FAME analysis inlet activation?

[sdegrace]

I am analyzing a mixture of FAMEs using an Agilent 6890, quantitating two components of the mixture against a C23:0 internal standard. The samples are fairly dilute solutions (e.g., 0.6 mg/mL for one component) in iso-octane + 0.05 g/L BHT. For a given fatty acid X, I am injecting a standard made from a known mass of the fatty acid and the C23:0 and calculating an RF as follows:

RF = (A[C23:0] * M[X]) / (A[X] * M[C23:0])

A is the area response of C23:0 or FAME X, and M is the mass.

The RF is then used to quantitate the samples. For system suitability, the RF must fall in the range 0.95 to 1.05 (I have no control over this, this is what was given to me, and the transferring lab is able to achieve this). The FAMEs in question are close in molecular weight and response in the chromatogram to the internal standard. The transferring lab is also using Agilent 6890 instruments.

My sample analyses have been closely matching the assays for known samples, which is great, but I can't get the system suitability to pass, which is something I absolutely have to overcome. I am injecting a 21 FAME standard from Nu-Chek additionally as a check... I know the mass ratios of the two FAMEs of interest to the C23:0 (the C23:0 is itself a component of the standard from Nu-Chek, I didn't add it) and I also have to calculate RFs for the 21-FAME mix. These RFs are badly off, as much as 1.11 (always high). These RFs also have to fall between 0.95 and 1.05. As well, the area percents of the C23:0 and two analytes have to be close to the expected values and I am off by a full percent on two of them.

This is consistent between different vials and dilutions of the the test mix, so I don't think my problem is likely due to the sample prep, unfortunately. I am using Agilent crimp cap vials with teflon-lined crimp caps. I am seeing every peak in the 21-FAME mix and my blanks have no interfering peaks, so I believe that coelutions cannot be responsible for my problems and therefore fiddling with the flow or over program would be a pointless exercise - besides which, I have done extensive work with the flow and oven program specifically to eliminate coelutions and my problem is consistent.

It has been suggested to me that if I can't get the RFs to work for the sys suit then I may have an inlet activation problem. I am using a split/splitless inlet in split mode with a split ratio of 100:1 (this is non-negotiable). I am using an SGE FocusLiner (Flow arrow pointing down), these have restrictions around the glass wool plug and come individually foil wrapped, I was told by the experience party at another lab that these liners were the best choice they had encountered to significantly reducing these kinds of problems. I am replacing the septum and the liner before every run.

Assuming inlet activation is my problem, can anyone suggest any further steps I could take to fix it and get my sys suit to work... unfortunately I have almost no latitude with the method, the inlet temperature of 250 C is a given and my oven ramp and flow rate, which I do have some latitude with, can't be touched because the paramters are already optimal - any change and I get unacceptable coelutions (trust me, I know!!). I don't believe any of these parameters are likely to be pertinent in any case.

I am not highly experienced with the mechanics of GC - I don't have a full grasp on what inlet activation actually is, and beyond replacing the liner with a new liner from a sealed package, which I've already tried and failed, I don't have a lot of other ideas as to what to do. Thanks for any help anyone can give!!

[chromatographer1]

Are you weighing your sample of FAs as free acids or as the esters?

I assume your FAMEs are all saturated. Is this correct? If not, are you derivatizing the FAs?

Have you analyzed your FAs for purity, not only for acid content but for water and solvents like hexane-CH2Cl2-CHCl3 etc.?

The datum that the NuCheck standards are all high in Rf should be a clue.

If your data is consistently wrong, then perhaps your assumptions are wrong somewhere.

best wishes,

Rod

[sdegrace]

The Nu-Chek standard contains all methyl esters of known mass ratio (i.e., Nu-Chek gives the w/w % for each component on their CofA).

This is diluted and injected, and the RF calculation performed using the w/w %'s as the masses for the components... under these circumstances, the other lab can get conforming RF values but we can't.

I am derivatizing my samples as methyl esters using boron trichloride - like I say, I can work out my RFs for my calibration standard (composed of methyl esters), which tend to be in spec but high or high and out of spec, quantitate my samples against that RF, and get a very good match to known values for those samples... but we are told that if we can't get the Nu-Chek standard to work then our values are not reliable, so we have to get the Nu-Chek to work in order for our results to be trusted.

