I am doing sterol analysis using:

Derivatiation with Sylon BFT (BSTFA+TMCS)
Varian 3800 GC
Saturn 2000 MS (ion trap)
1 uL injections, 1000 pg/uL. Would like to get lower.

Using a method developed for HP. Doing both EI and SIS.

Having servere problems with peak area variability. Peaks are sometime up to 2-3 X different using multiple injections of the same sample. Chromatography is otherwise good. Spot-on retention times, good separtaiton. Have tried new inserts.

Any ideas? I haven't had much help from Varian.

What diameter column are you using? What type of injection liner are you using? What is the split ratio? One of the first things to try is to decrease your injection volume as a test to see if your injector is overloaded. What is the injection solvent and injector temp?

You need to isolate the problem to the injector or detector. Both can cause problems such as this.

The problem may also be related to something happening to the sample itself. For example, could you be seeing some degradation of your derivitized sterol product in the vial, or could there be some separation occuring in the vial resulting in a nonhomogeneous mixture? Do the peak areas always increase or decrease, or is there just overall variability in area response (i.e. up and down).

Knowing more about the injection conditions are important as you may be seeing a problem with backflash in the injector. I'm curious to know the diameter of the liner being used, headpressure, injection port temperature and solvent.

I should have been more specific:

30m, .25mm, .25um (length, dia, film)

Splitless injection, 1 uL 2 mm liner.
split 1:50 at time 0 going to splitless at 0.01 for 0.7 min and back to 1:50.

Solvent DCM, Inj temp 280.

This method derivatizes the Sterols but uses a-Cholestane as an IS. Same problem with the undervatized a-Cholestane however.

Variability seems random, sometimes higher, sometimes lower.

Using the constant flow mode at 1 mL/min.

I have decreased injection vol to 0.5 uL without help.

Today I am increasing transfer line temp to 300 and increasing splitless time to 1 min.

Thanks for everyones input.

Changing from split to splitless at 0.01 minutes is a bit unusual. When are you injecting with respect to time 0? If you are injecting before the flow has time to stabilize, or as the system is switching from split to splitless I would expect very poor precision. I always make sure the system has time to stabilize in splitless mode before injection when I do splitless analysis.

I agree with Ron on the valve switching. I'd try some runs without the 0.01 valve switch -- in other words, inject with the purge valve off and leave it off until 0.7 minutes. I have accidentally performed some runs where I injected with split on for a short while then went to splitless and I experienced very strange results (poor peak shape, irreproducibilty, etc.).

The other thing I notice is that the vapor volume of your injected sample is likely to be around 400 to 500 uL (depending on head pressure) with a 1 uL injection. This volume is borderline when using a 2 mm ID liner. I'd recommend using a 4 mm ID liner to avoid potential backflash problems.

I am on Guam so I just completed my weekend!
Today I am doing the runs without the .01 split at the begining of the run. This was how I was advised to do the splitless runs by varian.

Some more data points:

A series of runs with 1000 pg/ul, 1 uL injections gave this (in order of injection):

Injection # Area

1 849482
2 705719
3 544666
4 453866
5 460609
6 352785
7 347929

A continous reduction in peak area with a couple of holds at 4/5 and 6/7.

Thanks

I am not sure I understand your response to J. Ellis' suggestion about preparation stability on July 26. For example, if you inject a "fresh" preparation after having made several injections from an older preparation, do the areas go back to the higher values or do they continue to decrease? Is the variability you mentioned from the same preparation or from preparations made at different times?

If you haven't changed to a different liner and column, I would try that. It is possible that you are losing some of the sample to active sites in the column or liner.

I have also seen the kind of response decrease in the 7 injections listed above when the multiplier is at the end of its useful life.

Hello again,
The samples are dervatized sterols, however the internal standard is a Cholestane. At least the cholestane should be stable. The same variability occurs in both the cholestane and the dervatized sterols.

Also I have done consecutive days with the same dervatized standard and have seen it higher on subsuquent days than on day one.

I have done series that have not shown the progression from high area to low area.

Also, I did a series of hexachlorobenzene injections. Very good reproduction of peak areas. HCB elutes at 16 minutes. My sterols and cholestane elutes at 52-65 minutes.

Lastly, I did injections yesterday with the split off at injection and off for 1 minute. Then turned the split on. Much better reproduction but still not great 25% defferent from hi on to low one.

The liner is new.

The multiplier looks good, at least is sets relatively low.

I am relatively new to GCMS. Perhaps it would be a good idea to change it.

Thanks,

I'd try using a 4 mm ID liner to try to determine if you're experiencing a backflash problem. My calculations show that the vapor volume of your DCM is going to be borderline for the volume of your 2 mm ID liner. If backflash is occuring, then you can condense these high-boilers on cooler areas of the injection port (outside of the liner) and experience area counts that bounce up and down. Another way to test this would be to inject less sample into the 2 mm ID liner -- for example, 0.5 uL injection instead of 1.0 uL.

USE INTERNAL STANDAR AND
MAKE A MAINTENANCE TO YOUR DETECTOR.
WE KNOW THAT ION TRAP IS INESTABLE
I HOPE THAT NO SO MUCH