Chromatography Forum

Dear all,

there is the following interest situation.

According to the official methods in Europe, USA, Mexican Federation, NIS (new independent states, ex-USSR) etc the alcohol drinks control laboratories should determine the volatile compounds in spirit drinks and express concentrations of these compounds in milligrams per liter (mg/L) of absolute alcohol (AA).
Careful analysis of published papers and official documents related with determination of volatile substances in spirit drinks showed the absence of any mention of direct determination of volatile compounds concentrations expressed in milligrams per liter (mg/L) of absolute alcohol. Result of search was surprising for us. And after long discussions we have decided to verify/revise this idea experimentally. Analysis of obtained data shows that this new method works, really. Indeed, it is possible at the present time to introduce this approach for routine practice of analytical laboratories due to modern GC with wide range signal registration from flame ionization detector (FID). The linear range of modern FID is generally more than 10^7. Signal registration from impurities compounds and from main component ethanol takes place without any distortions. Illustrative materials supporting reality of the proposed method is here http://www.inp.bsu.by/labs/lar/eis.html .

We look for researchers from alcohol drinks control laboratories to check this method in their daily practice. We propose to develop the obtained experimental materials and send the joint complete paper for publication in the leading international scientific journal, for example, Agriculture and Food Chemistry or Food Control.

Best regards,
Siarhei

The problem is with measuring small in big. The amount of ISTD here is extremely large, unproportionally large. As a rule of thumb, the response of ISTD should be comparable in response to analytes. The error of calibration will be the dominant component of uncertainty balance. Most people use 1:1 ration for precise measurements.
Secondly, composition of mixture may change in hot injector. So for accurate identification of esters, acids, and alcohols this method can be used only as tentative indication of quality rather than actual organoleptic property. Hot GC data are rather a marker which equivalency to methods that will not change the composition of drink has to be demonstrated. But due to complexity and variability of matrixes this can not be achieved in practice. Cold on column injection - may be. EtOH as ISTD for calculation - definitely not.

It seems that the Organisation International du Vin has published a GC method that we have adapted to our needs.
I suggest you to search their published methods. (OIV)
Salutations

The problem is with measuring small in big. The amount of ISTD here is extremely large, unproportionally large. As a rule of thumb, the response of ISTD should be comparable in response to analytes. The error of calibration will be the dominant component of uncertainty balance. Most people use 1:1 ration for precise measurements.
Secondly, composition of mixture may change in hot injector. So for accurate identification of esters, acids, and alcohols this method can be used only as tentative indication of quality rather than actual organoleptic property. Hot GC data are rather a marker which equivalency to methods that will not change the composition of drink has to be demonstrated. But due to complexity and variability of matrixes this can not be achieved in practice. Cold on column injection - may be. EtOH as ISTD for calculation - definitely not.

Dear Alexandre,

many years ago there was a problem of small impurities and basic compound (solution) measuring simultaneously. The linear range of modern FID is generally more than 10^7. Signal registration from impurities compounds and from main component ethanol takes place without any distortions.

Experimental illustration that the error of calibration will --- not --- be the dominant component of uncertainty balance is in the Fig.5-8 on the following page http://www.inp.bsu.by/labs/lar/eis.html .

In the case of determination of volatile substances in spirit drinks ethanol as ISTD is an elegant way to apply method of ISTD directly.

And the last. All people inject samples in hot injector. We do the same. We don’t propose anything new.

Best regards,
Siarhei

It seems that the Organisation International du Vin has published a GC method that we have adapted to our needs.
I suggest you to search their published methods. (OIV)
Salutations

Dear Guillermo!

Would You like to send to me some references about this GC method.

Sincerely,
Siarhei

The linear range of the detector is not the limiting factor in the proposed analysis, so quoting the claimed 10E7 range is not particularly relevant.

The practical limitation on the analysis scheme as a whole is the chromatographic step, not the detector. The problem is to separate a huge quantity of ethanol from relatively tiny quantities of other compounds such that both the ethanol and the other compounds can be properly integrated on a reasonably flat and clean baseline. Ethanol usually gives tailing peaks - you can see this in your chromatograms, so the minor peaks are riding on a sloping baseline, which you can also see, and this compromises repeatable integration. If the tail gets big enough the baseline slope is so steep that the minor peak is not even detected. With large enough quantities of ethanol on the column to allow the minor peaks to be detected you are very likely to get some solvent effects that will distort the minor peaks and affect their retention times. The figures that you pasted are not all that clear, but I would guess that you have a signal:noise (short term) of about 5:1 on your minor peaks - this is not going to give repeatable integration.

Bear in mind also that with the small, oxygenated molecules that you are analysing the FID relative response is not the same for all compounds.

To just inject and integrate is seductively simple, but I doubt that it will work.

Peter

It seems that the Organisation International du Vin has published a GC method that we have adapted to our needs.
I suggest you to search their published methods. (OIV)
Salutations

Dear Guillermo!

