problem about the split ratio linearity (urgent) Thanks!!

[biomass]

Dear All,

I have to do my methanol and phenol calibration at different split ratio (100:1, 50:1, 20:1, 10:1) . But I found the response factor does not change linearly as the split ratio change. The following are my calibration equations at different split ratio (Y is peak area, and X is concentration)
methanol
split 10 y=1125x
split 20 y=412.19x (For methanol, you can see, as split ratio decrease from 100: 1 to 50:1, the
split 50 y=165.61x response factor increased from 28.275 to 165.61 (around 7 times)
split 100 y=28.275x

phenol
split 10 y=2925.3x
split 20 y=1576.3x
split 50 y=477.87x
split 100 y=105.27x

For the calibration curve at each split ratio, the reproducibility is good, and the R^2 is >0.98.

I don't what's the problem? calibration solvent? or improper column length in the injector port so that too many sample go to the column instead of the split vent?

Thanks!

[Peter Apps]

The problem is that a flash vapourizing split-splitless inlet is intrinsically non-linear because of the complicated gas flow fluctuations that happen just as the sample is injected when the pressure pulses due to the sudden evaporation of a volatile solvent at high temperatures.

The upshot is that you should always run the calibration at the same split ratio (and with the same solvent and inlet temperature) as your samples.

Peter

[mrussell]

And I would say,even injection volume. Essentially,I would not change ANYTHING between injections. There is only one parameter I will change. Some of our systems have a A/D convertor box that digitizes the analog signal for our data system instead of having them directly controlled and connected and recording the digital data. In that case,and that case only,and ONLY when there is an internal standard used,I will occasionally change the range setting on a 6890 if the top of a peak clips off. This works because its a mathematical thing. The range setting divides by two for every one its increased. Since the instrument response really the ratio of the area of the two peaks,dividing the entire signal by two makes no difference. (I guess it WOULD be possible to multiply the entire signal by two at the end in the case of an external standard calibration as well). But the key there is that the change was an electronic/mathematical thing,not someting that changes the physical conditions of the run.

In theory,any phenomena that is reproducable can be charachterized and rolled into the whole curve fit. But the question is,how reproducable is it,and is it going to come back and bite you later when something subtle changes. For instance,that could be particularly sensitive to sample composition and pressure. A small septum leak or a third component in a sample could change things drasticly. The key here is that the split ration is NOT the amount of sample that goes on the column vs the the amount that goes out the split vent. Its the amount of carrier gas at the inlet that does that. Its related,but its perfectly possible for more analyte to go one way than the other.

[WillyOne]

Have you taken into account the flash vaporization volume? Does it fits the free volume of your insert?
Perhaps you are facing to a back flash problem. In the SGS web page there is an application to control (better measure) this problem.
Regards

[ThermoMarkB]

Hi Biomass,

Everyone's answer so far is dead on, the only thing I have to add is that the injectors are proportional, not necessarily linear. The biggest impact is pressures and injection volumes (as well as the liner design). I do simple examples of calibrations by injecting 0.1>1.0uL standards, and they usually work, but they are not as clean as actually making dilute standards.

Vary the split ratio to find the ideal conditions for your sample set and calibration ranges, then leave it. I have rarely found an instance where I needed to change the split ratio for a method, and most SOP's will not be valid when you change something so fundamental to the method.

-Mark