156 Anallysf, March, 1974, Vol. 99, $9. 156-162 A Rapid Method for the Semi-quantitative Determination of Volatile N-Nitrosamines in Alcoholic Beverages BY M. CASTEGNARO, BRIGITTE PIGNATELLI AND E. A. WALKER (Unit of Environmental Carcinogens, International Agency for Research on Cancer, 150 Cours Albert Thomas, 69008 Lyon, France) A method is described for the rapid extraction of N-nitrosamines from strong alcoholic drinks with methylene chloride after saturating the water - ethanol phase with magnesium perchlorate. The nitrosamines are then determined by oxidation to nitramines and clean-up of the oxidation products by adsorption chromatography on a column containing two layers of different grades of alumina , followed by gas chromatography with an electron-capture detector. The method is equally applicable to beverages with low alcohol content such as beer.THE N-nitrosamines have been widely investigated in studies on carcinogenesis, but while certain of them are carcinogenic to all animal species investigated, there is still no definite evidence of their carcinogenicity in man. In areas of Northern France, epidemiology studies are currently being carried out by us on the relationship between high alcohol consumption and cancer of the oesophagus.1 In order to further these studies, an investigation of analytical methods for the determination of nitrosamines in home-distilled spirits is being undertaken, particular attention being paid to reduction in the time of analysis. In a similar study of home-produced beers conducted in East Africa, Collis, Cook, Foreman and Palframan2$3 found no evidence of the presence of nitrosamines in concen- trations above 100 pg k g l , which was their minimum level of detection.However, such reliable evidence as is available on the presence of nitrosamines in foodstuffs usually indicates levels in the order of 1 to 10 pg kg-l. It would seem reasonable to require a useful screening method for determining nitrosamines in spirits to be capable of detecting similar levels. Isolation of nitrosamines from solutions containing appreciable amounts of ethanol presents problems in extraction and separation, as the lower homolopes of the dialkyl- nitrosamines are freely soluble in water, ethanol and solvent phases. Distillation also presents problems, which arise from the formation of azeotropes.Crosby, Foreman, Palframan and Sawyer4 attempted to concentrate the nitrosamines by using a spinning band column, but the concentration factor obtained was only modest and the technique is slow. An alternative method has been proposed by Sen and Dalpe5 in which the alcohol is removed as an azeotrope with either benzene or toluene. The method, however, is somewhat lengthy for screening large numbers of samples. The authors report a method of analysis for nitrosamines in alcoholic beverages, which involves the use of a simple liquid-liquid extraction of nitrosamines and is suitable for screen- ing large number of samples. The final determination of the nitrosamines is made by gas chromatography of the nitramine derivatives according to the method of Sen,6 after clean-up on a column of alumina.EXPERIMENTAL As methylene chloride has been found by most workers to be the most suitable solvent for the extraction of nitrosamines, attempts were first made to extract, with methylene chloride, a 1 + 1 water - ethanol mixture containing known amounts of seven different nitrosamines. After separating it, the solvent phase was dried over sodium sulphate and distilled in a Kuderna-Danish apparatus. The collected fractions of the distillate and the residue were then analysed for nitrosamines by gas chromatography. As expected, the co-extracted ethanol carried into the solvent phase an appreciable amount of water, with @ SAC and the authors.CASTEGNARQ, PIGNATELLI AND WALKER 157 the result that the nitrosamines passed into the distillate and were lost.The problem, therefore, was to prevent extraction of water and ethanol into the methylene chloride phase. Calcium chloride is known to bond loosely with ethanol; therefore, extraction was attempted with simulated spirit (1 + 1 water - ethanol), after first saturating the solution with calcium