On-line Flow Injection–Pervaporation of Beer Samples for the Determination of Diacetyl Jos�e M. Izquierdo-Ferrero, Juan M. Fern�andez-Romero and Mar�ýa D. Luque de Castro* Department of Analytical Chemistry, Faculty of Sciences, University of C�ordoba, E-14004 C�ordoba, Spain. E-mail: qa1lucam@uco.es A method for the determination of diacetyl based on the integration of a continuous pervaporation module and a flow injection manifold is presented. The chemical derivatization reaction involes condensation between diacetyl, a-naphthol and creatine in a basic medium to form a coloured compound, which is monitored photometrically at 530 nm.The method affords low detection and quantification limits (5 and 10 ng ml21 of diacetyl calculated as ±3s and ±10s, respectively, for n = 20), a wide linear range (10–2000 ng ml21, r2 = 0.9998, n = 7), good precision (RSD < 3%) and a sampling frequency of 8 h21. The method was applied to the determination of diacetyl in beer from bottled and medium-process samples and the results were in excellent agreement with those given by the conventional method.Keywords: Flow injection; continuous pervaporation; spectrophotometry; diacetyl; beer Diacetyl is an a-diketone responsible for the flavour of several foodstuffs (cheeses, yogurts, beers and wines) and contributes adversely to the taste quality. Diacetyl has been used for both testing food ripeness (e.g., in cheeses and wines) and monitoring of manufacture processes (e.g., in beers).1 Low levels of diacetyl and pentane-2,3-dione are desirable in beer fermentation, as these compounds confer an unpleasant taste similar to butter in the final product.This means that routine control is necessary during the fermentation of ethanol and other compounds in beer production. Beer with an acceptable taste requires the formation of esters and alcohols such as ethyl acetate, isoamyl acetate and isoamyl alcohol and the removal of undesirable compounds (such as a-diketones).Recently, research in this field has been focused on the development of new microorganisms which increase the metabolization of undesirable products during fermentation.2 The low concentration levels of diacetyl in beverages and the presence of potential interferents makes the development of analytical methods with higher sensitivity and selectivity essential. Previous methods for the determination of these species include a separation step, such as distillation, prior to the determination.Usually these methods have been implemented by using optical detection, such as photometry after condensation3,4 or chelating reactions5,6 and fluorimetry.7,8 The individual determination of diacetyl and pentane-2,3-dione has also been proposed using GC–FID,2,9 GC–ECD2,10,11 GC– MS12 or headspace GC.13 Several methods based on HPLC with continuous fluorimetric detection have also been reported.14,16 Recent developments in continuous flow analysis have led to the use of non-chromatographic continuous separation techniques for the determination of one or more analytes, the separation of components into groups and interference removal.Non-chromatographic continuous separation techniques can improve the sensitivity of the method through preconcentration and its selectivity by deleting matrix effects or particular interferents. Membrane-based non-chromatographic separation techniques are particularly useful for increasing selectivity.In some cases, they also allow analyte preconcentration, thereby resulting in improved sensitivity. Continuous gas-diffusion techniques, whether isolated or coupled to other separation techniques such as dialysis or ion exchange, make excellent tools for achieving high selectivity and sensitivity. However, these techniques suffer from serious drawbacks (e.g., clogging of the membrane pores by suspended particles or high molecular mass compounds occasionally present in the sample and deterioration of the membrane by contact with the sample).Both drawbacks can be overcome by using pervaporation.17 This technique has been employed for a long time in industry, in competition with other traditional processes (e.g., distillation, extraction and adsorption); by contrast, it has scarcely been used for analytical purposes. A laboratory-scale pervaporation unit was developed a few years ago in order to overcome problems arising from the use of biosensors in on-line fermentation monitoring.18,19 Several applications of laboratory pervaporation based on different approaches have been developed recently by Luque de Castro and co-workers.20–24 