Procedure for Alcohol Determination by Distillation

Transcription

1 Page 1 14 November 1998 Procedure for Alcohol Determination by Distillation 1.0 Introduction The following notes discuss the procedure for determination of the alcohol in beer based upon ASBC MOA Beer 4A as implemented in our laboratory. Comments are also included on the expected recovery and accuracy of determinations done in this way. 2.0 Method The method consists of execution of the following enumerated steps. See Figure 1 for an illustration of the configuration of the apparatus. 1. Pour beer into 100 ml volumetric flask (use the one labeled 7 as its mark is high up on the neck.) Use a 1 ml Tensette pipet to suck foam out of neck. Fill to the mark or just above. Stopper and place in 20 C water bath. 2. While beer is attemperating pull microwave oven away from wall. Place jack stand on top of microwave oven. Elevate so that top of platform is about 8 above top of microwave. Place 500 ml mantle on top of jack stand with control knob facing about 30 to the right of forward. 3. Place Cole-Parmer ringstand (with white base) to left of microwave oven. Base should be behind rod and rod should carry two clamps oriented away form base i.e. towards front of microwave. 4. After beer has attemperated, suck out excess beer and foam so level is at the mark. 5. Fill two 25 ml flasks with DI water to near the marks. 6. Pour beer into 500 ml distillation flask. Rinse with two 25 ml portions of DI water. Add a couple of carborundum boiling stones. 7. Place distillation flask in mantle. Stabilize with clamp on mantle rod. 8. Place Kjeldahl trap in flask neck. Stabilize with clamp on mantle rod. 9. Connect bridge adapter to top of trap. Be sure thermometer port is plugged with glass stopper. 10. Rinse 100 ml volumetric flask with DI water and place in 500 ml beaker on counter top.

2 Page Place straight adapter on end of 400 mm condenser and while holding it in place lower its end into volumetric flask. Place condenser into two clamps on Cole Parmer ringstand without tightening. Simultaneously lower jack and position receiver, ringstand and bridge adapter angle so that bridge adapter mates with top of condenser. Adjust postitions so that all glass joints are tight and so that there is a little clearance between the straight adapter and the mouth of the volumetric flask. Tighten the clamps. 12. Make and check the cooling water connections. 13. Surround the receiver with ice cubes in a little water, turn on the cooling water and set the mantle control to 10 for 10 minutes, then lower heat setting to 6. There should be some reflux from the neck of the flask during this time.

3 Page Recheck tighness of all ground glass joints. Figure 2.1 Arrangement of distillation apparatus

4 Page Boiling should commence within a minute or so and reflux from the walls of the Kjeldahl adapter should start. If all goes well, the alcohol in this reflux should collapse the foam. If it doesn t, well that s what the trap is there for. 16. After 20 minutes (total) foaming should be under control and alcohol should be seen dropping from condenser tip into straight adapter. Advance heat to 7. After stabilizing, condensate should appear at about drops per minute. 17. After 1 hour (total) raise heat to 8. Check tightness of glass joints. 18. During distillation tare pycnometer, if pycnometer is being used, and obtain weight measurement with DI water at 20 C 19. Collect just under 100 ml of distillate. This should take about 2 and three quarters hours. 20. Place volumetric flask in water bath. After attemperation, make to mark with DI water. Mix throroughly. Measure and record density, specific gravity and OIML ABV. Find ASBC ABV from ASBC tables. Calculate ABW from grams/100 ml in ASBC tables. 21. Transfer residue from distillation flask to same volumetric flask. Rinse distillation flask with 25 ml portions of DI water. Place in water bath and make to mark with DI water. Measure and record density, specific gravity and P. 22. Measure and record specific gravity, density and P of beer. 23. Do the sums! 3.0 Accuracy There are five factors which effect the acheivable accuracy Measurement Conversion Recovery The standard Interferences The alcohol content in the distillate is estimated by measuring the distillate specific gravity (or density) and then converting this reading to an alcohol concentration by entering a table. Error in measurement of specific gravity thus translates into an error in the reported alcohol content we take out of the table. The table itself has some errors in it as it was at some time based on a set of measurements. It is clear that the ASBC table, at least, contains quantizing errors.

