ANALYST MAY 1986 VOL. 111 507 Determination of Ethanol in Alcoholic Beverages Using a Headspace Procedure and Fuel Cell Sensor* W. J. Criddlet and K. W. Parry Department of Applied Chemistry UWIST P.O. Box 13 Cardiff CF? 3XF UK and 1. P. Jones Lion Laboratories Ltd. Barry South Glamorgan UK A procedure is described that allows the rapid determination of ethanol in a wide variety of alcoholic beverages. Dynamic headspace analysis is employed and a fuel cell sensor is used for the quantitative determination of ethanol. The method is rapid and compares favourably in accuracy with distillation and gas-ch romatograph ic procedures. Keywords Ethanol determination; alcoholic beverages; headspace analysis; fuel cell sensor The approved method of determining the ethanol content of a beverage in the UK and most other countries dates back to the early 1900s and is based essentially on studies made by Thorpe and Brown.1 Tables of data were established that related the specific gravity of distillates to original gravity and thus the ethanol content of the final product.Although improvements have been made over the years in distillation apparatus and in procedures for measuring specific gravity, the method is essentially unchanged. However certain disadvantages are inherent in the procedure including a substantial skill element on the part of the operator carrying out the analysis and the excessive time required for each determination. Cost can also be a factor when expensive products are being examined. In more recent times gas chromatography has become increasingly used for ethanoI assay2 and in some countries is accepted as a standard procedure.However gas chromato-graphs are relatively expensive and require the usual back-up facilities normally associated with a well equipped laboratory, as does the distillation procedure. Other techniques are available that depend either on density measurements3,4 or headspace analysis (Alcoltrol Ac3 L. H. Engineering Co., Stoke Poges Buckinghamshire) but both suffer from high cost and the need for a laboratory environment. In this paper we describe a prototype instrument and associated procedure that can be used to determine ethanol in alcoholic beverages with an accuracy comparable to those of the above-mentioned methods and that hopefully will provide the basis of a commercial portable and simple to use instrument for ethanol determination.procedure for charging the vessels was as follows. The standard vessel was half filled with water and a standard volume of an aqueous ethanol solution (2.0 cm3 5.0% V/V; 1.0 cm3 10.0% V / c 0.5 cm3 20.0% V/V; 0.25 cm3 40.0% VIV) was added to the vessel using a digital pipette e.g., Boehringer BCL 1000 DG and the solution was made up to the mark with water. All of the standard solution was removed from the pipette tip by injecting the sample below the water level and depressing the pipette plunger three or four times. After the addition of two drops of Dow Corning 1520 silicone antifoaming agent the sample vessel was filled identically using the appropriate volume of beverage.An air supply (ca. 300 cm3 min-1) was generated using a small air pump and passed alternately through the two vessels (isolated using the taps shown in Fig. 2) thus generating a dynamic headspace containing ethanol. The ethanol content was measured as described below. Headspace ethanol analysis The air emerging from the vessel was passed through a short polypropylene T-piece pierced by a microlance hypodermic needle fitted to a standard Lion Laboratories (Barry South Glamorgan) fuel cell sensor head. Sampling of the flowing air - ethanol stream was achieved by first pressing the RESET button and then the READ button. This resulted in a standard volume of ethanol - water vapour being drawn into the fuel cell where an electrical potential was developed.This potential was amplified and used as a measure of the ethanol concentration in the vapour. Experimental Instrument Construction The instrument consists essentially of two main components, viz. the headspace generation element and the electronic ethanol detection system. These two components will be considered separately. Headspace ethanol generation Two specially designed and calibrated (100 cm3 at 20 "C) vessels were constructed incorporating inlets and outlets for liquids air and a thermocouple inlet as shown in Fig. 1. The * Presented at Analyticon 85 London September 17-19th 1985. t To whom correspondence should be addressed. - Air - EtOH outlet - thermocouple inlet Calibration mark Fig. 1. vessel Diagrammatic representation of a headspace generatin 508 7 4 t 4 5 3 ANALYST MAY 1986 VOL.111 1 1 LY Fig. 2. Schematic flow diagram of the ethanol analyser. 