This paper reports the use of an automatic spectrophotometric method for the determination of diacetyl in beer samples based on the on-line coupling of a continuous pervaporation unit to a simple flow injection (FI) manifold.The chemical reaction is based on condensation of the analyte using an excess of anaphthol and creatine in a basic medium as described by Mattessich and Cooper,4 who improved a previous assay for the determination of this analyte.The condensation product is monitored spectrophotometrically at 530 nm. Experimental Instruments and Apparatus A Pye Unicam (Cambridge, UK) SP-500 spectrophotometer furnished with a Hellma (Jamaica, NY, USA) 178.011 QS flow cell and equipped with a Knauer (Bad Homburg, Germany) recorder was used.A laboratory-built pervaporation module equipped with a thermostated magnetic stirrer was also used. Two Gilson (Worthington, OH, USA) Minipuls-3 four-channel peristaltic pumps with rate selector, two Rheodyne (Cotati, CA, USA) Model 5041 injection valves and PTFE tubing of 0.5 mm id were also used. The pervaporation cell, designed in the laboratory was similar to others described by Mattos et al.20 Reagents and Solutions All reagents were of analytical-reagent grade.The donor stream was an aqueous solution with which the sample solution was mixed. A solution containing 125 mmol l21 a-naphthol (Sigma, St. Louis, MO, USA) and 100 mmol l21 creatine (Sigma) prepared in 750 mmol l21 sodium hydroxide (Merck, Darmstradt, Germany) was used as a reagent/acceptor solution. Aqueous diacetyl (Merck) solution was used as standard after suitable dilution. Beer samples provided by Compa�n�ýa Andaluza de Cervezas (C�ordoba, Spain) were used for testing the method.All solutions were prepared in doubly distilled water of Analyst, February 1997, Vol. 122 (119–122) 119high purity obtained from a Millipore (Bedford, MA, USA) Milli-Q Plus system. PTFE membranes of 5.0 mm pore size and 47 mm diameter purchased from Millipore were also used. Manual Procedure A 100 ml volume of beer is subjected to distillation to obtain 20 ml of distillate, of which 10 ml are mixed with 0.5 ml of 1 mg ml21 o-phenylenediamine and kept in the dark for 25 min.Then 2 ml of 4 m HCl are added and the absorbance of the mixture is monitored at 335 nm within 30 min after mixing. A blank containing 10 ml of deionized water instead of distillate is also measured photometrically. Manifold and Procedure Fig. 1 depicts the hydrodynamic system used. It consists of a flow injection manifold divided into two parts. The lower part acts as donor submanifold, which consists of a monochannel system provided with a main injection valve (MIV) that injects the samples into an aqueous stream which circulates through the lower part of the pervaporation cell, which is thermostated at 90 °C.The diacetyl evaporates to the air gap between the donor solution and the membrane and diffuses through the hydrophobic membrane to the upper part of the pervaporation cell where it is collected. The upper submanifold includes the acceptor chamber of the pervaporation cell, which is located in the loop of an auxiliary injection valve (AIV).This valve is filled with the acceptor solution, which contains the reagent mixture (AIV in injection position) and then is switched to the filling position, thus keeping static the loop contents during a preset interval. The analyte which passes through the membrane is collected in the acceptor/reagent solution, with simultaneous development of the condensation reaction. After a presetime (Tp), the AIV is switched again to the inject position and the reaction product is driven to the detector.The peak height (transient signal obtained when the plug containing the coloured product reaches the detector) is proportional to the concentration of diacetyl in the sample. Results and Discussion Optimization of Variables The optimization of the variables involved in the overall process, grouped into physical, chemical and hydrodynamic, was performed using the univariate method. Table 1 shows the ranges over which each variable was studied and the optimum value found.Physical variables Fig. 2(a) shows the influence of temperature in the pervaporation step, which was studied between 30 and 110 °C (boiling point of diacetyl 88 °C). A temperature of 90 °C was sufficient for appropriate transfer of the analyte through the membrane. This temperature was also suitable for the formation of the condensation product. Fig. 1 Manifold for the automatic–photometric determination of diacetyl in beer