Sample degradation and method parameter question

[Anthony Doyle]

Hi

We recently received a direct injection residual solvents method (using a split/splitless general purpose glass wool liner), to replace our headspace method as it showed lots of unidentified peaks, thought to be due to sample degradation.

The sample is dissolved in DMSO and tested against a 6-component mixture dissolved in the same solvent. The inlet temperature is set to 140C, while the boiling point of DMSO is 189C. Is this a fundamental flaw? While more, bigger degradation peaks are present at higher inlet temperatures, won't the 'plug' of unvapourised DMSO present in the liner affect reproducibility, sample adsorption, ghost peaks etc?

The problem seems to worsen with increasing number of injections. While initially the sample looked clean, peaks with random retention times and areas showed up as more injections went on.

What can be done to resolve this problem?!

[DR]

If the oven temp<BP for solvent and all analytes.
It's called "cold trapping" - it, along w/ very slow injection rates, lets you get away with making large injections of solvents that expand a lot without flashing the carrier gas plumbing. It can actually increase your resolution too.

That said, Yes, generally the injection port temp should be higher than the solvent BP. You really don't want to coat your liner w/ sample & liquid solvent...

[Peter Apps]

Hi Anthony

I assume that your samples are drugs or polymers or somesuch, hence the need to dissolve them in DMSO.

Having the inlet temperature below the solvent boiling point is unusual but not necessarily a problem. What is you column temperature programme ?, and where do the target volatiles elute ?

The contaminant peaks are probably not related to the low inlet temperature or the DMSO - they are likely to be decomposition of the sample matrix (which I assume is not volatile). As you inject more samples you get more matrix in the inlet, and some of it has been there for longer, so you see a progressive deterioration in the chromatogram.

As long as you are injecting a solution of non-volatiles the only way around this problem is to change the inlet liner before the crud builds up to unacceptable levels. If you can get through one batch of samples between liner changes this is really not such a problem.

Peter

[CE Instruments]

If your sample degrades in a Headspace sampler then surely use of glass wool in the liner is inappropriate as this can add to degradation
Running such a low injector temperature is a problem because the DMSO will not boil but will gradually get flushed from the injector by the carrier gas. Your compounds of interest are likely to trap in hot solvent until most of the DMSO has evaporated increasing the likelyhood of glass wool catalysed degradation of your compounds. Poor sample transfer and discrimination may make the trace look better but is it ? If you are going to carry out this type of injection you should be using a PTV. Start at a low temperature 100C, boil off the DMSO then hit the sample with a quick blast on the PTV (250C) to drive the compounds onto the column. No glass wool. Your compounds will "see" 100C for the time it takes to boil off the DMSO and hopefully just enough heat to drive them into the column.

I much prefer Headspace methods as they are much cleaner; are you sure the extra peaks are from sample degradation in the headspace ? Run it with MS detection to prove it ? Can you not modify the headspace method to minimise the incubation time/temperature does your sampler agitate the vial during incubation ?

Peak shape of these additional peaks ? Could they actually be from previous injections and be heavier compounds ? Direct injection methods can certainly introduce heavier compounds that would not have necessarily been seen by headspace.

[Schmitty]

...to replace our headspace method as it showed lots of unidentified peaks, thought to be due to sample degradation...

What was your headspace incubation temperature?

I recently aquired a Tekmar HT3 with a dynamic trap. I am currently analyzing for 15+ residual solvents in water (yeah, easy I know ).

My temperature is 60°C for 20 minutes or so. Were you perhaps too hot during your sample incubation? Or, do you have to focus on the new direct inject method?

An interesting ill-effect that I am noticing is that the EM voltage on my 5973 creeps up by 100eV after every run.

[Anthony Doyle]

Thanks to everyone

The headspace temp/time was 105C for 40 minutes. we don't have access to a mass spec to determine if the unidentified peaks are due to degradation, unfortunately. A 'trial and error' method might have to be employed.
The ‘degradation’ peaks are unlikely to be due to a previous injection, with direct injection, as the temp. ramp ends with the oven at 230C for 8 minutes.

[Peter Apps]

Hi Anthony

If you had decomposition at 105 C in the headspace method then for sure you have the same problem with injections to an inlet at 140 C.

What I think is happening with the direct injection goes more or less:

You inject a sample, the residual solvents (quite volatile) and the DMSO (less volatile) evaporate and tranfer to the column and appear as peaks.

Meanwhile the non-volatile matrix is thermally decomposing in the inlet to produce other volatile compounds, these also transfer to the column and depending on how fast the matrix decomposes, the MW of the breakdown products and the temperature of the column they either appear as peaks, give you baseline drift, or sit on the column because the temperature is too low to move them.

Next injection the same thing happens again, except that now the heavier contaminants from the first injection have a bit of a head start on your analytes, and appear as ghost peaks. Because the inlet is already contaminated the thermal decomposotion goes a bit quicker, or produces some different compounds, so now you have even more interfering peaks on your chromatogram.

After a few injections you have a continual bleed of breakdown products onto the column. Some of them accumulate on the column while it is cool, and move along it as it programmes, but it does not take long before you have such a wide range of heavy crud spread along the column that it does not bake off even at a fairly high temperature.

Your sample matrix is too sensitive to heat for the conditions in your methods. Some fancy work with a PTV inlet might help, but you are still going to have a build up of non-volatile residues. My choice would be to do headspace at a lower temperature and compensate for the lower concentration of analytes in the headspace by transferring a bigger headspace volume to the column and focussing the peaks by keeping the start temperature of the temperature programme as low as possible.

Peter

[gcguy]

If you are concerned about sample degradation in the injector have you considered using a cool on column injection. In the past I have found that making the switch to on column sorts out a lot of problems, poor repeatability, thermal stability, method sensitivity etc. However I work in an environment that allows me to "make up" methods, hence I am free to change analysis methods provided I can prove their validity.

GCguy