Precision of headspace method all over the place

[cmontgomery]

Hi everybody,

Been having difficulties with a method im developing for a few weeks now so i decided to seek advice.

Im running a headspace method on an Agilent headspace G1888 and an agilent 6890 gc.
Looking for toluene in DMSO.

Conditions Headspace
Oven 120
Loop 130
Transfer line 140
Pressurisation time 0.1
Loop fill time 0.5
Loop equilibration time 0.1
Inject time 0.2

Conditions GC
Carrier He
Flow 0.9 Constant flow
Inlet temp 200
Inlet spilt 5/1
Detector temp 300
H2 30ml
He 25ml
Air 400 ml

Oven 40 for 3 min and then ramp to 250 over 6 min

When i run a sequence of 20 standards the area for the last standard is 15% higher than that of the first.
The areas for the first 6 injections have a %RSD of 1.3 %. The last 6 injections have a %RSD of 2.7 % however the %RSD of the 20 injections is about 9 %.

The headspace has passed leak test as has the gc. The flows have been checked, new liner and septum installed.
I ran 20 liquid injections and %RSD was less than 1.3 %

I have no idea what to do next.

All help appreciated.
Thanks

[BHolmes]

When i run a sequence of 20 standards the area for the last standard is 15% higher than that of the first.

Sound like you only have minor integrity issues. From my experience I would say 20% is maximum allowed change throughout a run bracketed with standards. However, if these are all from the same vial, then you may have sample evaporation issues and I would suggest that you only inject once per vial from now on.

The areas for the first 6 injections have a %RSD of 1.3 %. The last 6 injections have a %RSD of 2.7 % however the %RSD of the 20 injections is about 9 %.

This sounds great to me! What are you worried about? Maybe your method validation criteria is too strict, %RSD <10% is acceptable.

Whenever you set out to develop a new method, you need to keep in mind its analytical capabilities AND real world applicability. A method can have %RSD less than 5%, but at what cost? For this topic cost means your time, analysis time, money, chemicals/reagents, and any other resources.

From what you have said about your method, it sounds good enough to move forward with validation. But if you want to gain a better understanding of the repeatability and reproducibility of your method, try a simplied version of a Gage R&R study: inject a standard 5 times on day 1, then once per day for the next 5 days for a total of 10 injections. Look at the variation between injections on day 1 to determine repeatability and overall (all 10 injections) to determine reproducibility. This is just a sample, you can increase the number of injections per day or lengthen the study to 10days. As I said before, this is a very simplified version of a Gage R&R study, a repeatability and reprocibility study first developed by the automobile industry to determine the variability of all parts associated with a process. If you want to learn more about this type of study, check out these links:
http://gagerandr.com/
http://www.qualitymag.com/articles/8352 ... o-gage-r-r
http://www.jmp.com/support/help/R_R_Mea ... tems.shtml

[Bigbear]

I remember an Agilent fellow telling me to not go lower than a 10:1 split. If you have enough signal try upping the split.

[rb6banjo]

I routinely run 2:1 split on my SPME methods. I find that the peak shape is better and I don't have any problems with reproducibility with a split ratio that small.

If you're looking for toluene in DMSO, is there a reason you can't just inject the sample directly?

[chromatographer1]

How long are you heating the vials?

Rod

[thohry]

What is the internal diameter of the capillary column ?

[Peter Apps]

So you have a trend of increasing peak area with vial number. How is the headspacer connected to the GC - through the inlet septum or by cutting into the carrier gas supply line ?

Why are you punishing your column by ramping so fast to such a high final temperature ?

Peter