FAME ECFs: how can I reduce deviation for saturates/monos?

[brewerbill]

I am trying to implement AOCS Ce 1i-07 [Determination of Saturated, cis-Monounsaturated, and cis-Polyunsaturated Fatty Acids in Marine and Other Oils Containing Long Chain Polyunsaturated Fatty Acids (PUFAs) by Capillary GLC] on my Agilent 6890. I've got it all set up and everything looks good -- separation is good, peak shapes are good. I have the GC set up exactly as specified in the method (see below for details). However, I'm having a problem showing that my system is performing adequately.

The second step of the method system performance check is to measure detector response to the various FAMEs measured relative to the 23:0 methyl ester internal standard. To put this in the language of the method, the instrument's "Empirical Response Factor" (ECF) for each analyte must be measured using a specific commercially-prepared standard, and the deviation from the "Theoretical Response Factor" (TCF) calculated. ECFs must not deviate more than 3% from the TCFs for saturated and monounsaturated FAMEs.

The ECFs I measure on unsaturated and monounsaturated FAMEs are between 7 and 9% lower than the TCFs, so my system does not pass. If this criterion is not met, the method says "it is suggested the GC parameters (e.g., detector gases, carrier gas flows, and column choice, etc.) be reoptimized to achieve a difference of 3% or less before continuing with analysis." This is pretty vague! My question is, what factors are likely to affect the ECFs for these analytes? Where should I start "reoptimizing"?

Instrumental details:
30m x 0.25mm x 0.25 um FAMEWAX (Restek #12497)
Split/splitless inlet at 235C
1 uL injected by autosampler, split ratio 1:100
Helium carrier at constant flow of 2.4 mL/min
Oven program: initial 170C, 1 C/min ramp to 225C
Detector temperature 325C

Detector gas flows are not specified in the method; I'm using 30 mL/min hydrogen, 400 mL/min air, and helium as makeup (total of carrier + makeup 25 mL/min).

Thanks!

[dblux_]

I would start with decreasing split ratio to 1:50 just to check if this improves coefficient ratios.

[Peter Apps]

If you are adding the internal standard yourself (i.e. if it is not included in the commercial mix) then a simple explanation is that you added too much.

If the internal standard is the last peak to elute (or at least is later than the peaks that you are determining response factors for) then inlet discrimination is likely to play a role. What kind of inlet liner are you using ?

Peter

[brewerbill]

dblux_: That was actually my first thought, but I saw no substantive difference with the split set to 1:50.

Peter: The internal standard is included in the commercial mix, so I'm afraid that's not the issue. You have a good point on inlet discrimination, however -- the ISTD is one of the last peaks to elute, and the ECFs in general deviate less from the expected value as the run goes on. I am using a general purpose split/splitless liner (Agilent #5183-4711) with a glass wool plug.

[Peter Apps]

Peter: The internal standard is included in the commercial mix, so I'm afraid that's not the issue. You have a good point on inlet discrimination, however -- the ISTD is one of the last peaks to elute, and the ECFs in general deviate less from the expected value as the run goes on. I am using a general purpose split/splitless liner (Agilent #5183-4711) with a glass wool plug.

OK, that's fairly diagnostic, we need to work on inlet discrimination. Often with standard methods I end up recommending changes to operating conditions that cannot be changed. So before we start, what can you change and what not ? I presume thatanything that is not specified can be altered ?

Peter

[brewerbill]

Of course -- the way that the method calculates ECFs had me thinking about it backwards. Once I correct for the TCF I'm getting less peak area per C atom as the run goes on, indicating discrimination against late-eluting peaks.

I have a fair amount of flexibility with this method (within reason, of course). Only the following conditions are specified, so anything not on this list is fair game.

• 30m x 0.25mm x 0.25 um capillary column coated with PEG (Restek FAMEWAX is among the columns suggested by the method)
  Split/splitless inlet at 235C
  Split liner with glass wool (e.g. Restek #21022)
  1 uL injection
  Split ratio 1:100
  Helium carrier at constant flow of 2.4 mL/min (53 cm/s) OR hydrogen at constant flow of 1.2 mL/min (43 cm/s)
  Oven program: initial 170C, 1 C/min ramp to 225C
  Detector temperature 325C

In addition, the part of the method that I quoted in my original post indicates to me that even these parameters are not written in stone ["it is suggested the GC parameters (e.g., detector gases, carrier gas flows, and column choice, etc.) be reoptimized to achieve a difference of 3% or less before continuing with analysis."] One of the three items in this list (carrier gas flows) is explicitly specified, yet "reoptimizing" it is not just allowed but recommended. I read this sentence as giving the analyst quite a bit of leeway -- "reoptimizing" any of these parameters is justified if it brings the ECFs into line.

[chromatographer1]

A slight deviation of the tip of the column in the inlet can affect the amount of discrimination seen.

Slight adjustment to the installation of the column may prove helpful.

Back in the day, a splitter was not used and

TA DA ! no discrimination at all.

But on column injection or direct injection may not be useful due to your instrument or experience.
I have used Restek injection liners to perform direct injections onto a megabore capillary column.

best wishes,

Rod

[Peter Apps]

Since the column is the heart of the GC, if you can change that you can change anything !

But let's start with operating conditions, and then work up to hardware.

Generally, discrimination against high boilers is reduced by higher inlet temperatures.

I presume that you are using the autosampler in the Agilent default fast cold needle mode. Try increasing the pre-injection and post injection dwell times (i.e. the time that the needle is in the inlet before and after injection) in 1 s increments to a max of 3 s. What solvent are you using ?

The carrier gas flow rates that you give are the wrong way around - the flow for hydrogen should be higher than for helium. This leads to another thought - if you are using hydrogen with anything other than constant volume flow the FID sensitivity could change during the run (not at all likely I'll admit, but easy to change).

Try these adjustments for now and see how it goes.

Peter

[jdezeeuw]

I have heard this problem now several times.
It looks like discrimination is happening.

Possible solutions are:

- use hot needle injection;
- reduce in let temperature
- inject a smaller volume and split a smaller amount (still get the same amount on the column)

Make sure that evaporation step is OK, Use deactivated quartz wool, so your temperature in the liner will not change to much when injecting solvent.

jaap de zeeuw, Restek corporation