samples.P, Peristaltic pump; MIV, main injection valve; IV, injection volume; AIV, auxiliary injection valve; PM, pervaporation module; m, hydrophobic membrane; TMS, thermostatic magnetic stirrer; D, detector; w, waste; R, acceptor/reagent solution; C, donor solution; and S, sample. Table 1 Optimization of variables Range Optimum Type Variable studied value Physical Temperature/°C 30–110 90 FI Donor flow rate/ml min21 0.1–0.5 0.3 Acceptor flow rate/ml min21 0.5–2.0 1.8 Volume injected/ml 100–3000 2000 Pervaporation time/s 30–480 360 Chemical Sodium hydroxide concentration/mmol l21 100–1000 750 pH — 12 a-Naphthol concentration/mmol l21 5–200 125 Creatine concentration/mmol l21 5–150 100 Fig. 2 Influence of temperature (a) on the pervaporation process and (b) on the derivatization reaction for concentrations of diacetyl of 20 and 100 ng ml21 (dashed and continuous lines, respectively). 120 Analyst, February 1997, Vol. 122Chemical variables The influence of the pH in both the donor and acceptor solutions was studied.An aqueous solution adjusted to different pH values was used as a donor stream. A higher pervaporation efficiency was achieved by using doubly distilled water as the donor stream. However, the pervaporation efficiency was increased when the acceptor solution contained sodium hydroxide. This was tested by preparing different aqueous solutions containing sodium hydroxide at concentrations between 100 and 1000 mmol l21.A concentration of 750 mmol l21 provided the best medium for collection of the analyte. This solution was also appropriate for the development of the derivatization reaction. Concentrations of 125 mmol l21 a-naphthol and 100 mmol l21 creatine in the acceptor/reagent solution were selected as optimum. Higher concentrations of these reagents caused higher blank signals, resulting in appreciable lack of sensitivity (estimated as 50%). On the other hand, lower concentrations resulted in poor development of the chromogenic reaction.Hydrodynamic variables The flow rate had a dramatic influence on the performance of the system. As the flow rate of the donor solution determined the time during which the analyte was in the pervaporation module, 0.3 ml min21 was selected as optimum for the donor stream (lower values yielded non-reproducible results). On the other hand, the flow rate had no influence on the upper FI subsystem as the collection process occurred under static conditions.When the flow was re-started the reaction plug was driven to the detector. A flow rate of 1.8 ml min21 was appropriate to rinse the flow manifold. In order to obtain the best analytical signal with an acceptable sampling frequency, different stop-times in the range between 30 and 480 s were tried. A pervaporation time of 360 s was chosen as a compromise between the best signal and an acceptable sampling frequency. Injection volumes over 2 ml did not increase the analytical signal significantly. Features of the Method Ten standard solutions of diacetyl with concentrations between 1 and 5000 ng ml21 were prepared and injected in triplicate into the FI–pervaporation assembly using the optimum values of variables found previously.Table 2 summarizes the features of the method (equation, regression coefficient, linear range, detection and quantification limits and RSD). The linear range achieved for the method was suitable for applying it to the determination of diacetyl in beer samples.The sampling frequency achieved under the optimum working conditions was 8 h21. Study of Interferents The study of potential interferents was aimed at those commonly present in beer samples which are structurally similar to diacetyl, such as pyruvic acid, butylene glycol, acetoin (a precursor of diacetyl) and ascorbic acid. All interferents were added to the sample at concentrations higher than usually found in beers.The results showed that pyruvic acid, butylene glycol and ascorbic acid are tolerated at levels up to ten times their content in beer (with a tolerated interferent to analyte ratio of 100 : 1 for pyruvic acid and butylene glycol and 10 : 1 for ascorbic acid). Acetoin interferred at five times its concentration in these samples; nevertheless, acetoin is easily oxidized to diacetyl under atmospheric conditions. Application to the Proposed Method The method was applied to the determination of diacetyl in bottled and medium-process beer.In order to eliminate the gas contained in the original beverage (mainly CO2), all the samples were previously ultrasonicated for 5 min, then injected directly in triplicate. Table 