help with 5890 GC (on column, FID, MSTFA, ASTM D6584, etc)

[jf]

our columns are currently connected with the butt connector. I imagine this is the lowest class of connectors.

So the compression connectors with ferrules are likely the best, and should be periodically checked for tightness?

what would be the first indication of a leaky column?

All butt connectors can fail under the strenuous thermal stress the connection undergoes during the oven temperature programming involved in the biodiesel analysis.

Some ideas:

Certainly any inert connector that uses graphite ferrules can be additionally tightened.

Using quartz connectors with polyimide glue might be an adequate solution.

I have no definitive solution. Many connecting system should work, can work, but certainly all CAN fail at one time or another.

Metal and FSOT columns may require different solutions for a leak-free connection.

Please share with the Forum and with me any special solutions you find work for you.

best wishes,

Rod

[chromatographer1]

Bleed higher than normal is one indication that a silicone phase is suffering from oxidation.

Loss of linearity of polar analytes (x-axis intercept is positive, not zero or less than zero) is another.

best wishes,

Rod

[ece]

Bleed higher than normal is one indication that a silicone phase is suffering from oxidation.

Loss of linearity of polar analytes (x-axis intercept is positive, not zero or less than zero) is another.

best wishes,

Rod

loss of linearity do you mean on the calibration standards or in general?

[chromatographer1]

In the standards you will see a positive x-axis intercept.

In samples you will calculate a value that is lower than the actual level.

Samples and standards will both be affected in the same manner.

best wishes,

Rod

[ece]

agreed, so basically to see if our column is possibly leaking (even though it seems to me the most obvious thing to do would be to check the helium pressure), check the calibration curves?

[chromatographer1]

That seems reasonable to me.

To check for a helium leak in a lab full of helium from running GCs might not be possible, but it would be worth a try if you have a helium leak detector. If the leak is tiny but damaging nevertheless, it will be difficult to confirm in this manner.

Performance changes are a better indicator of column condition.

Of course, in addition, as an early indicator of damage, look for a growing asymmetry (tailing) of the analyte peaks you are measuring.

It is always a good practise to confirm with standards that your measurement is accurate as often as you deem necessary (by method requirements or your own comfort level).

best wishes,

Rod

[ece]

gotcha, we'll do that if we suspect a leak. last time i calibrated, all the lines except for triolein were +slope lines, with a correlation coefficient of .99 or better. triolein had a .98 but we are sure it's our column which was designed for lower temperatures than our method dictates for the final ramp, column bleed.

the following are three chromatograms(phs) that display the column bleed at the end (380 C), since our column is designed for 370 C (our salesman claimed there was no column available with a higher stat. phase thermal limit) i figured this was just column bleed.

this was an initial run of a calibration std



this was a later run of a calibration std



this is a recent run, an actual sample



as you can see, lower bleed on a subsequent std of a calibration run, but also less intensity on the same standard. now that i am running actual samples, i am seeing less bleed. initially, we were seeing high column bleed.

i can come to a couple of conclusions on this:

1) the stationary phase in the column might be spent, although if it were spent then i probably wouldn't see any peaks or perhaps i would see a shift of the peaks' retention times, which isn't happening

2) samples (if they are on spec fuel) typically have lower concentrations of all of the glycerides than calibration standards, so the stationary phase retain fewer glycerides in an actual sample than in a calibration standard.

i am only concerned right now with the breakdown of the stationary phase to the point where the column becomes brittle and breaks. but since i am still seeing peaks where i am supposed to see them, i'm not too worried yet. fortunately, we are planning on investing in a more robust column appropriate for higher temperatures.

thanks a lot for your input as always, folks!

[Bruce Hamilton]

The bleed dropped from 7000 to 5000 and stayed there. That's probably normal, and is OK.

I'm much more concerned about the change in retention time of your peaks, as that's much larger than I ever experienced, suggesting that your instrument settings were settling down at somewhat different points.

You may need to ensure that you are fully removing all air from the system before warming the column. I'd recommend leaving it one 24/7 -with reduced carrier flow if you can.

The smaller size of the peaks is usually due to accumulation of rubbish that elevates the baseline and also increases activity of the system - assuming it's not instrumental problems, such as smaller injection, leaking septa etc,

I'd suggest that you are either injecting too much sample or the samples are fairly yucky. Your pre-column is going to need frequent replacement, and you need to set some parameters for the permitted area variation of the Int Stds.

The peak retention times should not drift significantly after the first 2-3 injections, if they do drift during a long sequence, you need to investigate the reason.

I'd inject the internal std about 10x and watch what happens to peak areas and retention times. If the areas increase, you have activity or rubbish present that needs a few runs to settle down.

If the retention times don't stabilise after several injections, you also have some instrument parameters that need attention.

As I noted earlier, you appear to have the high temperature version of the column, which is rated to 400C. The last two chromatograms suggest the column is coping with the temperature fine. The polyimide outer coating will degrade and make the column brittle over several months, not the stationary phase.