chloride. However, on shaking the mixture in a separator, emulsions formed that were difficult to break down. As an alternative, magnesium perchlorate, which is soluble in ethanol and has a high affinity for water, was tried, with success. On separation, the volumes of the two layers were only slightly changed and the solvent could be completely distilled off without significant increase in the temperature of distillation.Ethanol was mainly retained in the water. A minor problem was encountered in differentiating the layers when the concentration of ethanol was varied about the ratio 1 : 1. As the difference in density between the perchlorate-saturated water - ethanol solution and methylene chloride is small, for relatively small variations in the percentage of ethanol, the methylene chloride may be found as either the upper or the lower layer. On addition of sufficient ethanol to bring its content to about 55 per cent., the two layers separate cleanly, with methylene chloride as the bottom layer. All subsequent extractions were made on solutions containing 55 per cent. of ethanol. Alternatively, if concentrations of ethanol are adjusted with water to be less than 45 per cent., the phases readily separate with methylene chloride as the upper layer.The method can thus be varied to suit the type of product to be analysed. The partition between the two phases was checked on 10 p.p.m. solutions of nitrosamines in 1 + 1 water - ethanol and by measuring those in the methylene chloride extract in an ultraviolet spectrophotometer. Three extractions with methylene chloride were sufficient to remove all the nitrosamines from the perchlorate-saturated water - ethanol phase. This extraction technique was applied to solutions of ethanol in water containing 10 pg k g l of each of seven nitrosamines and the methylene chloride extract analysed for the nitrosamines. The recoveries were generally found to be of a similar order to those found by Sen and Dalpe.5 The method was then applied to samples of nitrosamine-free calvados, both with and without addition of the nitrosamines. Accurate analysis was difficult owing to the presence of a large number of interfering peaks.Attempts to reduce their number by distillation of the calvados from alkaline and acidic solution were ineffective and, furthermore, led to lower recoveries, particularly of the higher-boiling nitrosamines. The problem was overcome by using an alumina column similar to that used by Telling' for the determination of nitrosamines in foodstuffs. The method required some modification as the neutral-grade alumina used by this author did not remove all the interfering peaks in the chromatograms originating from the calvados. Those peaks which remained were removed by using an alkaline grade, and an effective composite column of the two types of alumina was devised .Fig. 1 illustrates a chromatograni obtained from a calvados sample that contained 10 pg k g l of each of seven nitrosamines before clean-up. Figs. 2 and 3 illustrate the chromatograms obtained from the same calvados sample with and without addition of the nitrosamines after clean- up on the alumina column. METHOD OF ANALYSIS FOR NITROSAMINES IN CALVADOS MATERIALS- N-Nitroso cosqbounds-Nitrosodimethylamine, nitrosodiethylamine , nitrosodipropylamine and nitrosodibutylamine were obtained from Eastman Kodak Ltd. and nitrosomethylpentyl- amine from Schuchardt, Munich, and all were distilled before use. Nitrosopyrrolidine was kindly supplied by Dr.Eisenbrand of the Krebsforschungszentrum, Heidelberg, and nitroso- ethylmethylamine by Mr. J. F. Palframan of the Laboratory of the Government Chemist, London. All samples contained gas-chromatographic impurities in concentrations of less than 1 per cent. SOLVEXTS- Methylew chloride-Merck, analytical-reagent grade, was redistilled from sodium car- bonate in 2-litre batches, the first 100 ml being rejected. All other solvents were of analytical- reagent grade. REAGENTS- Hydrogen peroxide solution, 50 per cent. m/V.158 CASTEGNARO et al. : SEMI-QUANTITATIVE DETERMINATION OF [Analyst, Vol. 99 Trijuoroacetic acid-Merck, analytical-reagent grade, was redistilled before use. Magnesium perchlorate-Prolabo, laboratory