We investigated the site of the International Organisation of Vine and Wine including all their manuals devoted to METHODS OF ANALYSIS FOR WINES. Nothing new is for us there. Our GC method calculation differs principally from the OIV one.

Could you mention some articles or documents devoted to GC method that you have adapted to your needs.

Cheers!
Siarhei

This post was been deleted.

The linear range of the detector is not the limiting factor in the proposed analysis, so quoting the claimed 10E7 range is not particularly relevant.

The practical limitation on the analysis scheme as a whole is the chromatographic step, not the detector. The problem is to separate a huge quantity of ethanol from relatively tiny quantities of other compounds such that both the ethanol and the other compounds can be properly integrated on a reasonably flat and clean baseline. Ethanol usually gives tailing peaks - you can see this in your chromatograms, so the minor peaks are riding on a sloping baseline, which you can also see, and this compromises repeatable integration. If the tail gets big enough the baseline slope is so steep that the minor peak is not even detected. With large enough quantities of ethanol on the column to allow the minor peaks to be detected you are very likely to get some solvent effects that will distort the minor peaks and affect their retention times. The figures that you pasted are not all that clear, but I would guess that you have a signal:noise (short term) of about 5:1 on your minor peaks - this is not going to give repeatable integration.

Bear in mind also that with the small, oxygenated molecules that you are analysing the FID relative response is not the same for all compounds.

To just inject and integrate is seductively simple, but I doubt that it will work.

Peter

Dear Peter,

the modern capillary columns give sufficient separation for minor compounds and for main compound, ethanol, too. See illustration here http://www.inp.bsu.by/labs/lar/eis.html .

The following experiments have been done to check metrological characteristics of the proposed method.

Six standard ethanol-water (96:4) solutions were prepared to generate calibration curves. There were the following levels of volatile compound concentrations: 13 mg/L, 20 mg/L, 500 mg/L, 1000 mg/L, 5000 mg/L and 20000 mg/L for methanol and 2 mg/L, 5 mg/L, 10 mg/L, 100 mg/L, 500 mg/L and 2000 mg/L for another eight volatile compounds. Every standard solution was injected three times. Analytical characteristics of the obtained calibration curves are presented in the Table. As can be seen in Table, correlation coefficients R^2 for all compounds are higher than 0.9996. Detection limits were determined by analysis of low level standards. The detection limits are between 0.235 mg/L for isobutyl alcohol and 0.394 mg/L for methanol.

Compound______Linear_range_(mg/L)____Slope____Correlation_coefficient_(R^2)
acetaldehyde____2.24_-_1990__________1.559_____0.9996
methyl_acetate__2.09_-_2000__________1.517_____0.9997
ethyl_acetate___2.20_-_2094__________1.247_____0.9998
methanol_______13.0_-_20045_________1.377_____0.9999
2-propanol______3.74_-_2033__________0.914_____0.9998
1-propanol______1.99_-_2094__________0.809_____0.9998
isobutyl_alcohol__2.23_-_2000__________0.674_____0.9998
n-butanol_______1.98_-_2000__________0.737_____0.9998
isoamyl_alcohol__2.18_-_2073__________0.681_____0.9999

Table. Analytical characteristics of the obtained calibration graphs for volatile compounds in standard ethanol-water (96:4) solutions.

And now we invite researchers from Alcohol Drinks Control Laboratories to check this method in their daily practice. We propose to develop the obtained experimental materials and send the joint complete paper for publication in the leading international scientific journal.


Best regards,
Siarhei

Why do you use an ethanol content in your standard that is about double what it is in full strength potable spirits ?

While your r-squared values are good (I presume that you took means of the triplicates at each level) you do not give us the residuals, the repeatability RSD for the minor compunds, or the repeatability RSD for the ethanol peak area.

Before inviting other labs to invest their time in testing your method I think that you need to run and publish a full validation on it in your lab, then invite other labs to take part in an inter-lab reproducibility trial.

Peter

I can send you a copy of the proceding we use. (Spanish version, I'm sorry...)
Send me your e-mail address.
Regards

I can send you a copy of the proceding we use. (Spanish version, I'm sorry...)
Send me your e-mail address.
Regards

Dear Guillermo!

Thanks, many.
My e-mail address is: chere@inp.bsu.by .

Best Regards,
Siarhei

In reference primary methods the requirement to conc. of ISTD is that it has to match as much as possible the concentration of the analyte. In the reference labs the procedure is - screen, estimation, adding of ISTD as 1:1 and calculation without calibration graph. In routine labs it has to be justified by validation and requirements to accuracy and precision set by your regulator. If it is +/- X% around 0.001%-0.01%, just tell us that X and show the data. You also need to provide a comparison of the data obtained by proposed method and the official reference method.

On the examples provided in the link the EtOH peak was tailing badly. It was also integrated as B to X% tail, not as Baseline to Baseline. Did the EtOH peak pass requirement for tailing and symmetry? Can a better column be used?