5 Page 5 Recovery refers to the fact that in this method we try to remove the alcohol from the beer sample in order to measure it. We cannot recover all the alcohol in the distillation receiver. That which is not recovered is not counted. If we can obtain an estimate of the fraction of the alcohol not recovered we can adjust the amount determined by specific gravity measurement in order to account for it. Our estimate of the recovery is subject to error and depends upon our knowledge of the purity of the standard used in making the estimate. Intereference refers to other substances in the beer which interfere with our ability to measure ethanol. Substances more volatile than alcohol and substances less volatile than alcohol but more volatile than water will be carried over into the distillate in some measure. We assume, for this work, that the amounts of such volatile substances are small enough that they need not be considered. DeClerk focuses on acetic acid noting that it is present in quantities insufficient to make much of a difference except in the case of lambics. He notes (Vol. II p429) that raising the ph of the beer with alkali dissociates the acid and prevents its being volatilized but that this leads to frothing and inability to determine the true extract. We will say no more about interference. 3.1 Measurement Two types of measurements are made when beer alcohol is determined using this methodology. The first tyhpe of measurement is of the volumes of beer and distillate. The second is in the specific gravity of the distillate. 3.2 Volume Measurement Volumes are measured with the same Class A 100 ml volumetric flask specified as 100 ± 0.08 ml. If this value (0.08) is interpreted as being the standard deviation of the random error in filling the flask it is clear that the alcohol determined should be multiplied by the factor f ε = d ε b where ε b represents the error in filling the flask with beer (to include the error in transfer to the distilling flask) and ε d the error in topping off the distillate with deionized water. Though common sense dictates that the standard devitation associated with ε b should be larger than that associated with ε d because of the transfer of beer to the distilling flask we have no feel for how much larger and thus, to keep the analysis simple, make no adjustment for this. Expanding Equation (3.1) and throwing away second order terms gives (3.1) f = 1 ε b 100 ε d 100 (3.2) 2σ from which we see that the expected value of f is one and the standard deviation in f is b where σ is the standard deviation associated with filling the flask (0.08 ml) and thus the standard deviation in f is Thus, at the 100 b 10% level the error budgeted to this source would be %.

6 Page Specific Gravity Measurement To determine alcohol content we measure the density of distillate from the beer sample which distillate is assumed to contain all the alcohol in the sample and nothing else except water. We use an Anton Paar DMA 5000 Density Meter calibrated against dry air and deionized, degassed water. The specified repeatability for this instrument is and its accuracy is specified as grams/cm 3. The error in specific gravity measurement is seen through its conversion to an error in alcohol content (see section which follows). Figure 3.2 shows the sensitivity (partial differential of) ABV to specific gravity taken from the ASBC (based on table Association of Official Agricultural Chemists) tables. For example, at an ABV of 5% Figure 3.2 shows a sensitivity of -740% ABV for each unit of error in specific gravity. The repeatability of alcohol concentration estimates for solutions near 5% is thus % and the overal accuracy % 1. At 9.5% ABV the uncertainty in alcohol content from a specific gravity uncertainty of is % and it is this value that we will budget for the error contribution from density measurement. 3.4 Conversion The measured specific gravity is enterend into the ASBC table (based on table Association of Official Agricultural Chemists) and the alcohol content taken out.the data from this table is plotted in Figure 3.1. In fact we use an Anton Paar DMA5000 density meter which measures the density of the alcohol solution and looks up the ABV value using tables prepared by the OIML (International Organisation of Legal Metrology) rather than the ASBC tables 2. The two tables are presumed close. We checked at specific gravity According to the ASBC tables the ABV for solutions at this specific gravity is %. Examination of ABV s determined by the DMA 5000 indicate that the OIML value for this same specific gravity is 9.984% or 0.016% lower than the ASBC table s reading. Chi squared in fitting the ASBC data is given on Figure 3.1 as for 501 points corresponding to an rmse of %. The OIML tables are not accesible. We assume that they are about the same and we thus budget % for conversion error. 1. Given that the instrument deviations are interpreted as standard deviations then these error values are standard deviations as well. The values should be increased sligtly (1/ i.e. about 0.2% - i.e. percent of a percent) because the instrument spec values are for density, not specific gravvity. 2. We note that the DMA 5000 does contain the AOAC tables but as the instrument was programmed for the OIML method as it came from the factory we have stayed with it. Note further that the OIML tables are based on ITS90 and, therefor, presumably more accurate than the AOAC tables which were published originally in 1930.

