option is to pull samples onto a trap and then desorb the trap with the flows you are working with. I was recently visiting a lab and they were taking samples in a large syringe and the hooking them up to the sampling system and having the pump pull the syringe into the GC sampling train when the syringe was installed. Worked well for them although they were pulling to a trap. With an 8:1 split ratio you should have enough flow across the trap that you would get a good injection band on column. The valve and the trap are upstream of the injector in this case. If you do not have a trap, it would be relatively simple thing to retro-fit and Varian has an application note on their web site on just such a configuration.

Best regards.

[GasMan]

Daffonce,

This seems a typical case these days when manufacturers think that everything should be done with a capillary column and splitter. I would question why you need to split your sample of 10ppm, unless there is evidence of overloading. Keep the system simple.

OK, you are stuck with the column and the capillary inlet. I would propose that you plumb the sampling valve DOWNSTREAM of the capillary inlet and use a smaller sample loop. You need to flush the sample loop with about 5 times the volume of the sample loop. Fill the loop, wait about 15 seconds for the pressure in the loop to come down to atmospheric, then start the run, assuming that the valve will inject the sample at start run time. This should give you good sensitivity and good reproducibility.

Gasman