3 summarizes the concentrations found and the results obtained after addition of two amounts of the analyte (50 and 100 ng ml21) to aliquots of the samples. The results provided by the proposed method were compared with those obtained by the conventional method used by Compa�n�ýa Andaluza de Cervezas. This method is based on previous separation of the analyte by distillation (vapour stream) followed by condensation of the analyte with ophenylenediamine, and photometric monitoring of the product at 335 nm.Six beer samples were used for this comparison and the results obtained produced a straight-line graph: y = 0.934x + 1.477 (r2 = 0.9901, n = 6) where y represents the FI–pervaporation method and x the conventional method.As can be seen, an excellent correlation exists between the proposed and the conventional methods. Conclusions The method proposed for the determination of diacetyl in beer samples shows the following features: simplicity in implementing the experimental set-up (i.e., a conventional FI manifold in Table 2 Features of the method Equation* A = 0.0199C + 0.0005 Regression coefficient 0.9998 (n = 7) Linear range/ng ml21 10–2000 Detection limit/ng ml21 5 Quantification limit/ng ml21 10 RSD (%)† — Low level 3.0 High level 2.6 * A denotes absorbance and C analyte concentration in ng ml21.† For 50 and 1000 ng ml21 of diacetyl. Table 3 Application of the method Concentration/ng ml21 Recovery (%)† Sample Conventional Proposed Difference No. Type* method method (%) Addition 1 Addition 2 1 A 18 17 25.5 100 96 2 A 15 16 6.5 94 98 3 A 25 24 24.0 98 97 4 B 30 29 23.3 100 99 5 B 25 27 8.0 107 99 6 B 52 50 23.8 100 100 * A and B denote bottled and medium-process beers, respectively.† 50 and 100 ng ml21 for additions 1 and 2, respectively. Analyst, February 1997, Vol. 122 121which an easy laboratory-built pervaporation unit is included); high selectivity as a result of the nature of the chemical reaction; lower sample consumption (2 ml); and an acceptable sensitivity. These features make it suitable for the routine determination of diacetyl in beer samples, as was demonstrated by comparison with the conventional method. The Comisi�on Interministerial de Ciencia y Tecnolog�ýa (CICYT) is thanked for financial support (project No.PB93- 0827). We gratefully acknowledge Carlos M. G�omez S�anchez of Compa�n�ýa Andaluza de Cervezas for the samples compared with the conventional method. References 1 West, h, A. L., and Becker, K., Am. Soc. Brew. Proc., 1952, 81, 65. 2 Mathis, C., Pons, M. N., Engasser, J. M., and Lenoel, M., Anal. Chim. Acta, 1993, 279, 59. 3 Westerfeld, W. W., J. Biol. Chem., 1945, 161, 495. 4 Mattessich, J., and Cooper, J. R., Anal. Biochem., 1989, 180, 349. 5 Ribereau-Gayon, J., Peynaud, E., Sudraud, P., and Riberau-Gayon, P., Sciences et Techniques de Vin, Dunod, Paris, 1972. 6 Walsh, B., and Cogan, T. M., J. Dairy Res., 1988, 41, 31. 7 Garc�ýa-Vilanova, R. J., and Garc�ýa Estepa, R. M., Talanta, 1993, 40, 1419. 8 Mariaud, M., and Levillain, P., Talanta, 1994, 41, 75. 10 Barbieri, G., Bolzoni, L., Careri, M., Mangia, A., Paroladi, G., Spagnoli, S., and Virgili, R., J.Agric. Food. Chem., 1994, 42, 1170. 11 Otsuka, M., and Ohmori, S. J., J. Chromatogr. B, 1992, 115, 215. 12 Otsuka, M., and Ohmori, S. J., J. Chromatogr. B, 1994, 654, 1. 13 Damiani, P., and Burini, G., J. Assoc. Off. Anal. Chem., 1988, 71, 462. 14 Yamaguisgi, M., Ishida, J., Zhu, X., Nakamura, M., and Yoshitake, T., J. Liq. Chromatogr., 1994, 17, 203. 15 Ulberth, F., J. Assoc. Off. Anal. Chem., 1991, 74, 630. 16 Gilson, T.D., Parker, S. M., and Woodward, J. R., Enzyme Microb. Technol., 1991, 13, 171. 17 Luque de Castro, M. D., and Papaefstathiou, I., in Encyclopedia of Environmental Analysis and Remediation, ed. R. A. Meyers, Wiley, New York, in the press. 18 Prinzing, U., Ogbomo, I., Lehn, C., and Schmidt, H. L., Sens. Actuators B, 1990, 1, 542. 19 Ogbomo, I., Steffl, A., Schumann, W., Prinzing, U., and Schmidt, H. L., J. Biotechnol., 1993, 31, 317. 20 Mattos, I. L., Luque de Castro, M.D., and Valc�arcel, M., Talanta, 1995, 42, 755. 21 Papaefstathiou, I., Tena, M. T., and Luque de Castro, M. D., Anal. Chim. Acta, 1995, 308, 246. 22 Papaefstathiou, I., Luque de Castro, M. D., and Valc�arcel, M., Fresenius’ J. Anal. Chem., 1996, 354, 442. 23 Papaefstathiou, I., and Luque de Castro, M. D., Anal. Lett., 1995, 28, 2063. 