[sdegrace]

Oh yeah, also, my FAMEs are mainly unsaturated, at least all the ones I care about are, except obviously for the ISTD.

[sdegrace]

The 21-FAME mix and all the FAMEs used in the preparation of standards are obtained directly from Nu-Chek and accepted based on their CofA. FWIW, the chromatograms are clean and contain only the expected peaks.

[chromatographer1]

From the information you have given the most likely problem is your derivatizing procedure. You are losing a fraction of your unsaturated FAs. This will give you high Rf factors and can account for your samples giving the correct values when calculated against YOUR Rf values but not NuCheck's Rf values.

You may have the FAs exposed to oxygen during heating and the loss being caused by partial decomposition. You may be losing the FAs in the extraction step. If you use an alkane as an internal standard during the processing of your sample and added to your calibration mix you will discover the truth.

Good luck in your research. I hope you find the cause of the problem.

best wishes,

Rod

[sdegrace]

That's probably part of the problem at least for the RFs of the calibration standards being high and thanks for your advice - but the rub for me is that the RF value itself for the Nu-Chek standard has to fall into that 0.95-1.05 range, without regard to what happens when you attempt to quantitate off it. We're not actually using the 21-FAME standard to quantitate the samples. So we're not actually using that RF value from the 21-FAME for anything. But, there is an arbitrary (?) specification I need to meet based on it. Even though I can quantitate known samples with very good results, my results for unkowns will nevertheless not be trusted unless I can meet this RF spec for the 21-FAME mix.

Since the 21-FAME Nu-Chek standard is never subject to the derivatization and is simply diluted and injected, then something is either wrong with the standard or with the GC. Meaning either I have a bad lot of the Nu-Chek, or the dilution damages the sample (standard is diluted in iso-octane + 0.05 g/L BHT, vialed into crimp cap vials with teflon-lined caps, and stored at -15 C), or it is the inlet activation problem the other lab suggested. The only suggestion I have thus far is to try the SGE FocusLiners in individual foil packs, which didn't help, so now I'm stumped.

[sdegrace]

I'm going to see if I can get my hands on a completely different lot of the Nu-Chek standard to see if it comes out any different, and I'm interested in any suggestions about how we are preparing it, although we are following as closely as possible the directions from the lab that developed the method, but I guess what I'm hoping for is some education about the other lab's suggestion that the problem could be activation of the inlet, what that means, and how it might be addressed (and any other instrument-related factors) so that I can pursue that line in inquiry, or eliminate it in a more informed fashion. Because the other lab believes that that is most likely to problem I am virtually obligated to at least look at it.

[chromatographer1]

Certainly the Nu-Check std could be in error.

But if that standard gives correct Rf values, correct in that the other lab gets the same Rf values for standards that they prepare, then if it were your injection liner causing decomposition of your standards, then why did this injector problem only occur with your standard preparation, and not Nu-Check's?

If your C23 FA is high in area or the unsaturated FAMEs are low in area you will have high Rf values. Since this seems to occur consistently with your samples and your standards, then the only connection is your esterification/extraction of the FAMEs.

Am I misunderstanding something here? Is what I said about the descriptions of the situation correct?

It doesn't seem to be an injector problem unless your injector is decomposing the unsaturated FAMEs. This COULD happen I suppose. And it might be discrimination. But to me, it does not seem LIKELY.

It seems you are doing something right, if you make standards, calculate Rfs, and get 'correct' values for samples of known composition.

Your process of analysis is CONSISTENT. It may not be correct but whatever bias you have in your lab is consistent.

Now, you want your analysis of samples to match that of a known standard prepared externally. Nu-Check has been a reliable source for many years. Of course, anyone can make a mistake.

I hope you discover the source of bias and are able to correct it.

best wishes,

Rod

[Consumer Products Guy]

OK, I admit I didn't read everything carefully. that stated, for your RSD/system suitability are you calculating using the ratio of peak of interest to peak of internal standard (as opposed to the peak areas themselves)? That's what we would do.

