best septa for headspace of ethanol

[Peter Apps]

I am using PTFE faced silicone septa for headspace analysis of ethanol in water, and it looks as if I am getting eratic absorption of the ethanol into the septum.

Does anyone have a recommendation for septum material for this particular analysis ? - many labs use butyl rubber but I had assumed that this was just because it is cheaper.

Thanks Peter

[chromatographer1]

Peter,

I tried them all many years ago and found the PTFE faced silicone septa was the best.

I doubt this septa is your problem unless you have a very bad lot.

More likely the vials are leaking from the glass lip being inconsistent or the crimping tool is not sealing the septa at the proper tightness and you are getting inconsistent sealing of the cap.

Butyl rubber will absorb many solvents quite inconsistently, especially esters and aromatics. DON'T USE IT.

Change the setting on your crimper until you get a good seal. Some lots of septa may be out of spec for thickness which can also be a cause of your problem.

best wishes,

Rod

[Peter Apps]

Thanks Rod

I am sure that it is not leaks that are the cause of the problem - the main symptom is that in batches of 12 replicates the repeatability is worse towards the end of the batch; rsds of peak area are e.g. 0.7% over the whole batch, 0.93% for the last 6, 0.45% for the first 6, another e.g. 0.87% for all 12, 1.14% for the last 6 and 0.27% for the first six. This is with all the vials loaded into the G1888 sampler at the same time, so they spend increasing times at room temperature (and not pressurized) waiting to drop into the sample oven. When I fill and load each vial just before it starts equilibriating (so all the vials spend the same short time at room temperature) rsds are consistently 0.28% over the whole batch. Deterioration in repeatability has been a problem despite drastic changes to the hardware and operating conditions.

Another odd thing is that the ethanol area varies independently of the t-butanol internal std, so it is not just a simple volume and pressure problem, there must be some ad- or absorption or some diffusion going on.

Have you ever tried aluminium-faced septa ?

Peter

[chromatographer1]

Peter,

Yes I have tried aluminium coated septa. Chlorinated analytes are a problem.

I would have some suggestions and questions.

First your repeatability is wonderful. No, leaks are not the issue.

Certainly there can be issues with the quality of septa.

I don't know how long you allow the samples to sit between loading and the last analysis of the last vial. If it is several hours then yes I have seen reproducibility become more variable. In general, samples should not be allowed to stand more than 2-4 hours before analysis. 12-24 hours is definitely a no-no.

Higher concentrations of analyte tend to have more variability.

Teflon is permeable. Esters, aromatics, and chlorinateds are the worst offenders.

You have to also consider if there are reactions going on with the matrix.

I never had to heat alcohols more than 15 minutes at 80° from an aqueous matrix to achieve an adequate reproducibility, ie less than 1% of measurement of a 100 ppm to 5% amount.

Aluminium faced seals should improve your results for ethanol but may not for other analytes.

You are quite a rigorist in your RSD expectations. I admire your standards and the quality of your work.

Cheers !

Rod

[Peter Apps]

Hi Rod

Thanks again. I have the considerable benefit of working in a metrology lab, and so I can focus on repeatability pretty well to the exlcusion of everything else.

My samples are ethanol in water, and so I can select septa that work for this analysis, even if they are no good for anything else. I will give aluminium-faced septa a try. With these samples I do not expect any reactions with the ethanol.

I also use a 15 min equilibrium (with shaking). This is as much to allow the oven temperature to settle after the vial drops into the oven as to allow the sample to equilibriate with its headspace. With a batch of twelve samples (preceded by a dummy vial to cycle all the valves etc) the last sample spends nearly 3.5 hours in the queue (GC cycle time is 17 min) which is probably too long. Ideally of course I would like to be able to run overnight, which implies 16 hours in the queue ! Interestingly enough, in a series of 50 samples the repeatability plateaud after the first 12 samples.

I am back into the lab today after the Christmas - New Year break and I have two batches of 12 samples that have been in the vials since before Christmas. I'll let you know how they turn out.

All the best for 2008.

Peter

[chromatographer1]

Using the PTFE faced septa there is an equilibrium for each and any solvent to which it is exposed. This takes time at room temperature to achieve.

When heating the septa on the vial another equilibrium is achieved (if heated long enough) as the ethanol in the septa material permeates through the PTFE layer, which may be in either direction, depending upon the type of material involved, the concentrations and number of analytes present, and the time and temperatures involved in storage and in heating prior to analysis.

Your heating temperature and time should make the recovery become consistently variable (not necessarily acceptably variable), although as you vary these factors the variability can change.

The plateau of RSD data duplicates results I saw 15 years ago using the Varian Genesis (Tekmar 7000) and the PE HS-40.

The biggest factor in variability I found in my research was the consistency of the seal of the septa, which was based on the proper thickness of the septa, the tightness of the crimper application, and the surface of the glass vial where the vial seals.

Once these factors were controlled other factors of time of heating, septa surface chemistry, and analyte interactions could be monitored and measured more accurately.

good luck in your research.

