MAY, 1966 THE ANALYST Vol. 91, No. 1082 The Quantitative Microanalysis of Carbonyl Compounds BY A. &I. PARSOYS ( Unilever Research Laboratory, Welwyn, H u t s ) Several published methods for preparing 2,4dinitrophenylhydrazones and determining them colorimetrically have been examined, and a practical system of analysis has been devised from them. The key step is the use of a 2,4-dinitrophenylhydrazine - 66 per cent. phosphoric acid column eluted with benzene, as well as light pctroleurn. The results of an analysis in the difficult case of a butter fat are described. IN connection with flavour and other work a procedure was required for the quantitative microanalysis of complex mixtures of carbonyl compounds. Winter et aZ.l found that at least twelve carbonyl compounds were removed from butter by steam distillation.The amounts involved ranged from 18-8 p.p.m. for acetoin to less than 0.01 p.p.m. for hexanal, nonanal and nonan-2-one. Low ~oncentrations~9~ of monocarbonvl compounds have been im- plicated in the reversion flavours of edible oils4q5 and in the fishy flavour of dairy product^.^^^^^ Although other suggestionsg ,lo have been made, the colorimetric determination of 2,4-dinitrophenylhydrazones was selected as being the most generally useful technique. A great deal of work has been published on the chromatographic separation and spectro- photometric examination of 2,4-dinitrophenylhydrazone~, but the strictlv quantitative preparation of the derivatiL7es with 2,4-tlinitrophenylhydrazine has received relatively little attention. In a comparative study, Begemann and de JongL1 showed that the reaction of a dilute ethereal solution of a carbonyl compound with 2,4-dinitrophenylhydrazine in mineral acid gave an incomplete conversion, the yield in homogeneous solution12 being worse than that obtained when the solutions were partially13 or almost completely imrni~cib1e.l~ Water was present in all three solutions, and unless sufficient carbonyl compound was present for the derivative to separate, the equilibrium mixture obtained presumably fell far short of complete conversion into the 2,4-dinitrophenvlhydrazone.Henick et aZ.15916 carried out the reaction under essentially anhydrous conditions in benzene with trichloroacetic acid as the catalyst. There seems to be no reason why this reaction should not go to completion; a Dean and Stark apparatus can be used toremove water if necessary.17 However, the blank values were high (optical density in benzene - ethanolic potassium hydroxide, E = 0.35) and it was not clear whether thiswas due to excess reagent or to an artifact consequent upon the use of trichloroacetic acid.In addition, the presence of dicarbonyl compounds causes a large error in the calculation of carbonyl content, and therefore the procedure is largely useless for the empirical examination of oiIs for which it was originally proposed.l* Pool and Kloselg claimed a quantitative reaction of aldehydes in benzene with 2,4-dinitro- phenylhydrazine that had been adsorbed on to an alumina column, but their extinction coefficient ( E = 19,200 at 435 mp in benzene - ethanolic potassium hydroxide at zero time) appears to include a correction factor; the now accepted value1* (6 = 20,930 at 430 mp after 10 minutes) is significantly higher.Hegemann and de Jongll obtained a 90 per cent. con- version of heptanal into its 2,4-dinitrophenylhydrazone (assuming E = 22,500 at 358 mp in chloroform) but only 20 per cent. for nonan-%one on this column. Keith and L)ayls obtained 75 per cent. yields with alkanals, 65 per cent. with alk-2-enals and 60 per cent. with alka-2,4-dienals. The loading of this column is very low. Pool and KloseI9 recommended 0-05 to 0.50 pmole and Keith and Day1* up to 1 pmole of carbonyl compounds for a 10-g column, although the amount of reagent present is about 25 pmoles. In addition, decomposition may occur; for example, Lea and Jackson20 found that hydroperoxides give some carbony1 cornpounds under these conditions.298 PARSONS : QUANTITATIVE MICROANALYSIS OF CARBONYL COMPOUNDS [Andyst, VOl.91 Forss et appear to have been the first to prepare 2,4-dinitrophenylhydrazones by shaking 2,4-dinitrophenylhydrazine in 2 N hydrochloric acid with solutions of carbonyl compounds in light petroleum. Under