1 Regulated air supply; 2 two-way ta 3 headspace generating vessel; 4 isolating taps; 5 polypropylene {Liece; 6 sampling valve fuel detector unit and microlance hypodermic needle; and 7 air - ethanol outlet Data Analysis The signal developed in the fuel cell was analysed using a Lion AE-D3 unit which was specially built for this study and derived from the company's microprocessor-controlled Auto Alcolmeter. In the AE-D3 four range controls are utilised for 5.0 10.0, 20.0 and 40.0 VfV. Depending on the nature of the unknown sample i.e. whether beer wine etc. the appropriate selector button is activated and the signal from the fuel cell obtained using the standard vessel is displayed (LCD).Note that the concentration of ethanol in the standard is always constant, but will be displayed as one of the four values indicated above. After calibration the air stream is passed through the sample vessel the new air - ethanol stream is sampled (after ca. 2 min to allow flushing of the previous sample) and the new signal is displayed on the LCD. As the fuel cell output is directly proportional to the ethanol concentration over a wide range of concentrations it is a simple matter to display the unknown ethanol concentration. Temperature Correction The temperatures of both sample and standard were moni-tored continuously using an RS Components Type 610-067 digital thermometer fitted with a Type K thermocouple and any difference in temperature was compensated for mathe-matically (see Discussion).In order to establish the relationship between the tempera-tures of the vessels and the instrument readings both vessels were filled with standard ethanol solutions (100 cm3 0.05-0.20% VfV) and the reference vessel was maintained at 25.0 "C in a thermostatically controlled bath while the temperature of the sample vessel was varied in the approxi-mate range 10-35 "C. The AE-D3 was calibrated in the usual way using the reference vessel and the apparent ethanol content of the sample vessel was recorded using the 5.00% Vf V range. Determination of Ethanol in Various Beers Wines and Spirits The ethanol contents of a wide range of alcoholic beverages were determined using the procedure described above with the appropriate volume of the various beverage types viz., beers (2.00 cm3) table wines (1.00 cm3) fortified wines (0.50 cm3) and spirits (0.25 cm3).Table 1. Determination of the slope (m) of the linear logarithmic plot of temperature versus displayed ethanol concentration Ethanol concentration Correlation Yo VIV Slope coefficient 0.05 0.08 0.08 0.10 0.10 0.10 0.10 0.13 0.15 0.20 0.0291 0.0285 0.0286 0.0272 0.0290 0.0279 0.0265 0.0275 0.0279 0.0290 0.998 0.997 0.999 0.998 0.997 0.999 0.999 0.999 0.999 0.998 Statistical analysis: n = 1O;X = 0.0281; S.D. = 8.27 x C.V. = 2.94% Standard / Unknown Tu Ts Temperaturei'C -Fig.3. Graphical representation of the mathematical involved in temperature compensation Irocess Gas-chromatographic Analysis of Reference Samples The ethanol contents of a wide variety of samples were determined using a Perkin-Elmer 8310 gas chromatograph fitted with a 3 m X 3 mm i.d. stainless-steel column contain-ing 10% Carbowax 20M on 60-80 mesh Celite and a flame-ionisation detector. The column was operated isother-mally at 80 "C using helium as the carrier gas (15 cm3 min-l) and propan-1-01 as an internal standard. Analysis of Reference Samples by Distillation The ethanol contents of all reference samples were also determined by the distillation procedure approved by HM Customs and Excise i . e . using gravimetric density rneasure-ment.5 Results and Discussion The technique of headspace analysis particularly in the context of alcoholic beverages has one major advantage over most other analytical procedures viz.that the number of possible interfering substances in the headspace is greatly reduced i.e. many substances present in beverages are non-volatile and would not contribute to the headspace composition. However the procedure does have two inheren ANALYST MAY 1986 VOL. 111 509 Table 2. Determination of ethanol in various alcoholic beverages Ethanol content,% VIV Classification Type Beers. . . . . . . . . . . Light ale Bitter ale Lager Chinese beer Claret Cream sherry Amontillado sherry White port Ruby port Spirits . . . . . . . . . . Whisky Rum Vodka Vodka Table wines .. . . . . Moselle Fortifiedwines . . . . . . Finosherry Liqueurs . . . . . . . . Irish whiskey cream liqueur Coconut liqueur Headspace analysis Gas chromatography 3.10 3.11 2.90 2.91 4.18 4.16 4.85 4.84 8.5 8.5* 12.2 12.2* 17.6 17.7* 17.7 17.7* 17.8 17.8* 20.2 20.3* 20.2 20.3* 40.6 40.0 40.5 40.0 38.0 37.5 46.4 45.8 17.0 16.9 27.7 27.8 Distillation 3.12 2.91 4.17 4.94 8.4 12.1 17.6 17.8 17.7 20.5 20.2 40.lt 40.1 37.4 45.6 16.91. 27.91. * Independently determined by J. Harvey & Sons Ltd. -t Independently determined by International Distillers and Vintners Ltd. Table 3. On-site results of ethanol determination in 1984 port wine vintage compared with laboratory distillation values. Samples 1-30 are red and 31-36 are white port wine Ethanol concentration,% V/V Sample No.