[sdegrace]

I'm probably not being very clear about my actual goal in this...

I have two standards, a Nu-Chek FAME mix, which I do not derivatize, and a calibration standard which I have prepared myself from the C23:0 saturated internal standard and two unsaturated FAMEs (all from Nu-Chek), which is subject to the same derivatization procedure as the samples. In both cases I calculate an RF value using the known mass proportions of the analytes to the internal standard and the area responses. The RF for the calibration standard is used to quantitate the samples, the RF for the Nu-Chek mixture is just used as an additional check.

I'm obligated to do the analysis exactly this way, and I'm obligated to throw out the results if the RF values calculated for both standards do not fall in the range of 0.95-1.05. Consistently I am coming in higher. While apparently whatever I'm doing wrong results in a consistent bias allowing me to accurately quantitate samples of known composition, I have to meet that specification or my results are not acceptable.

Basically, my problem is not the "real" problem of correctly analyzing these samples. My problem is the "artificial" problem that I have to do it and precisely replicate the customer's method and what they do for system suitability. If I don't succeed, my results will not withstand their scrutiny. They believe that failure of the Nu-Chek RF standard most often points to a problem with the GC system, especially the inlet - whether they are right or not, I'm under and obligation to at least pursue what they think. In order to do that I am hoping to obtain more information about what could be the basic scientific rationale for that point of view, so that I can at least speak and think intelligently about it.

If I understand the idea of "activity" correctly, it is that some kind of contaminant in my inlet area is degrading components in my sample, the most vulnerable being the unsaturated FAMEs, making the saturated internal standard proportionately greater than the unsaturated analytes in the material which reaches the column and hence making my RF significantly different than 1. What I am hoping to gain is some notion of where such "activity" might come from and how to address it. I want to eventually understand and solve the problem (that being to exactly replicate the customer analysis to their specifications), but more immediately I want to put myself on firmer ground regarding the science of their pet theory.

Thanks for your answers, Rod - I appreciate your time and I hope I'm not being too obtuse!

Stephen

[chromatographer1]

I believe I understand what you are saying.

I believe I am not communicating well how I see the problem and what may be wrong.

The activity problem under discussion refers to oxidized 'junk' from the esterification reaction in the injection liner causing decomposition of the unsaturated fatty acid esters in the standard mix.

This decomposition SHOULD affect the purchased Nu-Check std as well as the in-house prepared standard.

In other words, IF the problem is based in hardware activity, the effect should be seen in both standards, both should be giving you the same Rf values for the identical mixes.

For example:

If I inject a Nu-Check standard containing Linoleic acid ME, Linolenic acid ME, and tricosanoic acid ME, each at 500µg/mL and the calculated Rf of the C18:2 ME is 1.03 and the Rf of the C18:3 ME is 1.04,

and then I prepare a standard in-house which has the same concentrations based on the final weight as a ME, not as a free acid, I should get the same Rf values.

If I get high values then somehow, my weights of the unsaturated FAs were wrong (they contained impurities such as residual solvents), my conversion of the unsaturated FAs to the MEs was incomplete, their extraction was not 100%, or I partially decomposed the acids in my attempt to make ME from them. All of these factors might cause the actual concentration of the unsaturated FAMEs to be low in comparison to the concentration of the saturated C23 FAME. My Rf values would then be slightly high.

Let's assume the Rf values are off by 5%.

I suspect that either your unsaturated FAs are impure (your 10mg may be 9.5mg of actual FA) or your esterification reaction may be decomposing linoleic acid and linolenic acid by 5%, or you are losing the unsaturated FAMEs in your extraction somehow.

Any of the three causes would mean a loss of 5% and an error of the Rf by 5%. Of course, it could be that each has a partial role in the loss of 5%. The FAs may be 98% pure, you may be decomposing them 2% by weight in the esterification process, and you might have a 1% loss in extraction.

My guess would be the esterification IF the more unsaturated FA (Linolenic acid, for example) has a larger error in Rf than a less saturated FA (linoleic acid). Otherwise, I would guess the purity of the unsaturated FAs is incorrect.

The saturated FA is not decomposed by the esterification reaction. Unsaturated FAs can be, especially if they are partially oxidized from storage, OR if air is present in the reaction vessel in which they are converted to the ester.