grade. Sodium sulphate, adzydrous-Merck, analytical-reagent grade.50 40 30 20 10 Time/minutes I Fig. 1. Chromatogram illustrating the masking of nitramine peaks by other oxidation products : arrows indicate the position of the peaks for dimethylnitramine (a) ; ethylmethyl- nitramine (b) ; diethylnitramine (c) ; dipropylnitramine (d) ; methylpentylnitramine (e) ; dibutylnitramine (f) ; and pyrrolidylnitramine (g) EXTRACTION- The approximate percentage of ethanol in the calvados is first measured by means of a hydrometer. To a 50-ml aliquot of sample sufficient ethanol is added to bring its concen- tration to about 55 per cent. To this solution is then added sufficient anhydrous magnesium perchlorate to saturate the solution, the magnesium salt being added in successive amounts of about 10 g and the solution cooled under a tap after each addition (interaction between water and anhydrous magnesium perchlorate is exothermic).The saturated solution is then extracted with three successive 50-ml portions of methylene chloride, which are bulked and shaken in a second separator with 35 to 40 g of anhydrous sodium sulphate so as to remove any of the water - ethanol phase that may be carried through in suspension, and the methylene chloride layer is run off into a suitable apparatus for removal of the solvent. The sodium sulphate is finally washed with 1 O m l of methylene chloride, which is added to the bulked extracts. Two methods of evaporation have been carried out, (a) by using the Kuderna-Danish apparatus, and (b) by evaporation from a 250-ml round-bottomed flask immersed in a water- bath at 60 "C and flushed with a gentle current of nitrogen.In each instance the volume of solvent is reduced to between 0.5 and 1 ml. The results shown in Table I indicate that better recoveries for the more volatile nitrosamines are obtained by using the Kuderna-Danish apparatus. The second method, however, is much more rapid and shows little disadvantage for the higher nitrosamines.March, 19741 VOLATILE N-NITROSAMINES IN ALCOHOLIC BEVERAGES 159 PREPARATION OF NITRAMINES- The nitrosamines in the residue are oxidised to nitramines with a mixture containing 5 ml of trifluoroacetic acid and 4 ml of 50 per cent. hydrogen peroxide solution.6 After gently shaking it overnight at room temperature, the solution is cooled in an ice-bath, the pH adjusted to between 10 and 11 by dropwise addition of a 20 per cent.solution of potassium carbonate and then extracted with three successive 50-ml volumes of methylene chloride. The extracts are bulked in a separator and dried by shaking with 15 to 20 g of anhydrous sodium sulphate. The methylene chloride is then transferred into a flask as before. The sodium sulphate is washed with 20 ml of methylene chloride, which is added to the bulked extracts, and the volume of solvent reduced in the water-bath at 60 "C to slightly less than 10ml. The concentrate is then transferred, with washing, into a graduated test-tube and its bulk is further reduced in a current of nitrogen to slightly more than 6 ml (it is important not to reduce the volume below this level as nitramines may be lost); 2 ml of hexane are then added and the bulk further reduced to between 0-7 and 06ml (again, care must be taken not reduce the volume below 0.5ml).The volume is finally adjusted to 5ml with pentane and the solution is ready for the final clean-up by column chromatography. I I I I I U 50 40 30 20 10 0 Time/rninutes Fig. 2. Chromatogram illustrating the seven nitramines a to g (Fig. l), gas chromatographed a t 130 "C, as indicated by the arrows, after clean-up on the alumina column COLUMN CHROMATOGRAPHY- The column consists of three layers: a bottom layer of 3 g of basic aluminium oxide, activity 11; a middle layer of 3 g of neutral aluminium oxide, activity 111; and, at the top, a layer of 1 g of anhydrous sodium sulphate. A glass column, 1 em in diameter and 25 cm in length, is used.Each type of oxide is prepared by first activating the appropriateMerck grade (ie., basic or neutral) of activity I by heating it in an oven for 3 hours at 240 "C. After cooling in a desiccator, 100 g of the activated material are shaken in a stoppered flask for 3 hours with the appropriate