Bruce Hamilton

[ece]

hey bruce

I'm much more concerned about the change in retention time of your peaks, as that's much larger than I ever experienced, suggesting that your instrument settings were settling down at somewhat different points.

thanks. we are cheap so we used the same calibration standards and tested them multiple times for repeatability over a week. from what i've heard, typically they last only 4-8 hours, even though we refridgerate them. so that could be a contributing factor, i'm not sure.

You may need to ensure that you are fully removing all air from the system before warming the column. I'd recommend leaving it one 24/7 -with reduced carrier flow if you can.

check.

The smaller size of the peaks is usually due to accumulation of rubbish that elevates the baseline and also increases activity of the system - assuming it's not instrumental problems, such as smaller injection, leaking septa etc,

we frequently do blank solvent runs to correct this and sometimes i see additional peaks (which i've attributed to the needle not being clean). both if there is junk accumulating on our column it's not eluting, so i will take note.

I'd inject the internal std about 10x and watch what happens to peak areas and retention times. If the areas increase, you have activity or rubbish present that needs a few runs to settle down.

i'll get started on that today. i've noticed with previous chromatograms that the peak areas for both internal standards are actually the same every time, or really close. the shift in retention times i've noticed, but at the time i was really looking for relativity and correspondence more so than baseline shifts, since they all drifted by the same amount. still, it is something i take note of more now especially if it affects intensity as that ultimately affects peak area.

thanks for your comments!

[ece]

excellent advice giacomo

I will purchase mstfa in 1 ml vials from now on.
I might try to put the 23 mL of mstfa we have left into ampules anyway.
We did end up getting good correlation with our standards (>99% on all but triolein, which is obscured due to column bleed at the end of the run. Our next column will be a high-temp metal column, which should resolve this)

the one question i have is with your last statements

The pre-column will need to be changed at least every 100 injections, and perhaps more frequently than that.
You will need to use new ferrules on the pre-column every time you change the pre-column.
The analytical column will last AT MOST for 100 samples, and maybe for far fewer, depending on the quality of the biodiesel.

i can see changing out the guard column every 100 or so injections, using tailing of initial solvent peak as an indicator.

But surely, with a good guard column in place, we would expect to get more than 100 samples out of a gc column? I'm guessing 90 of the samples we inject will be astm-quality (<0.25% free and total glycerin). I find it hard to beleive that the labs with an autosampler that run their columns 24/7 are changing out their columns every week.

We use manual cool on column injection instead of split, where people run these things 24/7 on autosamplers with 25-30 samples at a time. Still, I'd rather clip the column than change the column if we can extend it's life.

For manual injection it seems that there would be more accumulation at the inlet, however someone mentioned accumulation at the end of the guard column also. So when you need to clip it where do you clip it? At the beginning? The end? Both? Either way I'd rather clip it to extend it's life than change the whole thing out every 200 injections if that's possible.

btw bruce we did as you suggested earlier and I believe some of those previous samples may have had some flow control problems on the helium end, which we corrected. but thanks b/c it gives us some other way to trouble shoot in the future!

Thanks!

[chromatographer1]

Bruce wrote:

"The bleed dropped from 7000 to 5000 and stayed there. That's probably normal, and is OK. "

Normal, true, but not completely OK in my opinion.

For a FSOT column as they are using the 5000 counts of bleed is quite typical. However, there are columns being used in the industry that have 1/4 the bleed (some even less) and allow the more accurate measurement of the triglycerides.

I would suggest that a different column that offers a lower bleed be selected for his next purchase.

I have mentioned three vendors (and I do NOT work for any of them) and another poster mentioned a fourth one from Japan which I suspect will offer similar performance although I do not have direct experience to confirm that conclusion.

This column is a good column and will do an adequate job. Yet I do believe there are even better solutions out there.

best wishes,

Rod

[Bruce Hamilton]

With regard to trimming the precolumn/guard you just remove a short length from the front. I used to remove about 20 cm each time, but it really depends on the cleanliness of your samples, and the completeness of the derivatisation - eg free fatty acids and decomposition products will smear further than underivatsied TAGs..

I'd certainly agree with Rod that the baseline bleed could be improved, and alternative suppliers of columns will offer columns with improved bleed. The criteria is always whether you can achieve the required resolution and sensitivity. My feeling was that your column should be adequate, given that it had been used for initial learning and method development. Only you can ascertain whether the column meets your requirements.

If you have the money available, a new column is almost always superior to a used column, and spare consumables ( including columns ) are virtually mandatory for critical applications like IPC and product release.

The huge increase in Biodiesel analysis already has GC vendors falling over themselves to provide dedicated instruments, improved columns, and comparative data about performance, so you should select your next column after reviewing some of the recent data, or talking to other users.

Bruce Hamilton