24 Papaefstathiou, I., and Luque de Castro, M. D., Anal. Chem., 1995, 67, 3916. Paper 6/06401I Received September 17, 1996 Accepted October 28, 1996 122 Analyst, February 1997, Vol. 122 On-line Flow Injection–Pervaporation of Beer Samples for the Determination of Diacetyl Jos�e M. Izquierdo-Ferrero, Juan M. Fern�andez-Romero and Mar�ýa D. Luque de Castro* Department of Analytical Chemistry, Faculty of Sciences, University of C�ordoba, E-14004 C�ordoba, Spain. E-mail: qa1lucam@uco.es A method for the determination of diacetyl based on the integration of a continuous pervaporation module and a flow injection manifold is presented.The chemical derivatization reaction involes condensation between diacetyl, a-naphthol and creatine in a basic medium to form a coloured compound, which is monitored photometrically at 530 nm. The method affords low detection and quantification limits (5 and 10 ng ml21 of diacetyl calculated as ±3s and ±10s, respectively, for n = 20), a wide linear range (10–2000 ng ml21, r2 = 0.9998, n = 7), good precision (RSD < 3%) and a sampling frequency of 8 h21. The method was applied to the determination of diacetyl in beer from bottled and medium-process samples and the results were in excellent agreement with those given by the conventional method.Keywords: Flow injection; continuous pervaporation; spectrophotometry; diacetyl; beer Diacetyl is an a-diketone responsible for the flavour of several foodstuffs (cheeses, yogurts, beers and wines) and contributes adversely to the taste quality. Diacetyl has been used for both testing food ripeness (e.g., in cheeses and wines) and monitoring of manufacture processes (e.g., in beers).1 Low levels of diacetyl and pentane-2,3-dione are desirable in beer fermentation, as these compounds confer an unpleasant taste similar to butter in the final product.This means that routine control is necessary during the fermentation of ethanol and other compounds in beer production. Beer with an acceptable taste requires the formation of esters and alcohols such as ethyl acetate, isoamyl acetate and isoamyl alcohol and the removal of undesirable compounds (such as a-diketones).Recently, research in this field has been focused on the development of new microorganisms which increase the metabolization of undesirable products during fermentation.2 The low concentration levels of diacetyl in beverages and the presence of potential interferents makes the development of analytical methods with higher sensitivity and selectivity essential. Previous methods for the determination of these species include a separation step, such as distillation, prior to the determination.Usually these methods have been implemented by using optical detection, such as photometry after condensation3,4 or chelating reactions5,6 and fluorimetry.7,8 The individual determination of diacetyl and pentane-2,3-dione has also been proposed using GC–FID,2,9 GC–ECD2,10,11 GC– MS12 or headspace GC.13 Several methods based on HPLC with continuous fluorimetric detection have also been reported.14,16 Recent developments in continuous flow analysis have led to the use of non-chromatographic continuous separation techniques for the determination of one or more analytes, the separation of components into groups and interference removal. Non-chromatographic continuous separation techniques can improve the sensitivity of the method through preconcentration and its selectivity by deleting matrix effects or particular interferents.Membrane-based non-chromatographic separation techniques are particularly useful for increasing selectivity.In some cases, they also allow analyte preconcentration, thereby resulting in improved sensitivity. Continuous gas-diffusion techniques, whether isolated or coupled to other separation techniques such as dialysis or ion exchange, make excellent tools for achieving high selectivity and sensitivity. However, these techniques suffer from serious drawbacks (e.g., clogging of the membrane pores by suspended particles or high molecular mass compounds occasionally present in the sample and deterioration of the membrane by contact with the sample).Both drawbacks can be overcome by using pervaporation.17 This technique has been employed for a long time in industry, in competition with other traditional processes (e.g., distillation, extraction and adsorption); by contrast, it has scarcely been used for analytical purposes. A laboratory-scale pervaporation unit was developed a few years ago in order to overcome problems arising from the use of biosensors in on-line fermentation monitoring.18,19 Several applications of laboratory pervaporation based on different approaches have been developed recently by Luque de Castro and co-workers.20–24 