amount of water (3g for basic aluminium oxide, activity 11, and 6 g for neutral aluminium oxide, activity 111). The stoppered flask is then left to stand overnight so as to enable the mixture to equilibrate. Three grams of the basic aluminium oxide (activity 111) are then agitated gently for half an hour with a sufficient volume of a 1 per cent. solution of &ethyl ether in pentane160 CASTEGNARO et al. : SEMI-QUANTITATIVE DETERMINATION OF [Analyst, Vol.99 I I I I Time/m i nu tes Fig. 3. alumina column. (Fig. 1) would be found Chromatogram obtained for a blank calvados after clean-up on the Arrows indicate the positions a t which the nitramines a to g to cover the solid and desorb gases from the alumina. The washed alumina is then added to the glass column and solvent run off from the bottom of the column until the top of the solvent layer is level with the upper surface of the alumina. This procedure is repeated with the neutral aluminium oxide (activity 11) and then finally 1 g of anhydrous sodium sulphate is added to the top of the column. The pentane solution containing the nitramines is then added to the column and liquid run out until the upper level coincides with the upper surface of the column.The column is eluted successively, first with 25 ml of a 1 per cent. solution of diethyl ether in pentane, then with 40ml of a 25 per cent. solution of diethyl ether in pentane and finally with diethyl ether alone. The first 25 ml of eluate are rejected, after which it is collected in 5-ml fractions. From each fraction a 5-pl portion is injected into the gas chromatograph. Finally, the solutions are bulked and the volume is reduced to 5ml by evaporation on a water-bath. A suitable aliquot is then injected into the gas chromatograph. The minimum level of detection is about 1 pg k g l of nitrosamine in the original sample. The volume of ether can be further reduced to 0.5 ml without loss of nitramine, giving a factor of 10 on the peak size. This step is preferable for accurate measure- ments of the concentrations below 10 pg k g l .So far it has not been possible to adjust conditions so as to obtain complete separation of the individual nitramines on the alumina column. However, as the retention volume of each nitramine is reasonably reproducible, analysis of each fraction separately increases confidence in identification. The nitramines are eluted from the column in order of polarity, the least polar being eluted first, i.e., in the order: dibutyl, dipropyl, diethyl, methylpentyl, ethylmethyl, dimethyl and, finally, the pyrrolidyl compound. The total volume of eluate is normally about 55 ml but varies slightly from batch to batch of the prepared alumina, and it is therefore necessary to run nitramine standards for each batch.The variation has been found not to be more than 5 ml for the nitramines that have the longest retention times when using freshly prepared alumina. Recovery of the nitramines after having passed through the column, the fractions having been bulked and the volume reduced, is 100 per cent.March, 19741 VOLATILE N-NITROSAMINES IN ALCOHOLIC BEVERAGES 161 GAS CHROMATOGRAPHY- Pye 104 gas chromatographs fitted with nickel-63 electron-capture detectors were used with 6 foot x 4 inch 0.d. glass columns packed with 10 per cent. Carbowax 20M on DMCS- treated Chromosorb W. The columns were conditioned at 220 "C for 2 days before use. The determination of nitramines in the series dimethyl to the methylpentyl derivative was performed isothermally a t 130 "C with a nitrogen flow of 50 ml min-l through the column and a scavenge flow of 20mlmin-l through the detector.The injection temperature was 170 "C and the chart speed 5 mm min-1. The determination of the dibutyl- and pyrrolidyl- nitramines was performed on a second chromatograph at 160 "C. With a second instrument available this procedure was found to be more convenient than temperature programming. RESULTS The recoveries obtained in the replicate analyses of the aqueous ethanolic solution containing 10 pg k g l each of seven nitrosamines are illustrated in Table I, where the results shown were obtained by using both methods for evaporation of the solvent. Table I1 shows the recoveries for the same concentration of nitrosamines in four different samples of calvados; in this instance only the rapid method of evaporation in the water-bath was used.Recoveries were good, with the exception of nitrosodimethylamine and nitrosopyrrolidine, losses of which, even when using a Kuderna-Danish apparatus, still tended to occur. TABLE I RECOVERY OF THE SEVEN NITROSAMINES FROM THE WATER - ETHANOL MIXTURE CONTAINING 10 pg kg-1 OF EACH NITROSAMINE Recovery of each nitrosamine, per cent. A f 1 (4 Nitrosamine - Nitrosodimethylamine . . .. .. 