7 Page x ABV, % 5 4 ABV vs. Specific Gravity Fit to ASBC Tables W_coef={1.7417e+05, e+05, e+05, e+05} V_chisq= ; V_npnts= 501; W_sigma={5.06e+03,1.53e+04, 1.54e+04,5.17e+03} Alcohol: ABV vs SG Specific gravity (20 C/20 C) Figure 3.1 Polynomial Fit to ASBC (AOAC) Alcohol Table Recovery To check on recovery we made up 1 L of an approximately 10% v/v solution of EtOH and measured the alcohol content of 7 aliquots using the method described here. Class A glassware was used to measure out 100 ml of ethanol and DI water was added to make up to 1000 ml. Alcohol and water measurements were made with volumetric flasks in a constant temperature bath at 20 C. The 100 ml volumetric flask used to measure the alcohol is rated ±0.08 ml and the 1 litre volumetric flask used to make up with water is rated ±0.3 ml. We interpret these numbers as standard deviations. The strength of the solution by volume is, A ( p + ε p )( V a + ε a ) = V s where p is the purity of the ethanol standard (fraction close to 1), ε p is the uncertainty in the purity, V a is the vol- ε s (3.3)

8 Page 8 ABV/ Specific Gravity, % ABV Error From S.G. Uncertainty To Estimate ABV Error Multiply Value Read From Graph Against Alcohol Content By SEM in SG Reading Alcohol:dabvdsg vs ABV ABV, % Figure 3.2 ABV Error Sensitivity to Error in Density vs ABV ume (100 ml in this case ) of alcohol transferred quantitatively to the V s (1 L) volumetric flask and ε a and ε s represent the standard errors in the volumes of the two volumetric flasks. Rearranging Equation (3.3) A pv a + pε a + ε p V a + ε p ε a ε s pv a + pε s ε a + ε s ε p V a + ε p ε a ε s = V s V 2 s Separating the actual strength from the errors we have (3.4) pv A = a V s pε a + V s ε p V a V s ε s pv a V 2 s (3.5) where we have dropped all second order and higher terms.in the uncertainties and, thus, assuming that the three

9 Page 9 uncertainties are small and independent we find EA = V a 1000 (3.6) σ A = ( pε a ) 2 + ( ε p V a ) 2 + ( ε s pv a 1000) (3.7) 1000 Both these values are fractional values. If V a is 100 ml then the expected strength of the solution is 10% and the standard error is %. The accuracy of the 100 ml volumetric flask is the major contributor. Were it the only source of error the uncertainty in the concentration would be %. The second largest contributor is the uncertainty in the volume of the 1L volumetric flask. Were this the only error source the uncertainty would be 0.003%. The ethanol used to prepare this solution is not 100% pure. It is guaranteed by the label to have a minimum assay of 99.5%. We assayed it by measuring its density at 20 C. The measurement ensemble is given in the upper left portion of Figure 3.3. Two of the readings in this set appear to contain errors three times that of the rest of the group. These are discarded in the upper right grouping of measuremets which give an ensemble average of is % with an SEM of %. This uncertainty is clearly dominated by the uncertainty from the volume measurements if the two errors are RSS d. Thus we consider a solution prepared from 100 ml of this ethanol to be ± % ethanol. Measured values from the individual distillation runs are compared to % in the spreadsheet and the average readings used to determine the recovery for our lab. Thus the uncertainty in the strength of the standard becomes and uncertainty in every measurement of a sample which is adjusted by the average recovery determined by these calibration runs. At the 10% level this uncertainty would be % and it is that value which is budgeted..

10 Page 10 Alcohol Assay Alcohol Assay ABV Density ABV dev Dens Dev ABV Density ABV dev Dens Dev u u sd sd sem sem Solution (100 ml in 1L): % Replicate Distil. Assay Recovery Err Err^2 Adjby Rec. Err Err^ % % E % E % E % % % u % RMSE: RMSE: sd % cv 0.24% Figure 3.3 OIML Measurements of Alcohol Standard and Recovery from Solution of 100 ml Standard in 1000 ml DI Water Thus we model the recovery as ± 0.23% i.e. it represents a bias plus a random component. As we did in the tests with standard solutions we would multiply all determined values by the factor (the reciprocal of ) to account for the alcohol held up in the distillation apparatus and lost to evaporation. This value is relevent only to our laboratory and our implementation of the procedure. At the same time we add the random part of the determined recovery to our error budget. At an alcohol content of 10% this would be 0.023% 3.6 Summary of Errors The table which follows summarizes the error sources. The RSS of all of them is about 0.03% and that is the estimate of the error level of which we think we are capable in the absence of interferences.

11 Page 11 Table 1: Error Source RMSE Volume Measurement % Specific Gravity Measurement % Tabulation Error % Recovery 0.023% Calibration Standard % RSS % 4.0 References 1.