these conditions the derivatives are obtained free from excess reagent in a surprisingly short time. Begemann and de Jongll found that the reaction with undecan-%one was frequently complete in 4 hours. However, the results were more reproducible and the reaction still more rapid if the petroleum solution was percolated through a column of the aqueous phase supported on Celite. In this way, 0.3 to 5 pmoles of heptanal, heptan-2-one, non-2-enal, undecan-2-one, tridecanal and pentadecan-%one gave almost complete conversion into their respective 2,4-dinitrophenylhydrazones in 1 hour with 2,4-dinitrophenylhydrazine (about 250 pmoles) in 7-5 ml of 2 N hydrochloric acid on 15 g of Celite.Subsequently, Schwartz and Parks22 suggested that the solution of 2,4-dinitrophenyl- hydrazine in 2 N hydrochloric acid could be replaced advantageously by 2,4-dinitrophenyl- hydrazine in 66 per cent. phosphoric acid. This column contained a relatively large amount of reagent (about 100 pmoles per g of stationary phase), from which impurities were removed by washing with benzene. A further advantage, particularly for the analysis of oils, is that the column does not decompose hydroperoxides to carbonyl whereas the 2 N hydrochloric acid column2* does (80 to 88 per cent.), either because hydrochloric acid is stronger than phosphoric acid, or because of the catalytic effect of the chloride ion.25~26527 All workers in this field have stressed the importance of the ubiquitous nature of carbonyl compounds and the necessity for the careful purification of the solvents used.EXPERIMENTAL 2,4-DINITROPHENYLHYDRAZINE- The commercially available material (e.g., 2,4-dinitrophenylhydrazine AnalaR, British Drug Houses Ltd., m.p. 196" to 199" C) was suitable for most purposes, any impurities being removed from its solution in aqueous mineral acid by filtration and washing with carbon tetrachloride28 or a carbonyl-free petroleum solvent.However, this method of purification was not possible for solutions in ethanol or benzene, and the reagent was purified as described under Method. Xn alternative procedure, extraction with light petroleum in a Soxhlet apparatus,ll gave material which still imparted a colour to the organic phase when distributed between 2 N snlphuric acid and purified light petroleum. Little purification could be effected by simple re-crystallisation from methan01.l~ CYCLOHEXANE AND LIGHT PETROLEUM- Useful indications of the carbonyl contents of these solvents were obtained by shaking samples with a saturated solution of 2,4-dinitrophenylhydrazine in 2 N hydrochloric acid overnight, in a Griffin wrist-action flask shaker (Messrs. Griffin and George Ltd.).The resulting concentration of 2,4-dinitrophenylhydrazones in the organic phase was obtained by dividing the value of the optical density at its maximum (about 340 mp) by the value of E = 23,700. It should be noted that not only is the adsorption maximum 20mp lower in hesane than it is in chlor~form,~~ ethanol or benzene, but the extinction coefficient is 5 per cent. higher. \Ye found that the lowest value for E occurred in a 4 to 1 v/v mixture of benzene and ethanol. The heavily contaminated samples of cyclohexane and light petroleum gave precipitates which were taken up in benzene and determined after dilution with ethanolic potassium hydroxide.17 The results obtained are shown in Table I. TABLE I CARBONYL COMPOUNDS IN SOME LABORATORY SOLVENTS Solvent Carbonyl concentration, pmolar Hexane .. . . .. .. . . . . 3000 Hexane (spectroscopic grade) .. .. 14 Light petroleum (b.p. 60" to 80" C) . . 320 Light petroleum (b.p. 40" to 60" C) . . 330 Light petroleum (aromatic free) . . .. 17 Cyclohexane . . . . . . .. .. 