This paper reports the use of an automatic spectrophotometric method for the determination of diacetyl in beer samples based on the on-line coupling of a continuous pervaporation unit to a simple flow injection (FI) manifold.The chemical reaction is based on condensation of the analyte using an excess of anaphthol and creatine in a basic medium as described by Mattessich and Cooper,4 who improved a previous assay for the determination of this analyte. The condensation product is monitored spectrophotometrically at 530 nm. Experimental Instruments and Apparatus A Pye Unicam (Cambridge, UK) SP-500 spectrophotometer furnished with a Hellma (Jamaica, NY, USA) 178.011 QS flow cell and equipped with a Knauer (Bad Homburg, Germany) recorder was used.A laboratory-built pervaporation module equipped with a thermostated magnetic stirrer was also used. Two Gilson (Worthington, OH, USA) Minipuls-3 four-channel peristaltic pumps with rate selector, two Rheodyne (Cotati, CA, USA) Model 5041 injection valves and PTFE tubing of 0.5 mm id were also used. The pervaporation cell, designed in the laboratory was similar to others described by Mattos et al.20 Reagents and Solutions All reagents were of analytical-reagent grade.The donor stream was an aqueous solution with which the sample solution was mixed. A solution containing 125 mmol l21 a-naphthol (Sigma, St. Louis, MO, USA) and 100 mmol l21 creatine (Sigma) prel21 sodium hydroxide (Merck, Darmstradt, Germany) was used as a reagent/acceptor solution. Aqueous diacetyl (Merck) solution was used as standard after suitable dilution. Beer samples provided by Compa�n�ýa Andaluza de Cervezas (C�ordoba, Spain) were used for testing the method.All solutions were prepared in doubly distilled water of Analyst, February 1997, Vol. 122 (119–122) 119high purity obtained from a Millipore (Bedford, MA, USA) Milli-Q Plus system. PTFE membranes of 5.0 mm pore size and 47 mm diameter purchased from Millipore were also used. Manual Procedure A 100 ml volume of beer is subjected to distillation to obtain 20 ml of distillate, of which 10 ml are mixed with 0.5 ml of 1 mg ml21 o-phenylenediamine and kept in the dark for 25 min.Then 2 ml of 4 m HCl are added and the absorbance of the mixture is monitored at 335 nm within 30 min after mixing. A blank containing 10 ml of deionized water instead of distillate is also measured photometrically. Manifold and Procedure Fig. 1 depicts the hydrodynamic system used. It consists of a flow injection manifold divided into two parts. The lower part acts as donor submanifold, which consists of a monochannel system provided with a main injection valve (MIV) that injects the samples into an aqueous stream which circulates through the lower part of the pervaporation cell, which is thermostated at 90 °C.The diacetyl evaporates to the air gap between the donor solution and the membrane and diffuses through the hydrophobic membrane to the upper part of the pervaporation cell where it is collected. The upper submanifold includes the acceptor chamber of the pervaporation cell, which is located in the loop of an auxiliary injection valve (AIV).This valve is filled with the acceptor solution, which contains the reagent mixture (AIV in injection position) and then is switched to the filling position, thus keeping static the loop contents during a preset interval. The analyte which passes through the membrane is collected in the acceptor/reagent solution, with simultaneous development of the condensation reaction. After a preset pervaporation time (Tp), the AIV is switched again to the inject position and the reaction product is driven to the detector.The peak height (transient signal obtained when the plug containing the coloured product reaches the detector) is proportional to the concentration of diacetyl in the sample. Results and Discussion Optimization of Variables The optimization of the variables involved in the overall process, grouped into physical, chemical and hydrodynamic, was performed using the univariate method.Table 1 shows the ranges over which each variable was studied and the optimum value found. Physical variables Fig. 2(a) shows the influence of temperature in the pervaporation step, which was studied between 30 and 110 °C (boiling point of diacetyl 88 °C). A temperature of 90 °C was sufficient for appropriate transfer of the analyte through the membrane. This temperature was also suitable for the formation of the condensation product. Fig. 1 Manifold for the automatic–photometric determination of diacetyl in beer