I hope I have been clear. I only seek to help.

best wishes,

Rod

[sdegrace]

OK that makes sense - but my in-house and Nu-Chek standards are both off (high) and non-conforming by a broadly similar amount. I should try making an in-house standard with the exact same mass ratios as the Nu-Chek standard - that would be really interesting. Bearing in mind that my in-house standard is made from individual > 99% pure standard oils from Nu-Chek... I am not attempting to precisely replicate the ratios in the Nu-Chek mix in my in house standard, but if I did try and I compared the results it might tell me something.

I really appreciate your help, but I'm also looking for any contributions anyone has to make on the other aspect of my questions, which is understanding the concept of inlet activity and how it is addressed, because reducing my ignorance in this area where I have to communicate with the customer regarding their theory is a legitimate aspect of my problem as well.

I really am benefiting from this discussion and I appreciate your help, Rod.

Sort of an aside to this discussion that explains why I am locked in to so much with this method while apparently having free rein with the flow and oven program... the customer actually uses hydrogen as the carrier gas, which in my opinion is smart and I wish we could do that. We were under a lot of pressure to use hydrogen carrier and replicate the method exactly, but for good reasons that I can't go into, that is very impractical for us and we had to insist on using helium. Meaning I had to mess with the method to get it to work in an equivalent fashion. Despite the fact we are going to validate, that's really the only leeway I have, though.

Stephen

[chromatographer1]

The helium should be less reactive than hydrogen for carrier in the presence of unsaturated FAMEs. So I don't see that being an issue.

If your column or injector had contamination that would react with the unsaturated bonds causing a loss of unsaturated FAMEs then your Rf values would be high.

If you had a reaction using hydrogen gas which caused an unsaturated FAME to hydrogenate in one or more double bonds which would then co-elute with C23 FAME then your Rf values would be high. (but you are using helium)

Are all your high Rf values high to the same degree? Do you have any saturated FAMEs with Rf values calculated? Are they high, low, or acceptable?

I have had analytes react with my metal needle of the syringe.

Does changing the injector temperature make any difference in the Rf values?

Does changing the brand of injection liners used make a difference?

(hey, I am struggling here, looking for a cause. I will really enjoy hearing the reason for this variance when you find it, and you will.)

best wishes,

Rod

[Bruce Hamilton]

I"m not sure I've understood all of the discussion but a couple of points.

C23 Int Std sound like a AOCS method, or similar. I agree the response factors sound like some sort of inlet or sample decomposition issue.

However, the frequent replacement of liner should address most issues.
I'd mainly be concerned about activity resulting from your derivatisation, possibly traces of the reagent in the injected sample, so ensure your samples are clean.

I don't think you've said the Carbon Number range for the FAs ( and I can't be bothered checking NuChek site ), but please ensure that you aren't seeing discrimination, especially if all your peaks elute earlier than the C23 FAME.

Activity may be obvious if you inject a dilute pure reference mixture ( eg 30% Unsaturates, 30% PUFA, and 40% saturates ) about 10 times.
Initially with your normal volume for about 3 injections, then 3 injections of larger injection volumes, then back to the initial volume.

If the latter samples are larger, and the PUFA and Unsaturates increase in area, it's likely you have activity, and you need to look for the source. I'd start with residual/excess derivatisation reagent.

NuChek had an excellent reputation, and in 20 years of using their standards I never found a composition or purity error, unlike several other Lipid standard suppliers.

It's easy to prepare very stable FAME mixtures from common oils, and I'd suggest preparing three - a wide CN range, eg coconut oil, typical C14 : C22 unsaturated vegetable oil, and a typical PUFA marine oil ( eg cod liver, menhaden ), and make up mixtures and use those to determine whether you are losing unsaturates. Compared to saturated. the marine PUFA are the best indicator.

I've virtually always used hydrogen, but the few times I've used He the extra head pressure caused different discrimination when I've had to adjust for the same retention.

I'd look at the AOCS manual, as it has someinformation, as doesn't W.W. Christie's lipid analysis WWW site.

Please keep having fun,

Bruce Hamilton