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Distillation 20.0 19.8 20.3 20.2 20.0 19.6 18.8 18.9 17.5 17.3 18.3 19.3 19.9 18.5 19.6 20.0 17.8 19.6 16.0 16.8 19.7 19.7 19.1 19.8 18.7 19.7 20.1 20.1 19.8 18.2 17.7 16.1 16.3 17.6 19.3 19.3 * Sample diluted using river water. Analyser 20.0 19.8 20.2 20.4 20.1 19.6 18.9 19.0 17.9* 17.2* 18.9* 19.6 20.1 18.3 19.9 20.0 18.1 19.6 16.0 16.5 19.7 19.6 19.0 20.0 18.6 19.6 20.0 20.0 20.0 17.9 17.3 16.8 16.5 17.3* 19.7* 18.9 disadvantages viz.extreme temperature sensitivity the concentration of ethanol in the vapour phase increasing exponentially with increase in temperature and the effect of dissolved solids on the composition of the vapour phase i.e., the well known “salting-out” effect. Considering the latter problem no great difficulty was experienced in overcoming this. The simple act of dilution, necessary in any event to produce a headspace ethanol concentration suitably low for use with the fuel cell sensor, overcame the problem the “salting-out” effect being un-detectable at the drink dilutions used. Compensation for temperature variation between the stan-dard and sample solutions is however more complex. As indicated above the concentration of ethanol in a headspace, dynamic or static is exponentially dependent on temperature, the slope of the linear logarithmic plots being independent of concentration in the range 0.05-0.28% V/V.To avoid the necessity for thermostating the two vessels it is necessary to compensate mathematically for any variation. Interestingly, however it is not necessary to know the actual temperatures of the liquids but only the temperature difference between them. The following mathematical treatment illustrates this point. Table 1 shows the slopes of such plots the mean slope of these lines being 0.0281. A graphical representation of the process involved with temperature correction is given in Fig. 3 which shows logarithmic plots for the variation of head-space ethanol (as measured on the AE-D3) for standard and unknown samples.If we let S be the AE-D3 response to a standard at a temperature T (note that S is therefore the value of the range selected) R be the AE-D3 response to an unknown sample at a temperature T and U be the AE-D3 response to the unknown solution at the temperature of the standard solution T, i.e. the required corrected response value then from Fig. 3 and L o g S = m T + C1 . . . . . . (1) LogR = mT + C2 . . . . . . (2) where rn is the slope of the logarithmic plots and C1 and C2 are the intercepts for the standard and unknown lines respec-tively. From equation (2), C2 = logR - mT . . . . . .(3) and Log U = mT + C2 . . . . . . (4) i . e . , Log U = mT + log R - mT, = m(T - T,) + log R .. ( 5 510 ANALYST MAY 1986 VOL. 111 It follows that only the difference in temperature is involved in correction log R and m being determined experimentally. Results obtained using this correction were excellent and data obtained for actual determinations of ethanol (under laboratory conditions) in a wide variety of drinks are shown in Table 2. The agreement between standard ( i . e . gas-chromatographic and distillation) procedures and the head-space procedure described here shows clearly that the method is applicable to most beverages and has a precision of the same order as those of gas chromatography and distillation. The analysis times are considerably shorter if times are measured from the initial preparation of samples and standards. It is worth noting that data obtained for the analysis of high-ethanol beverages (spirits) are less satisfactory in absol-ute terms than for the other beverages studied.We attribute this to difficulties in reproducibly dispensing the small (0.25 cm3) sample volume associated with the analysis of spirits. However for general screening purposes the accuracy achieved coupled with the rapidity of the determination may make the procedure of some use to analysts in the spirits industry. If a precision similar to that obtainable by gas chromatography or distillation is required we do not recom-mend the headspace procedure as described in this paper. Extensive field trials of the determination of ethanol under non-laboratory conditions were undertaken during the port wine vintage of 1984 where a large number of wines were analysed on site in Portugal under the adverse conditions that operate in most port wine-producing farms.Water taken directly from the Douro River was sometimes used for dilution with no adverse affects as the results in Table 3 show. Under these conditions power was provided by the integral sealed, re-chargeable lead - acid battery in the AE-D3 unit which provided power for up to 8 h of continuous use. We would particularly