I am currently using a HP 5890 II GC and a FID detector to analyze my gas samples. I injected gas samples automatically through a 100ul sample loop via a solenoid valve. The gas samples were prepared by flowing helium through a cyclohexene saturator hold at 0oC to get a saturated cyclohexene vapor pressure. Theoretically, the concentration of the diluted cyclohexene sample should be around 2%. In order to quantify the diluted cyclohexene, I injected 1000ppm cyclohexene(certified gas) through the same sample loop and get a corresponding GC peak area, then I calculated a response factor of cyclohexene. By using this response factor of cyclohexene from the certified gas injection, I can calculate the concentration of the concentration of my gas sample, which was prepared by helium flowing through cyclohexene at 0oC. But the concentration of cyclohexene is 3.3% based on the response factor, it is a big difference from the 2% theoretical value. Assuming cyclohexene has very linear response for FID detector in a large concentration range, can any one advice me on this question? All opinions are highly appreciated! Thanks!

From what you have told us only a sampling error could be the cause if your calculations are correct.

I assume you are allowing both samples to vent to atmospheric pressure prior to injection and are not pressurizing the sample while not pressurizing the standard gas mix.


Assuming all hardware is correctly operating and the sampling is not in error it appears you have a 3.3% concentration.

Two things that I noticed immediately are that you are using a relatively large sample loop for this concentration range, and that you are determining the concentrations of unknowns far above your calibration level. You are calibrating at 1000 ppm and measuring a concentration you calculate at approximately 20,000 ppm. Even assuming FID linearity, any error in calibration will be very significant working so far above the calibration point.

I agree with the above posting that a sampling error due to diffent pressures is a likely cause. You could try putting a back pressure regulator after the loop to make sure you are always sampling at the same pressure.

Thank you all very much for the quick response. These are definitely very good suggestions. I was considering before there might be a pressure difference between the two injected samples. But I tried the following to minimize the difference. For example, I introduced about 10cc/min flow for both samples to fill the 100ul sample loop, the resulting pressures inside the sample loop should be quite similar. Back to the questions why the measured sample is so far from the theoretical value( from antoine equation), are the sensitivity of FID detector and linear range of response are the main reasons? I quote some characteristics of a FID detector in a book:
MDQ---0.01 nanogram(~50ppb), Linearity---1000000. I am not sure if my understanding is correct: the quotation means in the range of 0.01 nanogram and 0.00001g, the FID response linearly, right? Calculate the mass of cyclohexene injected into the column through the 100ul sample loop, 1000ppm corresponding to 3.7 exp(-7) gram, 2% cyclohexene corresponding to 7.3 exp(-6)g, both of them should fall into the linear range I mentioned above. Which means the size of the sample loop might be fine. I am not sure if I am right at this point. More recommendation and advice are highly appreciated! Thanks a lot! Melanie

Do you have your sample loop at ambient temperature? You SHOULD NOT. C6s at a reasonable concentration may be adsorbing in the sample loop or on the diaphram or rotor body unless they are extremely clean and/or at a reasonable temperature higher than ambient. Higher concentration samples will leave more sample adsorbed and do it more quickly than a more dilute sample.

Do you have a valve which shuts off the sample flow on a path ahead of the valve and have the sample loop vent to atmospheric pressure? If you do not, you SHOULD.

Are your integrator and amplifier range set up correctly?

I still suspect you are not sampling properly. good luck

Melanie
The true vapour pressure of cyclohexene may be even lower than you have calculated. The physical properties database http://esc.syrres.com/interkow/physdemo.htm gives an experimental vapour pressure of 89 mm at 25ºC so at 0ºC it would be well below 1% by volume. As to FID linearity I agree with you that 10 micrograms is towards the upper limit but I would prefer to say 10 micrograms/sec. You may be close to this limit depending on your column (but is it splitless chromatography - you didn't say), but even if you are, your bias is in the wrong direction for non-linearity to be the explanation. You should also allow for the fact that sample and reference std are measured at the same loop temperature. The reference certificate presumably states a temperature of 25ºC or 20ºC, but your sample stream has expanded in its passage through the loop. Unfortunately this is again a bias in the wrong direction for your result. I don't think that sorption of the sample stream in the loop can be a factor because the result would yet again be biased low. Have you tried to detect sorption and carry-over by running sample and zero air consecutively?
Sorry! after all this I don't have an explanation except that the published value might be wrong and you are right.

Melanie
Sorry, too hasty -I can't do simple arithmetic! 89mm vapour pressure is 11.7% by volume at 25C. Did you calculate 2% from this experimental value?