With this integration the amount of the EtOH was obviously under estimated, and if integrated properly the result of impurities to EtOH could be almost 2 times lower. I see a problem with this integration in settling disputes with customers, who work hard to make good booze? In their labs they may integrated B to B.

Why do you use an ethanol content in your standard that is about double what it is in full strength potable spirits ?

While your r-squared values are good (I presume that you took means of the triplicates at each level) you do not give us the residuals, the repeatability RSD for the minor compunds, or the repeatability RSD for the ethanol peak area.

Before inviting other labs to invest their time in testing your method I think that you need to run and publish a full validation on it in your lab, then invite other labs to take part in an inter-lab reproducibility trial.

Peter

Dear Peter,

the RRMS values in the case of ISTD for all minor compounds are presented, for example, in the Fig. 5 and 7.
Now let’s compare RRMS values for method of external standard (ES) and for method of ISTD (ethanol, main compound). See Fig. 8 on the page http://www.inp.bsu.by/labs/lar/eis.html .
The value of RRMS for methanol is 7,9812% and 9,4101% for ethanol, respectively, in the case of external standard. The value of RRMS for methanol is 1,2776% in the case when we use ethanol as internal standard (see Fig. 7). There is essential reduction of RRMS more then six times from value 7,9812% to value 1,2776%. Feel the difference.

In order to study accuracy of the proposed methodical approach in the case of large ranges of volatile compounds concentrations 6 – 20000 mg/L for methanol and 1 – 2000 mg/L for another eight volatile compounds reference ethanol-water solutions (96:4) were prepared with known concentrations of volatile compounds. Every reference solution was injected 30 (15 × 2) times. The repeatability in the worst case for lower concentrations 1 mg/L did not exceed 3.6 %. The obtained experimental results are presented below in the Table.


Compound____Known_conc._of_i-th_____Conc._measured_by_ISTD____Relative_discrepancy
_____________compound_in_standard___________________________________________
________________(mg/L)________________(mg/L)_______________________%______

acetaldehyde______1.158_________________1.129_______________________-2.50____
________________5.137_________________5.182_______________________+0.88____
________________10.11_________________9.921_______________________+1.87____
________________99.64_________________93.86_______________________-5.80____
________________497.6_________________481.1_______________________-3.32____
________________1989__________________2037_______________________+2.42____
methyl_acetate____1.000__________________1.005______________________+0.50____
________________5.000__________________5.121______________________+2.42____
________________10.00__________________9.905______________________-0.95____
________________100.0__________________96.35______________________-3.65____
________________500.0__________________484.9______________________-3.02____
________________2000___________________2042______________________+2.10____
ethyl acetate______1.047__________________1.072______________________+2.39____
________________5.234__________________5.374______________________+2.67____
________________10.47__________________10.45______________________-0.19____
________________104.7__________________102.0______________________-2.58____
________________523.4__________________512.2______________________-2.14____
________________2093___________________2115______________________+1.02____
methanol_________5.975__________________6.044______________________+1.15____
________________53.07__________________53.51______________________+0.83____
________________103.2__________________102.97_____________________-0.22____
________________1005___________________988.1______________________-1.68___
________________5013___________________4987_______________________-0.52___
________________20045__________________20118______________________+0.36___
2-propanol_______2.636___________________2.645______________________+0.34___
________________6.698___________________6.754______________________+0.84___
________________11.78___________________11.77______________________-0.08___
________________103.0___________________101.0______________________-2.13___
________________509.0___________________503.2______________________-1.22___
________________2033____________________2047______________________+0.69__
1-propanol_______1.047___________________0.997______________________-4.78___
________________5.234___________________5.223______________________-0.21__
________________10.21___________________10.23______________________+0.20__
________________103.2___________________100.2______________________-3.10__
________________523.4___________________513.6 _____________________-1.87__
________________2094____________________2125 _____________________+1.51__
isobutyl_alcohol___1.000____________________0.971_____________________-2.90__
________________5.000___________________5.033_____________________+0.66__
________________10.00___________________9.82______________________-1.80__
________________100.0___________________97.7______________________-2.30__
________________500.0___________________491_______________________-1.80__
________________2000___________________2032______________________+1.60__
n-butanol________1.000___________________0.991_____________________-0.90___
________________5.000__________________5.061______________________1.22___
________________10.00__________________9.89_______________________-1.10__
________________100.0__________________97.10______________________-2.90__
________________500.0__________________491.0______________________-1.80__
________________2000___________________2036______________________+1.80__
isoamyl_alcohol____1.036__________________1.003______________________-3.19__
________________5.182__________________5.169______________________-0.25__
________________10.37__________________10.21______________________-1.54__
________________104.0__________________101.0______________________-2.60__
________________518.0__________________510.0______________________-1.58__
________________2073___________________2110______________________+1.78__

Best Regards,
Siarhei

Hi Siarhei

You have impressive data - but posting them on the internet will not advance your cause. You need to write them up in a paper for a peer reviewed journal.

Peter