recommend the use of digital pipettes for field use as their plastic construction makes them extremely robust. Turning to the fuel cell and its role as the sensor in this analytical study its mode of operation is such that a two-stage oxidation of ethanol occurs giving ultimately acetic acid as the final product.6 It is apparent therefore that compounds other than ethanol may be involved in similar oxidation reactions, and thus interfere quantitatively with data obtained.This is clearly true for most alcoholic beverages which contain a wide variety of volatile oxidisable organic species notably acetal-dehyde methanol and primary and secondary alcohols up to Cg. However the occurrence of these substances is such that in total they are not likely to exceed 500 mg 1-1 in beer wine or spirits and can therefore be discounted as interfering at a measurable level. The same argument is used to justify the distillation procedure where volatile components would affect the final distillate density and we would concur with this line of thought.It is worth noting the modification to the normal use of digital and similar pipettes. Standard usage was found to give erratic delivery of standard volumes depending on the nature of the beverage used. Certain of the more viscous liquids ( port wines liqueurs etc.) often left small globules in the disposable pipette tips thus affecting the precisely delivered volume. To overcome this problem the total contents of the tip were removed as described under Experimental. Although the precisely displayed volume may not therefore be delivered, reproducible volumes certainly were. Bearing in mind that a comparison procedure is involved in the analytical method described here approximate but reproducible volumes are the basic requirement not absolutely precise volumes as is usually the case.The use of an antifoaming agent is recommended for eliminating the carry-over of liquid from the calibrated vessel to the sampling chamber. Some alcoholic beverages particu-larly beer give solutions that show severe frothing and if no antifoaming agent is added bubbles form in the neck of the calibrated vessels and are carried to the sampling chamber. This clearly must be avoided and the recommended anti-foaming agent is extremely efficient in eliminating this effect. Whilst considering the possible carry-over of liquid from the vessels to the sampling chamber it should be noted that the bubbles of air passed through the solution are coarse and sinters are not used to produce fine bubbles as might be expected in an analytical procedure of this type.Sinters were tested in the first instance but it was found that a fine almost invisible spray was produced that was carried to the sampling chamber by the air flow. This resulted in the solution finding its way into the fuel cell with the obvious disastrous results. Removal of the sinter eliminated this effect and it should be noted that bubbles of large dimensions equilibrate very rapidly with a water - ethanol system.7 It would appear that the headspace procedure compares in precision with both gas-chromatographic and distillation methods for ethanol determinations. It should be pointed out that the determination of ethanol by either of the above established procedures requires a high degree of practical analytical expertise. In addition neither procedure can be conveniently operated outside a laboratory environment.We feel therefore that the procedure described in this paper has much to commend it in terms of simplicity of operation i.e. a relatively low skill requirement and its ability to be operated virtually anywhere and with very short warm-up periods. On these grounds alone the procedure scores well against the established procedures. In conclusion the prototype instrument described in this paper offers a rapid inexpensive (below &3000) and portable means for the determination of ethanol in most alcoholic beverages. It is hoped that it will in due course and with appropriate packaging and automation appear as a commer-cially available instrument for the alcoholic beverage indus-tries. The authors thank Mr. R. W. Goswell of John Harvey and Sons Ltd. and Dr. David Clutton of International Distillers and Vintners Ltd. for samples and analytical data and Dr. H. P. Reader of Cockburn Smithes and Sons Oporto for test facilities in Portugal. 1. 2. 3. 4. 5 . 6. 7. References Thorpe E. and Brown H. T. J . Znst. Brew. 1914 20 569. James A. T. and Martin A. J. P. Biochem. J . 1952,50,679. Stabinger H. Leopold H. and Kratky O. MLF Chem., 1967 98 436. Leopold H. Electronik 1970 19 297. “Laboratory Alcohol Table RDC 80/267/04,” HM Customs and Excise London 1979. Williams P. M. PhD Thesis University of Wales 1978. Jones A. W. PhD Thesis University of Wales 1974. Paper A51305 Received August 27th 1985 Accepted September 24th I98