samples.P, Peristaltic pump; MIV, main injection valve; IV, injection volume; AIV, auxiliary injection valve; PM, pervaporation module; m, hydrophobic membrane; TMS, thermostatic magnetic stirrer; D, detector; w, waste; R, acceptor/reagent solution; C, donor solution; and S, sample. Table 1 Optimization of variables Range Optimum Type Variable studied value Physical Temperature/°C 30–110 90 FI Donor flow rate/ml min21 0.1–0.5 0.3 Acceptor flow rate/ml min21 0.5–2.0 1.8 Volume injected/ml 100–3000 2000 Pervaporation time/s 30–480 360 Chemical Sodium hydroxide concentration/mmol l21 100–1000 750 pH — 12 a-Naphthol concentration/mmol l21 5–200 125 Creatine concentration/mmol l21 5–150 100 Fig. 2 Influence of temperature (a) on the pervaporation process and (b) on the derivatization reaction for concentrations of diacetyl of 20 and 100 ng ml21 (dashed and continuous lines, respectively). 120 Analyst, February 1997, Vol. 122Chemical variables The influence of the pH in both the donor and acceptor solutions was studied. An aqueous solution adjusted to different pH values was used as a donor stream. A higher pervaporation efficiency was achieved by using doubly distilled water as the donor stream. However, the pervaporation efficiency was increased when the acceptor solution contained sodium hydroxide. This was tested by preparing different aqueous solutions containing sodium hydroxide at concentrations between 100 and 1000 mmol l21.A concentration of 750 mmol l21 provided the best medium for collection of the analyte. This solution was also appropriate for the development of the derivatization reaction. Concentrations of 125 mmol l21 a-naphthol and 100 mmol l21 creatine in the acceptor/reagent solution were selected as optimum. Higher concentrations of these reagents caused higher blank signals, resulting in appreciable lack of sensitivity (estimated as 50%).On the other hand, lower concentrations resulted in poor development of the chromogenic reaction. Hydrodynamic variables The flow rate had a dramatic influence on the performance of the system. As the flow rate of the donor solution determined the time during which the analyte was in the pervaporation module, 0.3 ml min21 was selected as optimum for the donor stream (lower values yielded non-reproducible results). On the other hand, the flow rate had no influence on the upper FI subsystem as the collection process occurred under static conditions.When the flow was re-started the reaction plug was driven to the detector. A flow rate of 1.8 ml min21 was appropriate to rinse the flow manifold. In order to obtain the best analytical signal with an acceptable sampling frequency, different stop-times in the range between 30 and 480 s were tried. A pervaporation time of 360 s was chosen as a compromise between the best signal and an acceptable sampling frequency.Injection volumes over 2 ml did not increase the analytical signal significantly. Features of the Method Ten standard solutions of diacetyl with concentrations between 1 and 5000 ng ml21 were prepared and injected in triplicate into the FI–pervaporation assembly using the optimum values of variables found previously. Table 2 summarizes the features of the method (equation, regression coefficient, linear range, detection and quantification limits and RSD).The linear range achieved for the method was suitable for applying it to the determination of diacetyl in beer samples. The sampling frequency achieved under the optimum working conditions was 8 h21. Study of Interferents The study of potential interferents was aimed at those commonly present in beer samples which are structurally similar to diacetyl, such as pyruvic acid, butylene glycol, acetoin (a precursor of diacetyl) and ascorbic acid. All interferents were added to the sample at concentrations higher than usually found in beers.The results showed that pyruvic acid, butylene glycol and ascorbic acid are tolerated at levels up to ten times their content in beer (with a tolerated interferent to analyte ratio of 100 : 1 for pyruvic acid and butylene glycol and 10 : 1 for ascorbic acid). Acetoin interferred at five times its concentration in these samples; nevertheless, acetoin is easily oxidized to diacetyl under atmospheric conditions.Application to the Proposed Method The method was applied to the determination of diacetyl in bottled and medium-process beer. In order to eliminate the gas contained in the original beverage (mainly CO2), all the samples were previously ultrasonicated for 5 min, then injected directly in