biodiesel analysis - strange peaks

[clairem]

Hi

I'm doing glyceride analysis on biodiesel using the standard analysis (FID, capillary column, inj 350, det 370, temperature ramps up to 370 from 50 deg C) on a Unicam proGC. After a long time to set up - including service engineer visit - it was working well, but now after 3 months I get a series of peaks - regularly spaced approx 1 min apart, about 12 units high ( cf internal standard at 2 units) from 7 mins into the analysis to 25 minutes, then they disappear again. This happens on samples, internal standard only and no injection. I've changed the septum, leak checked and cleaned injector, but dont know what to do next, so any help is much appreciated!

Claire

[Bruce Hamilton]

Can you post a chromatogram? and are you using an autosampler?, and is the Gc carrier gas always flowing, or is the instruemnt turned off with no carrier flow sometimes?.

Do the peaks on the chromatogram start small size, get larger, then drop off, or are they all of similar height. Did cleaning the injector help at all?. Has the problem gradually or suddenly appeared?.

Also, if you don't inject for a while, do the peaks get larger, and then get smaller as you quickly repeat the injections. If you condition the GC at final temperature for a few hours, do you see larger or smaller peaks, and what trend after a couple of injections?. If you perform blanks solvent injections are the peaks larger or smaller?.

If you see the peaks on no injection, then they can only come from:-
1. carrier gas - very unlikely, but need to check you have correct traps in line. Peaks should be largest after instrument has been standing. A few injections should see a decrease in area.
2. septum, try a different batch. Make certain it's rated for the high injector temperature.
3. injector, you've cleaned it, but what type of liner are you using.
4. Column - it's degrading with time ( especially if air has got in whilst it's warm ), or it is contaminated. Conditioning the column at high temerature may help remove contamination, but will aggravate any degradation. Have you go another identical ( new ) column you can try?.

If the peaks appeared larger after solvent injection, then solvent may be the cause of contamination . Make sure it is a good, low residue, grade. Ensure your needle wash solvent is effective and frequently changed. Ensure you are using clean ( new ) sample vials.

Also ensure samples aren't picking up contamination such as diesel, hydrocarbon, or silicone grease. If the problem has suddenly appeared, I'd suspect sample contamination as first cause to investigate.

Good luck

Bruce Hamilton

[clairem]

Hello!

I can't work out how to post a chromatogram, but the peak size stays constant no matter what, solvent or not, after standing or in a series of injections. Usually I use an autosampler , but now I'm taking things apart I inject manually.

So in terms of what to look for
1. Carrier gas, no decrease in peak area on repeated injection so unlikely
2. Septum - ok to 350 deg
3. Injector - has a glass liner
4. Column - has been conditioned with no improvement, and I've ordered a new one.

Is it likely to be a problem with the detector? OR wait and see what happens with the new column.


Claire

[DR]

Hello!

I can't work out how to post a chromatogram, ...

Look Here

[Bruce Hamilton]

If the peaks look like they have passed through a column ( with normal J reverse J peak profile and a decent peak width ), they are unlikely to be from the detector. If they are very sharp spikes or very very broad, they could be, but I very much doubt it.

I'd put a short ( 1-2m max ) thin-film, non-polar column between the injector and detector, reduce head pressure to obtain similar column flow, and perform some runs. If the peaks are still appearing ( will elute at earlier time/temperature, and will be closer together ), I'd suspect your sample solvent and/or injector.

What style is your glass inlet liner?, is it straight through, or have quartz wool, cup, or packing?. After reasonable use, is there much amber/black gunge on the liner.

What are your samples, do they contain derivatising reagents, and are there mono, di, triglycerides and/or sterols also present?. If you are injecting derivatising reagent, you could be reacting with non-volatile material from earlier samples in the injector and column, in which case you're probably doomed . But you would need to solve the problem before subjecting your new column to the samples, and trying a short column should indicate whether the problem is the injector and/or column.

Sorry there are so many questions, but we need to know what the contamination looks like, and where the peaks fit in your normal profile.
Sterols tend to give large single peaks, MAGs and DAGs give couplets a couple of carbon numbers apart, and TAGs give multiple, overlapping, peaks a couple of carbons apart. All will be broader if they come from earlier injections and less obvious if from reactions on the column or injector.

Please keep having fun,

Bruce Hamilton

[Peter Apps]

Hi Claire

Try taking the column temperature ramp out of the run programme. Then run it with no injection. If the peaks are still there, the same size and at the same retention time then they pretty well have to be electronic.

We REALLY need to see a chromatogram.

Peter

[clairem]

ok, here goes with uploading the image!


the white trace was before the problem started - around 2 weeks ago, and is the internal standard alone which elutes at 23 minutes. (slight ghosting form the biodiesel main peak at 11 minutes)

The blue trace is from when the problems started - the first troubleshooting run i did, and again is only the internal standard at 23 minutes.

The yellow trace is a typical no injection run - each consectutive one looks exactly like that one.