37 , - - - - I - I . ~ - - ~ - ~ I ~ . . ~ ~ + ~ ~ ~ ~ ~ ~ ; ~ m,.3ao\ i n .. ..May, 19661 PBRSONS QUANTITATIVE MICROANALYSIS OF CARBONYL COMPOUNDS 299 Thus a good grade of cyclohexane or light petroleum was adequate for most purposes. By using the procedure described by Van der Ven and de Jonge30 the carbonyl content of aromatic-free light petroleum was still further reduced to 2 pmolar. BENZENE- Carbonyl compounds in benzene were found to react incompletely when shaken overnight with a solution of 2,4-dinitrophenylhydrazine in mineral acid.They were therefore determined by adding 1 mg of purified 2,4-dinitrophenylhydrazine and 10 mg of trichloroacetic acid to a 10-ml sample, leaving to stand overnight and measuring the adsorption at 430mp after dilution with 1-8 per cent. ethanolic potassium hydroxide. The results of an investigation into the accuracy of this procedure are described in the next section. I t was found that consistent results could best be obtained by adding the benzene solution to an equal volume of freshly prepared 1.8 per cent. ethanolic potassium hydroxide (obtained by shaking 5 potassium hydroxide pellets with 25ml of re-rectified ethanol at room tem- perature, and filtering) and reading the optical density with a self-recording spectrophotometer (Cnicam SP800) after exactly 10 minutes.At low carbonyl concentrations, absorption due to unchanged reagent was too large to be ignored31; so measurements were also made at 360 mp, and calculations performed after determining the appropriate extinction coefficients- %som~* E ~ ~ o ~ I J . 2,4-Dinitrophenylhydrazine . . . . 1860 1180 2,4-Dinitrophcnylhydrazones. . . . 3400 22,300 The last figure is significantly higher than that used by Keith and Day,19 but is consistent with the values obtained by Jones et aZ.32 if it is assumed that solvents, unlike oils, are liable to contain more ketones than aldehydes. The formulae derived from the above extinction coefficients were as follows- 2,4-Dinitrophenylhydrazine = (594 E,,,, - 90 E,,,) pmolar 2,4-Dinitrophenylhydrazones = (49-7 E,,, - 31.4 E,,,) pmolar The carbonyl content of sulphur-free benzene proved to be variable (10 to 500 pmolar), even with bottles labelled with the same batch number.One or two distillations from 2,4-dinitrophenylhydrazine - trichloroacetic acid17 reduced this value to less than 4 pmolar. Details of this purification are given under Method. REACTION OF CARBONYL COMPOUNDS \VITH 2,4-DINITROPHEh'YLHYDRAZINE AND TRICHLORO- ACETIC ACID I N BENZENE- An attempt was made to check the Henick p r o c e d ~ r e l ~ ? ~ ~ by following the reaction with a model compound at room temperature. Purified 2,4-dinitrophenylhydrazine (2.6 mg, 12.9 pmoles), in 10 ml of carbonyl-free benzene, was added to nonan-2-one (101 pg, 0-72 pmole) and trichloroacetic acid (186 mg, 1-14 pmoles) in 15 ml of the same solvent.At intervals a sample was withdrawn, diluted with an equal volume of 1.8 per cent. ethanolic potassium hydroxide, and examined after 10 minutes in the self-recording spectrophotometer (time of scan was 2 minutes) against a blank of trichloroacetic acid in benzene - ethanolic potassium hydroxide. A control was run under the same conditions, the nonan-%one being omitted. The net apparent quantities of saturated and unsaturated carbonyl 2,4-dinitrophenylhydra- zones obtained are shown in Table 11. TABLE IT APPAREXT EXTEST OF REACTIOB BETITEEN NONAN-2-ONE AND 2,4-DINITROPHENYLHYI)KAZINE I N BENZENE IN THE PRESENCE OF TRICHLOROACETIC ACID Time of reaction a t 23" C 5 minutes 30 minutes 23 hours saturated carbonyls, pmoles .. . . 0.79 0.79 1.03 Saturated carbonyls, per cent. . . . . 110 110 143 Unsaturated carbonyls, pmoles. . . . - 0.05 - 0.07 - 0.24 Unsaturated carbonyls, per cent. . . - 7 - 10 - 33 The negative values are a consequence of the magnitude of the readings given by the control. In another experiment, purified 2,4-dinitrophenylhydrazine (2.5 mg, 12-5 pmoles) in 10 ml of carbonyl-free benzene was added to trichloroacetic acid (146 mg, 900 pmoles) in the same solvent. Samples were withdrawn and examined as before, and the optical densities obtained are shown in Table 111.300 PARSONS : QUANTITATIVE MICROANALYSIS OF CARBONYL COMPOUKDS [ A d y s t , Vol. 91 TABLE I11 REACTION BETWEEX 2,4-DI?;ITROPHEKYLHYDRAZINE AND TRICHLOKOACETIC ACID I N BENZENE Time of reaction at 23" C E360 Ed30 &60 5 minutes 0.63 0.30 0.18 30 minutes 0.6 1 0.30 0.18 6 hours 0.63 0.32 0.2 1 24 hours 0.68 0-39 0.26 47 hours 0.70 0.43 0.30 120 hours 0.75 0.5 I 0.37 The explanation of the high readings of the control appears to be that 2,4dinitrophenyl- hydrazine reacts with trichloroacetic acid to give a pigment which is green in alkaline solution.It is possible to allow for this effect by assuming that the pigment has similar absorption (at 360 and 430 mp) to 2,4-dinitrophenylhydrazine and by using the formulae given in the preceding section. This method for computing the results was applied to the original experiment and the results are set out below- Time of reaction a t 23" C 6 minutes 30 minutes 23 hours Carbonyl 2,4-dinitrophenylhydrazones, pmoles .. 