33 49 46 Nitrosoethylmethylamine . . .. 52 64 89 Nitrosodiethylamine . . .. .. 86 78 76 Nitrosodipropylamine . . .. .. 83 98 82 Nitrosodibutylamine . . .. .. 96 110 100 Nitrosomethylpentylamine . . .. 97 92 79 Nitrosopyrrolidine . . . . .. 70 70 43 (4 7 - 42 26 48 38 22 79 33 60 47 43 71 95 64 64 57 79 83 100 100 72 75 86 100 100 95 104 100 89 100 100 40 30 60 70 60 (a) Methylene chloride removed by using a Kuderna-Danish apparatus.(b) Methylene chloride removed by evaporation on a water-bath. As acidic and alkaline distillation stages are frequently used in the clean-up of nitrosamines extracted from food, a number of analyses of both water - ethanol and calvados containing nitrosamines were carried out with the inclusion of a distillation stage. Distillation was carried out nearly to dryness and the distillate analysed by using the method described above. The results shown in Table I11 indicate considerable losses of higher nitrosamines, which, we feel, indicates the value of a liquid - liquid extraction technique in analysing alcoholic samples for nitrosamines. TABLE I1 RECOVERY OF SEVEN NITROSAMINES, ADDED AT THE 10 pg k g l LEVEL, FROM FOUR DIFFERENT SAMPLES OF CALVADOS* Nitrosamine Nitrosodimethylamine .. .. .. .. Nitrosoethylmethylamine . . .. .. Nitrosodiethylamine . . .. .. .. Nitrosodipropylamine . . .. .. .. Nitrosomethylpentylamine . . .. .. Nitrosodibutylamine . . .. .. .. Nitrosopyrrolidine . . .. .. .. Recovery of each nitrosamine, per cent. it3 82 2s 2; 76 83 76 73 45 71 44 49 88 99 93 83 99 102 80 98 102 103 63 68 40 40 26 50 * Methylene chloride was removed by simple evaporation on a water-bath.162 CASTEGNARO, PIGNATELLI AND WALKER TABLE I11 RECOVERY OF NITROSAMINES FROM THE DISTILLATE OF A WATER - ETHANOL Each sample was initially spiked with nitrosamines at the level of 10 pg k g l MIXTURE AND FOUR DIFFERENT SAMPLES OF CALVADOS Nitrosamine Nitrosodimethylamine .. Nitrosoethylmethylamine Nitrosodiethylamine . . Nitrosodipropylamine . . Nitrosomethylpentylamine Nitrosodibutylamine . . Nitrosopyrrolidine . . Recovery of each nitrosamine, per cent. A f- > (4 (b) - f 1 .. .. 20 31 34 14 85 16 8 .. .. 21 83 58 26 84 77 106 .. . . 29 Trace 59 Trace 71 19 27 I . . . 33 Trace 66 Trace 91 28 32 .. .. 45 65 75 Trace 90 18 22 .. . . 22 Trace Trace Trace 100 31 29 .. . Trace Trace Trace Trace Trace 25 37 (a> Water - ethanol mixture. (b) Four different samples of calvados. As an example of samples with low alcohol content, the liquid - liquid extraction technique was used for the analysis of a number of beers spiked with the seven nitrosamines as before. Recoveries again were in the same range as those for calvados. However, as interference from other oxidation products in the gas-chromatographic analysis was considerably less than for calvados, it was possible to eliminate the column clean-up procedure, thus considerably diminishing the time of analysis. REFERENCES 1. Tuyns, A. J., I d . J . Cancer, 1970, 5, 152. 2. Collis, C. H., Cook, P. J., Foreman, J. K., and Palframan, J. F., Lancet, 1972, ii, 441. 3. ---- , Gut, 1971, 12, 1015. 4. Crosby, N. T., Foreman, J. K., Palframan, J. F., and Sawyer, R., in Bogovski, P., Preussmann, R., and Walker, E. A., Editors, “N-Nitroso Compounds, Analysis and Formation. Proceedings of a Working Conference held a t the Deutsches Krebsforschungszentrum, Heidelberg, 13-15 October, 1971,” IARC Scientific Publication No. 3, International Agency for Research on Cancer, Lyon, France, 1972, p. 38. 5. Sen, N. P., and Dalpe, C., Analyst, 1972, 97, 216. 6. Sen, N. P., J . Chromat., 1970, 51, 301. 7. Telling, G. M., Ibbid., 1972, 73, 79. Received July 17tk, 1973 Accepted August 29tk, 1973