Reproducibility problems - alcohols and headspace

[Anthony Doyle]

I have observed reproducibility problems while analysing alcohols (methanol, iso-propanol, butanol) by headspace, using DMSO as diluent. On 2 occasions, using 2 different methods (and 2 different solvent mixes that contain alcohols) the reproducibility has been a lot worse for the alcohols than for the other components. (Typically >5 for alcohols, using 6 replicate injections). The reproducibility problem is only observed for headspace analysis. I'm using an Agilent 6890 (who isn't?!) and a Gerstel MPS-2 Autosampler, where the headspace syringe is seated in a heating block. What might these symptoms indicate and how might the problems be resolved?

Also, while using similar method parameters, I changed the diluent from DMSO to Benzyl alcohol. The response for the alcohol nearly doubled. Is this phenomenon often observed?

Thanks for your help

[Peter Apps]

Hi Anthony

If only alcohols are affected the most likely culprit is adsorptive activity, probably to silanols on the glass surfaces of the syringe, the inlet liner or the column. The benzyl alcohol temporarily binds to the active sites, leaving more of the analyte alcohols to get to the detector. If the adsorption is in the inlet or the column the peaks will have tails, if it is in the syringe they might still be sharp.

Increasing the syringe temperature might help, or you could deactivate the inside of the syringe with dichlorodimethylsilazane (DCDMS) or something similar.

If the activity is in the inlet, change the liner (of course there should not be glass wool in the liner), if in the column, try a new one ( there is not a lot that you can do to resuscitate active columns).

When you say that the reproducibility is "> 5", what are you calculating ?, is this the relative standard deviation (sd / mean) ?, and is it in % ?.

Are you using an internal standard ? If not and your reproducibility is 5% for raw peak areas of alcohols (and better than this for other components) you are doing about as well as the hardware you are using will allow.

You mention solvent mixes. Why not do liquid injections ?

Peter

[Anthony Doyle]

Hi

The alcohol peak shows tailing, manifest as a shoulder peak. This shoulder isn't in the blank run. I was using a deactivated general purpose split/splitless liner, that contains glass wool. Why not glass wool?

The system precision is measured by sd/mean * 100 (Coefficient of Variation). I'm using a capillary column. I'm not using an internal standard. We received this method and a validation report from a supplier, and are using this starting point for optimisation.
Any feedback would be much appreciated

Anthony

[Peter Apps]

Hi Anthony

Tailing is not the same as a shoulder on a peak; a tailed peak descends slowly and smoothly back to the baseline, a shoulder is a marked change in the rate of descent.

I have done an intensive optimisation study on the MPS2 - 6890 combination for headspace analysis of ethanol in water and if you are getting shoulders on your peaks it is most likely due to flow disturbances in the inlet. If the injection rate from the syringe is more than the carrier gas volume flow rate the extra volume flow as the sample is injected disrupts the Agilent EPC pressure control at the inlet and you get erratic split ratios and/or sample diffusing out through the septum purge. If you are doing split injections this should not be a major cause of problems unless you are injecting close to the maximum syringe rate.

The peak area repeatability performance spec for the MPS2 is 3.5 or 4 % (forgive me if I do not take the time to look it up !), so your 5% rsds are not far off the performance limit of the instruments.

The reason not to use glass wool is that it has a huge surface area with correspondingly abundant active sites, and since your sample is already in the gas phase you do not need the glass wool to help the evaporation of the sample or to retain bits of crud and high boilers that might be in it.

Regards Peter

[Anthony Doyle]

Hi Paul

The carrier gas flow rate is 1.0ml/min, the split is 5:1, so the 'total flow' in the inlet is 9ml/min. To double check - should I be looking at the total inlet flow or the carrier gas flow to determine the injection speed? We are currently working near the maximun injection rate (500ul/s where the max is 1ml/s) so this is surely a problem cause.

Unrelated - I've noticed, when using glass wool liners with headspace, that the septa tend to bleed and pieces collect on the liner. I was going to try high temp/low bleed septa to deal with this, as liners without glass wool are recommended. Any thoughts?

Appreciate you help

Anthony

[CE Instruments]

Are you sure your GC will control a split ratio of just 5:1 accurately. I was always told that it was not advised to run below 10:1 ?

[Anthony Doyle]

Yes thanks

I have actually validated methods with lower splits so this is not a problem

[Peter Apps]

Hi Anthony

Yes.I think that your injection speed is too high, try dropping it to 150 - 200 ul/s to match the total flow. If you go much slower than this you might see peaks that are broad and flat on top. You could also reduce the injection volume or increase the split if you have big enough peaks - the plan is to reduce pressure fluctuations in the inlet.

Little bits of septum falling into the inlet are not what is usually understood by septum bleed, which is the slow evaporation of volatile components from the septum, not its physical disintegration. With a dome- or conical-tipped needle I would be surprised to see much septum disintegration unless the septum nut is screwed down too tight. Check the alignment of the MPS with the injection port just in case.

Septum debris in the inlet can cause peak tailing, but it usually affects non-polar high MW compounds much more than the low MW alcohols that you are looking at.

It is a good plan to insert the column about 5 mm deeper into the inlet than Agilent recommends. At the recommended insertion depth the column tip sits just at the top of the narrow gooseneck at the bottom of the inlet liner, and all the solid crud gets funnelled down on top of it. If it sticks up a bit the crud falls alongside it.

I am still wodering why you are struggling with headspace when it sounds as if you could be doing liquid injections ??

Regards Peter

[Anthony Doyle]

HI Peter

Apologies for the tardy reply. Your advice came in very handy. The reason we are going with headspace is that the method is for checking residual solvents levels in active pharmaceutical ingredients. The sample is reasonably concentrated, and when injected directly, the chromatogram contains a lot of unknown peaks, with the toluene having very poor peak shape incomparable to the standard.

Thanks
Anthony

[Ron]

Anthony,

One of the main factors affecting the reproducibility in headspace is the sealing of the vials. This is especially true with the headspace units that use a magnetic vial transport. Steel seals are difficult to crimp consistently. The best results are usually obtained using screw cap vials.

Many headspace methods use an internal standard. This can have a very significant effect on precision. If you look at the USP general chromatography section the use of internal standard is not requred, but is allowed.