Cutting the ends off the column reduced the peak size to about half - does that signify anything?

The liner is glass, straight through, and still looks pretyy clear, theres no gunge on it.

The samples are crude biodiesel so mono, di and tri glycerides, methyl esters and a little bit of the other impurities ( though I'm using refined oil as a feedstcok so there shouldnt be too much). Theres also likely to have been some phospholipids present in some of the samples.

Thanks for all your help, and I'll keep going with your suggestions

Claire

[clairem]

oh, and the samples are silylated with MSTFA prior to anlaysis - is that doom?!?

[Peter Apps]

Hi Claire

Thanks for posting the chromatogram. From the peak sizes going down by a half when you trimmed the column I am pretty sure that you have some contamination at the inlet end (i.e. the beginning) of the column. Usually this builds up slowly, but you had a sudden onset of contamination which points to a drop of unvapourized sample getting into the column.

There are two things that you can do. Put a small plug of glass wool in the inlet liner just above the column tip to help evaporation and intercept drips, and cut an extra chunk off the column to eliminate the contamination that is already there. You could also add a "retention gap" between inlet and column - this is just a 1 m piece of uncoated deactivated silica column that you take out and throw away when it gets dirty and replace with a clean piece. Connect it to the column with a press-fit connector.

If you cannot afford to lose any column length you could try rinsing the column with solvent or reversing it in the oven (i.e. put the detector end into the inlet and leave the other end lose) and baking it out, but these are measures of last resort.

Peter

[GOM]

Claire,

It does look like column contamination at the front end. Without going into the equations, you can cut off half of your column and still retain 70% of the efficiency. This would be a drastic measure but it means don't be afraid of cutting 1-2metres off the front end in this instance, even if it is only a 10-12 metre column. Then in future, as Paul suggests, use a precolumn.

Regards,

Ralph

[GOM]

Sorry, I meant Peter - darn this heat!

[Bruce Hamilton]

Hi Claire,

Thanks for posting the chromatogram - it's very helpful, and you've already got some good suggestions. It's definitely not from the detector.

That's not unvapourised sample, unless you are using butter or coconut oil feed for your methyl esters, that looks to me far more like either ( best guess ) some of your reagent reacting on the column, or column stationary phase breaking down ) or ( less likely ) high boiling alkane hydrocarbon fraction - which you can check for by injecting a very dilute solution of kerosine or diesel and seeing if any of the peaks co-elute. My guess is that your reagent collecting on the column has caused those peaks, or alternatively your column has seen some oxygen while warm.

Still, as Peter says, you need to provide an obstruction so the reactive and low volatility materials, and using a insert with a barrier ( quartz wool, cup, etc. ) will help prevent future incursions, as will the precolumn. If you had along period of good performance, you might like to look at your procedure and see if something has changed ( contamination of a reagent, new bottle of reagent etc ), and also change the insert. The precolumn would be a good idea for any new column.

Yes, you can cut the column to reduce the peaks, but you need to keep sufficient resolution to easity separate the various critical groups. It's easier if your feed is a standard C16,C18, C18:x oil, but if you are using a wider range of fatty acids, you will need to ensure you don't get co-mingling of families.

Are you following a standard method ( such as the DIN procedure - I don't have it handy, but I think it uses MSTFA, I can check if needed ), especially with regard MSTFA derivatisation details etc.?. I'd try to minimise the amount of reagent seeing the injector and column.

For the curious, the reason why MAGs, DAGs, and TAGs are critical in Biodiesel is because they ( especially C16:0 and C18:0 ) can precipitate out of the final biodiesel blends in cold weather, so their content is controlled.

Please keep having fun,

Bruce Hamilton

[clairem]

Hello

I think you may have hit on something with the derivisation - I've recently (say 3 weeks ago) ran a load of sample of pretty much 100% triglyceride (to prove there was no reaction to biodiesel), using the EN method, which is the one taken from DIN standards with the MSTFA addition. Theres no OH bonds for the MSTFA to react with in a ~100% TG sample, so its just gone straight through to the column.

That seems to fit with the deterioation in performance suddenly.

The only other possible source would have been from the phospholipids present in a jatropha oil feed - but I've only run 2 jatropha samples so the effect shouldnt have been so large.

So in future, I'll reduce the amount of MSTFA on high TG samples and try to avoid running them.

Thanks to everyone for your great advice, you've all been so helpful! Have a great weekend

P.S. Bruce - I'm coming to New Zealand this Sept, for a conference and of course, to travel - I'm really looking forward to it!

[JHZR2]

but the peak size stays constant no matter what, solvent or not, after standing or in a series of injections.

Ive found this curious... I get the same thing. If I run jet fuel through my FID straight, I get peaks of a certain height. If I add some drops of C8 to serve as a marker, I get a higher spike where C8 is, but the resultant dilution has no effect on the peak height of everything else.

To me this is strange, as it is being diluted...

JMH