High-level Ethanol Method?

[chromy]

I have a Thermo Focus GC with a TriPlus autosampler and a Stabilwax 15x0.32x0.25 capillary column, and an FID. Right now we are using it to determine % ethanol on beer samples, using headspace, on undiluted samples.

My lab would like to use this same GC (we could change the column if necessary) to analyze higher-level alcohols for % ethanol. The beers tend to run from 0.5% to 6% ethanol, but we'd like to be able to analyze liquors which run from 6% to about 98% ethanol.

Evidently, before I came to this lab, they worked on dilution schemes, but were not able to get consistant results.

I've looked in many different places, including the archive here, to try to find a method we can use to do this. My lab is willing to send me for training on method development, but from my (very limited) knowledge of GC, it seems like we'd be re-inventing the wheel! Our lab is quite small, and we only need this one method. I don't see the point in spending huge amounts of money and time on this, IF we can acquire a method that is already in use.

There are two sources that look promising to me: first, an ASTM Method D5501 (GC ethanol), and second, an AOAC method in their Official Method of Analysis, 17th Edition.

Bearing in mind that I haven't actually laid eyes on either of these methods yet, can anyone tell me if I'm on the right track?
Sorry this post is so long!

Chromy

[chromatographer1]

How are you doing the headspace analysis?

That is the critical component of your analysis.

I have a method I developed over 10 years ago using a Tekmar 7000 (Varian Genesis) or a Perkin Elmer HS40 that will do 100% ethanol to ppm using the same methodology.

The linearity depends upon diluting EVERY sample the same amount while not overloading the detector or the concentration of ethanol in the headspace.

I would place 10µL of sample and dilute it with 990 µL of water in a 2mL vial. (1:99 dilution) Then I would take 1µL of that and add 99µL of water into a 6mL headspace vial. That will give you a 10000 fold dilution which should give you a linear range easily of 1% to 100% alcohol in the first sample. One could double this concentration if a lower or more accurate determination is required by a reduction in volume of the final sample.

I would heat the vial for 12-15 minutes @ 85°C and then inject a 100 to 500µL gas sample.

This should give you a place to start. Most published methods use an excessive amount of sample which contributes to non-linearity and variability of results. It also requires an excessive amount of equilibration time to achieve useable results. Read my article in 1997 June Analytical Chemistry about fast equilibration and accurate 1ppm analysis using only 1mg of sample. Analytical Chemistry Vol.69, No. 11, p2221-3

best wishes,

Rod

[chromy]

Thanks, Rod! I'll look for your article and see if it might work for us. I don't know how accurate their dilutions may have been.

For the beer, we use 20 mic's of straight sample in 2 ml of internal standard (t-propanol). (I hope that no forum members will take this as a challenge to debate the use of internal standards. ) The autosampler incubates the samples at 90 degrees C for 5 min., then injects with 1 ml gas sample. The split ratio says 40.

We don't need to get down to ppm level, we are only looking for 4% error.

I do realize that this is terribly mundane and dull, but you'd be saving my state's taxpayers a lot of money (it's a state lab). Does anyone have feedback on the 2 sources I mentioned?

[Consumer Products Guy]

Chromy - this is straightforward, but I can't provide all details as some parts are proprietary. We use a modified AOAC Official Method 973.23, Official Methods of Analysis of AOAC International (16th edition, 36.1.01.). Remember, tons of published or official procedures are antiquated (this one specifies a packed column). Here we assay up to about 75% ethanol (v/v). We make up an internal standard solution of n-propyl alcohol in water, and add 25 ml of that to deliver the 1.00 ml n-propyl alcohol internal standard AOAC states (or you could add separately). Sample size is about 1 ml, into a capped vial. Shake the stuff, GC on PEG-type 0.53 mm column at 40C, split injection, automatic injector; KEEP INJECTION SIZE TO 0.50 ul since your solvent is water. We get excellent reproducibility. We use 99+ ethanol (SD-40B 200 proof) as our standard, remember some reagent or HPLC alcohols are only 90% actual ethanol, a problem we found with a contract lab.

[chromatographer1]

CPG post is helpful and I would agree that doing high concentrations one is better off doing direct injections instead of headspace for ease of use.

If you want additional information and options

you might call Supelco Technical Service for assistance for measuring alcohol in beer wine etc and they can give you methods that others have done for this kind of analysis.

800-359-3041

techservice@sial.com

best wishes,

Rod

[chromy]

Thanks, Consumer Products Guy, I really appreciate your help. Just knowing where to start looking is a big step forward. I'll check our libraries here in town and see if I can locate the AOAC 16th, and maybe 17th, Editions. It will be interesting to see if the two editions differ much for this method. Of course we would not use the method verbatim without paying AOAC for it. But I need to see it first, to know if we can adapt it or not.

Evidently the ASTM method was for ethanol in fuels.