triplicate. Table 3 summarizes the concentrations found and the results obtained after addition of two amounts of the analyte (50 and 100 ng ml21) to aliquots of the samples. The results provided by the proposed method were compared with those obtained by the conventional method used by Compa�n�ýa Andaluza de Cervezas.This method is based on previous separation of the analyte by di (vapour stream) followed by condensation of the analyte with ophenylenediamine, and photometric monitoring of the product at 335 nm. Six beer samples were used for this comparison and the results obtained produced a straight-line graph: y = 0.934x + 1.477 (r2 = 0.9901, n = 6) where y represents the FI–pervaporation method and x the conventional method.As can be seen, an excellent correlation exists between the proposed and the conventional methods. Conclusions The method proposed for the determination of diacetyl in beer samples shows the following features: simplicity in implementing the experimental set-up (i.e., a conventional FI manifold in Table 2 Features of the method Equation* A = 0.0199C + 0.0005 Regression coefficient 0.9998 (n = 7) Linear range/ng ml21 10–2000 Detection limit/ng ml21 5 Quantification limit/ng ml21 10 RSD (%)† — Low level 3.0 High level 2.6 * A denotes absorbance and C analyte concentration in ng ml21.† For 50 and 1000 ng ml21 of diacetyl. Table 3 Application of the method Concentration/ng ml21 Recovery (%)† Sample Conventional Proposed Difference No. Type* method method (%) Addition 1 Addition 2 1 A 18 17 25.5 100 96 2 A 15 16 6.5 94 98 3 A 25 24 24.0 98 97 4 B 30 29 23.3 100 99 5 B 25 27 8.0 107 99 6 B 52 50 23.8 100 100 * A and B denote bottled and medium-process beers, respectively.† 50 and 100 ng ml21 for additions 1 and 2, respectively. Analyst, February 1997, Vol. 122 121which an easy laboratory-built pervaporation unit is included); high selectivity as a result of the nature of the chemical reaction; lower sample consumption (2 ml); and an acceptable sensitivity. These features make it suitable for the routine determination of diacetyl in beer samples, as was demonstrated by comparison with the conventional method.The Comisi�on Interministerial de Ciencia y Tecnolog�ýa (CICYT) is thanked for financial support (project No. PB93- 0827). We gratefully acknowledge Carlos M. G�omez S�anchez of Compa�n�ýa Andaluza de Cervezas for the samples compared with the conventional method. References 1 West, D. B., Lautenbach, A. L., and Becker, K., Am. Soc. Brew. Proc., 1952, 81, 65. 2 Mathis, C., Pons, M. N., Engasser, J. M., and Lenoel, M., Anal. Chim. Acta, 1993, 279, 59. 3 Westerfeld, W. W., J. Biol. Chem., 1945, 161, 495. 4 Mattessich, J., and Cooper, J. R., Anal. Biochem., 1989, 180, 349. 5 Ribereau-Gayon, J., Peynaud, E., Sudraud, P., and Riberau-Gayon, P., Sciences et Techniques de Vin, Dunod, Paris, 1972. 6 Walsh, B., and Cogan, T. M., J. Dairy Res., 1988, 41, 31. 7 Garc�ýa-Vilanova, R. J., and Garc�ýa Estepa, R. M., Talanta, 1993, 40, 1419. 8 Mariaud, M., and Levillain, P., Talanta, 1994, 41, 75. 10 Barbieri, G., Bolzoni, L., Careri, M., Mangia, A., Paroladi, G., Spagnoli, S., and Virgili, R., J. Agric. Food. Chem., 1994, 42, 1170. 11 Otsuka, M., and Ohmori, S. J., J. Chromatogr. B, 1992, 115, 215. 12 Otsuka, M., and Ohmori, S. J., J. Chromatogr. B, 1994, 654, 1. 13 Damiani, P., and Burini, G., J. Assoc. Off. Anal. Chem., 1988, 71, 462. 14 Yamaguisgi, M., Ishida, J., Zhu, X., Nakamura, M., and Yoshitake, T., J. Liq. Chromatogr., 1994, 17, 203. 15 Ulberth, F., J. Assoc. Off. Anal. Chem., 1991, 74, 630. 16 Gilson, T. D., Parker, S. M., and Woodward, J. R., Enzyme Microb. Technol., 1991, 13, 171. 17 Luque de Castro, M. D., and Papaefstathiou, I., in Encyclopedia of Environmental Analysis and Remediation, ed. R. A. Meyers, Wiley, New York, in the press. 18 Prinzing, U., Ogbomo, I., Lehn, C., and Schmidt, H. L., Sens. Actuators B, 1990, 1, 542. 19 Ogbomo, I., Steffl, A., Schumann, W., Prinzing, U., and Schmidt, H. L., J. Biotechnol., 1993, 31, 317. 20 Mattos, I. L., Luque de Castro, M. D., and Valc�arcel, M., Talanta, 1995, 42, 755. 21 Papaefstathiou, I., Tena, M. T., and Luque de Castro, M. D., Anal. Chim. Acta, 1995, 308, 246. 22 Papaefstathiou, I., Luque de Castro, M. D., and Valc�arcel, M., Fresenius’ J. Anal. Chem., 1996, 354, 442. 23 Papaefstathiou, I., and Luque de Castro, M. D., Anal. Lett., 1995, 28, 2063. 24 Papaefstathiou, I., and Luque de Castro, M. D., Anal. Chem., 1995, 67, 3916. Paper 6/06401I Received September 17, 1996 Accepted October 28, 1996 122 Analyst, Feb