non linear calibration with brand new 7890A agilent GC-ECD

[dalarie]

Hi,

I am currently trying to set up a new 7890A agilent GC with dual u-ECDs.
I have a 7 level calibration with levels 1.55, 3.125, 6.25, 12.5, 25, 50 and 100 ng/ml.
The calibration curve falls off over about 20ng/ml. I have run a split method and a splitless method and get the same results. I use average response factors and the majority are reading about 15%.

does anyone have any suggestions? It seems that the ECDs are very sensitive and I am perhaps looking for ways to decrease sensitivity?

[MMJ88]

Seems like a somewhat common thing with ECD. If you search the literature, you'll see quite a few papers on it. What is your analyte?

It's interesting if splitless vs. a high split doesn't affect it. Can you elaborate more on that point? If you're getting the calibration falling off at the same point with quite different amounts of analyte on column, that would indicate something other than the detector.

[dalarie]

thanks for the reply,

I am analysing about 30 organochlorinated pesticides including DDT and the calibration curves look about the same for each analyte.
The split I was using was 5:1, I ran the standards on splitless and split the same day and the average response factors were about the same. I inject 1ul- could injecting a lower volume help? I also use 0.32um columns. I have a 6890 gc and the calibration is perfect although they are different columns and carrier gas is helium (new one is hydrogen)

You said it could be something other than the detector? Do you have any idea what it could be?

[MMJ88]

I mentioned it could be something other than the detector because you specified you had done split and splitless and gotten similar results (just speculation on my part). However, a 5:1 split isn't very much. Try a much greater split, like 50:1. How good are the peaks in your lowest standard? Maybe you can find a split ratio that will keep your lowest peaks in a comfortable range, while getting your higher ones down out of the non-linear range.

Agilent claims a range of linearity of 10^4, but most other places and people suggest 10^3. In any case, your concentrations only vary by 10^2, so you should be able to achieve linearity.

Can you explain what you mean on the 6890? You run a similar analysis on a different instrument (but with different column and carrier gas) and have no problems with calibration linearity? Using the same exact solutions?

[Peter Apps]

I have at the back of my mind a faint recollection that ECDs do not like hydrogen

You might be going off the top end of the approximately linear range of the detector - try diluting the samples x 100, or increase the split substantially.

Peter

[Bigbear]

Please do not forget that an ECD detector has a very short dynamic range. They work "backwords" if you will in the fact that detected compounds reduce the detector signal vs adding signal in most detectors. Sounds to me as if you are overloading the detector. A quick check would be to up your split rate to at least 50:1 and see if the curves improve.

You could also try increasing the make up flow.

[dalarie]

thanks for the great suggestions.

I have tried a split of 10:1 and I cannot see my lowest standard so I guess increasing the split will not help.
the other instrument I have is agilent 6890. I run the same compounds on that one but I use two totally different columns and a longer temp program (12 min for the 7890 and 30 min for the 6890). I use the same standards to calibrate and consistently get average response factors of less than 10 % for every compound. The make up flows to the detector are the same in each instrument at 30ml/min.

I can try to increase the make up gas, do you know what a good flow would be to try?

[Bigbear]

Some thing's fishy. What do you mean by totaly different columns. Do you have an extra column for the 6890 method? If so try installing it in the 7890 with the same parameters you use with the 6890. The results should be very similar.