Rod

[Peter Apps]

I have been trying different lots of septa - all teflon faced silicone from one manufacturer (no names no pack drill). The repeatability over batches of 12 varies from lot to lot, and the pattern of deviations from perfect repeatability also varies - some lots give a uniform trend of area with run number, some give eratic variability but no trend. This behaviour is reproducible from batch to batch within a septum lot.

Running seven levels of CRMs, with six replicates at each level gave the following results for % rsd at each level:

10mg/100ml: 0.088%
20mg/100ml: 0.48%
50mg/100ml: 0.18%
100mg/100ml: 0.096%
200mg/100ml: 0.080%
300mg/100ml: 0.11%
500mg/100ml: 0.65%

The whole set of 42 was run as one continuous sequence BUT the 20 and 500 mg/100ml sets of 6 were queued (at room temp) for about twice as long as the others because they were loaded at the same time as the 10 and 300mg/ml vials respectively. The others were filled to vials and loaded into the autosampler just as the last vial of the previous batch of six was about to drop into the oven to equilibriate. Plainly time spent at room temperature before equlibriating has a marked influence on repeatability. These were all done with septa from one lot.

I tried aluminium faced septa but I could not get them to seal gas tight.

I have now ordered as many kinds of septa from as many sources as I can in order to run a comparison.

Peter

[matthieu]

Hi Peter, Rod,

Interesting topic here.
I also faced this RSD problem during qualification on the same equipment (agilent G1888-6890GC)
I ran a serie of 10 samples, (aceton and ethanol in water) and noticed a variation in areas as peter mentionned. With the first 6 samples, I had RSD around 0.5% on both analytes, the last four gave 0.8%. However the global RSD on the 10 samples is around 2%!!
This phenomenon is more important with aceton than ethanol.

Here are some parameters:
sample: 1mL of 0,1% v/v solution of ethanol and aceton in water, in 20mL headspace vials (I know i should use smaller vials, but sop is sop!!!)
Heating at 70°C for 20 minutes (with shaking)
GC run time 4 minutes.
Vial Septum: PTFE

So Peter, if you find a better septum please let me know!!

Cheers,

Matthieu

[Peter Apps]

Hi Matthieu

Very interesting - If you plot peak area vs run number do you see a trend ?. How repeatable was the ethanol: acetone area ratio ? What was the temperature of the 6-port valve and the sample loop ?, I found that if the loop and valve are set at 120 C I get stronger trends and poorer repeatability than with it set at 90 C (the oven is at 70C). Initially I thought that this was due to changes in volume flow and pressure as the sample flow heated up in passing from vial to loop, but the effect was stronger if the vial septum was held on the needle for longer by increasing the pressurization time (the pressurization flow was off since I control the pressurization solenoid from the GC, not from the 1888) and so I now think that the effect was due to the septum heating up while held against the port below the hot valve block, and bleeding absorbed ethanol back into the vial headspace.

I should mention that the whole setup is drastically modified from how it came out of the factory.

With a 4-minute cycle time I doubt that the sample oven temperature is stable - it takes 6-7 minutes for the effects of dropping a vial at room temperaure into the 70 C oven to wear off, this is a disadvantage of the otherwise good technology of a stirred air oven. You do not see this on the 1888 front panel readout, which shows the temperature of the oven wall, not the air inside.

Time to build a better mousetrap .....

Peter

[matthieu]

Peter,

For Infos: The loop temp is set at 85°C and transfert line et 95°C. I did not try any higher setting points.
I agree with your comment about oven temp stabilisation, my first tought was that it doesn't has any real influence on my samples as the incubation time is quite long (20min) compare to the time needed to reach equilibrium under my conditions (aroud 12 minutes for ethanol, 9 minutes for aceton). but it's just a guess..I will take a look on this parameter.
But i still think that if it has influence, i will see variation much earlier than at the 6th sample. However this could explain why I have 0,5 to 0,8%RSD where you have 0,2 (This and the "black-hand manipulator" factor!!)

Regarding the results: I don't see a trend in areas, but a big"step".
To be clear: for ethanol I had areas around 240 (atlas unit..if i remember well is µV.min) for the 6 first samples with good RSD, then i suddenly drop to 227 for the last four.
I notice the same "step" with aceton, although it appears one sample earlier (5 samples at aroud 940, 5 samples at 890-900)

The RSD on area ratio is 2%.

Next week I will go to another customer site to qualify the same equipment (G1888-6890), I will let you know if I find "surprising" results!

Matthieu

[Peter Apps]

Hi Matthieu

I get an upwards trend in areas, while you get a downward drop, so I think that we are looking at two different problems.

With an equilibrium time of 20 min and a cycle time of 4 min, the 1888 (and others) has to wait for sample 5 or 6 to inject before it can start equilibrating 6 or 7, because 20 is an exact multiple of four. There is then a difference in the queuing at room temperature, which might give the step in area that you see. If you decrease the equilibrium time to 19 min or increase it to 21 min you avoid the clash between injection and equilibrium, and the step might disappear.

Peter

[matthieu]

hi Peter,

Obviously we are not facing the same problem.
Thank you for the tips about time managment, I'll take a look on results with 19 min heating instead of 20. (this week qualification is cancelled, FIDs don't reach temp!)..more news in few weeks!

Cheers,
Matthieu