Am I using the right type of column, I'm assuming that a column that separates ethanol and t-propanol well with low ethanol levels, will also separate well at high levels? It does do a good job at the 0.5-6% level.

Also, thanks for the heads-up on the reagent alcohols!

Thanks again, Rod, I'll check that out also. I'd hate to spend months on something that already exists somewhere else.

Any other advice or input would be greatly appreciated.
Chromy

[Peter Apps]

Hi Chromy

If you are happy with your beer method, then your simplest option for the higher concentrations is to dilute them down to beer levels and then analyse the diluted samples as if they were beer.

Can you clarify your use of internal standard - as I read it you add 20 microlitres of sample to 2 ml of standard (propanol), and then run the headspace. Doesn't this give a huge propanol peak ?. Or is the propanol a solution in water ?

Peter

[chromy]

Thanks Peter, I will give it a try. It is possible that when they tried dilution before, perhaps they were not as careful as you need to be when working with multiple dilutions. And we have a grand total of three pipettors here and I'm not sure these are calibrated or maintained...yeah. It might make sense for me to go back and work with dilution before totally abandoning our current method...

The propanol we use IS diluted in water, sorry that was unclear.

Rod, I'm curious about your comment that we might want to do direct injections instead of headspace. It's my impression that we chose headspace because in beer/liquor there are lots of sugars and things that could gunk up our injector, column, etc. But possibly the amount of dilution that we'll be doing would make that less of a problem? But then, I also thought that large amounts of water might be hard on the column? This is all based on lots of reading and no experience , unfortunately.
Chromy

[Peter Apps]

Hi Chromy

You will not need multiple dilutions - you can dilute 10:1 in one step to bring even pure ethanol down into the beer range.

As long as you dilute your standards and samples using the same pipettes it will not matter if the pipettes are calibrated because all the systematic errors will cancel out.

If you do not have the pipettes etc then do everything by weight, record the actual weights, and throw it all into a spreadsheet to calculate the actual dilution. This is usually a lot quicker and more accurate than volumetrics.

Injecting aqueous samples makes me nervous, but other people are quite happy with it. Distilled spirits are certainly safe to inject on wax columns, and it would be more straightforward than dilution and headspace. The drawback would be that you would need to change your GC setup for different types of samples.

Peter

[chromatographer1]

For decades alcohol and other related compounds have been analyzed using Carbowax on Carbopack supports, usually 5 or 6.6% loadings.

Bourbon, brandy, beer, wine, are all tested this way.

If diluted and the injection liner is replaced regularly, especially if a packed injection liner is used, a packed column will accept hundreds of injections. It is a very cost effective means of doing a routine analysis.

However, headspace is also a very viable option, although great care must be taken in producing the samples, both dilution, and crimping the vials reproducibly.

I did not use volumetric means of making dilutions. It is much more accurate to use volumetric flasks and to weigh additions to an already existing solvent (such as water) in the flask, then add water to the precise volume of the flask (say 50 or 100mL).

If you calculate the density of any solution (weight of the water and weight of each compounded added) per volume of solution, this first dilution can be weighed into another dilution flask for lower concentrations.

Since most balances will give you 4 or 5 significant figures of accuracy, that is much better than the 1% accuracy of most pipetters or pipettes assuming perfect transfer of liquids.

The greatest error is in the flask (type A are 1% accuracy) and that cannot be helped. It does allow you to make many dilutions from the same stock solution and eliminate a great many opportunities for error using this method.

I have found that headspace analyzers [both Varian Genesis (aka Techmar 7000) and Perken Elmer's HS40] are more precise and reproducible than are the lab techs who prepare the samples. I have no experience with other brands of analyzers, but I would expect the results to be the same.

I wish to emphasize that most errors in headspace analysis are caused by analysts trying to use too large a sample at too large a concentration.

Smaller samples and 50 to 5000ng amounts of solvent in a 50 to 100µL volume will give you the best results, although higher and lower concentrations are possible. For example in my article I used 0.5ng to 1µg of each of 18 solvents in 25µL of water using a 3mL headspace vial.

This was to demonstrate 1 ppm solvent determinations in 1mg samples of drug or other matrices, the LOW end of such analyses. But much higher levels can still be ascertained accurately and quickly, even as high as 100µg amounts using a 6mL vial.

Good luck and good hunting.

Rod

[chromy]

Doh! You're right Peter, no multiple dilution needed. *Banging forehead repeatedly on desk* Good point about using the same pipettes, my boss has been pre-mixing the internal standard, but we could easily change that. At this point I'd like to get our pipettes on a maintainance & calibration schedule, since we would eventually like to get accredited and I'm interested in dragging our lab out of the cave it's in currently. Heh.

Rod, I like your point about the Carbowax. For now I'd like to try our current column (since we already have it) and headspace with dilution, and see how that works out. If it's too shaky, we can go back to more basic ways (probably where we should have started) and see how that does.

Thanks again for your helpful input!