0.67 0-(i9 0.73 Carbonyl 2,4-dinitrophenylhydrazones, per cent. . . 93 96 101 Applied in this way the procedure gives reasonable results with much larger concen- trations of carbonyl compounds. By using 2,4-dinitrophenylhydrazine (2.5 mg, 12.5 pmoles) and nonan-2-one (1-01 mg, 7-2 pmoles) under conditions similar to those described in the foregoing section, the results are shown below- Time of reaction a t 23" C 5 minutes 30 minutes 24 hours Carbonyl 2,4-dinitrophenylhydrazones, pmoles . . 4.7 6.9 7 . 1 Carbonyl 2,4-dinitrophenylhydrazones, per cent. . . 65 96 99 For analyses involving subsequent fractionation of the derilratives by chromatography, however, the presence of artifacts from the reagents was clearly undesirable. For example, it was observed that a solution of 2,4-dinitrophenylhydrazine in 2 s aqueous trichloroacetic acid deposited a crystalline precipitate on standing at room temperature, and after 2 weeks the supernatant liquid was almost colourless.REACTIOY BETWEEN 2,4-DINITROPHEXYLHYDKAZINE IN AQLJEOCS ACID AND SOh'AX-?-ONE IN \'A RI 0 LT S SOLVENTS- (a) Light PetroZeztm-By using the wrist-action shaker under standard conditions it was possible to follow reactions in systems involving two immiscible liquids. The reaction between nonan-2-one (210 pg, 1.46 pmoles) in 5 ml of light petroleum and 2,4-dinitrophen~~lhydrazine (about 5 pmoles) in 5 ml of 2 N sulphuric acid proceeded as shown in Table 1iT. TABLE IC: REACTION BETM'EEN NONAN-2-OX;E 13 LIGHT PETIIOLEUM AKD 2,4-DINITKOPHENYLHYDK.~Zl~E I N 2 h' SVLI'HrRIC ACID Time, in minutes Percentage conversion Time, in hours Percentage conversion 0 0 1 39 5 2.8 2 58 10 5.0 5 85 30 24.7 16 102 Thus the reaction was essentially (95 per cent.) complete in 6 hours, as was also the case with the carbonyl compounds studied by Forss et aZ.21 and by Begemann and de Jongell under similar conditions. ( b ) Hexune - benzene-The reaction between nonan-2-one (470 pg, 3.3 pmoles) in 10 ml of hexane - benzene and 2,4-dinitrophenylhydrazine (about 4 pmoles) in 10 ml of 5 per cent.v/v sulphuric acid was studied in a similar manner with the ethanolic potassium hydroxide procedure to determine 2,4-dinitrophenylhydrazones in the presence of 2,4-dinitrophenyl- hydrazine. The results, set out below, clearly show the retarding effect of benzene on the reaction- Benzene, per cent.. . 0 20 50 90 Percentage conversion in 16 hours . . 33 10.2 4.1 2.5 Percentage conversion in 65 hours . . 56 28 14.6 9.1May, 19661 PARSONS : QUANTITATIVE MICROANALYSIS OF CARBONYL COMPOUNDS 301 REACTION BETWEEN 2,4-DINITROPHENYLHYDRAZII\;E IN AQUEOUS ACID AND VARIOUS CARBONYLS IN LIGHT PETROLEGM- Shaking experiments-A saturated solution of 2,4-dinitrophenylhydrazine (about 5 pmoles) in 5 ml of 2 N sulphuric acid was shaken with the carbonyl compound (1 to 3 pmoles) in 5 ml of light petroleum for a suitable time. A sample was withdrawn from the organic phase and examined in the spectrophotometer. The concentrations of the 2,4-dinitrophenylhydrazones of nonan-2-one, acetophenone and benzophenone were calculated, assuming E = 22,600, 24,100 and 28,300, re~pectively.~~ IVith benzaldehyde, however, a precipitate separated at the interface ; benzene was therefore added and the concentration of 2,4-dinitrophenylhydrazone was obtained by the ethanolic potassium hydroxide procedure, assuming E = 33,300.32 The results are set out in Table 1'.The relative rates of reaction were benzaldehyde > nonan-2-one > acetophenone > benzophenone. These results agree with those of Belcher and Fleet,33 who found that benzophenone reacted very slowly with hydroxylamine in homogeneous solution. REACTION BETWEEN 2,4-L)INITROPHENTLHIDKAZINE IN 2 N SULPHURIC ACID AND VARIOUS CARBONYL COMPOCKDS I N LIGHT PETROLElJM Concentration (time = O), pmolar Carbonyl compound Renzaldehyde .. . . 570 Nonan-2-one . . . . 197 Acetophenone . . .. 490 .4cetophenone . . . . 490 Acetophenone . . . . 490 Benzophenone . . . . 268 Benzophenone . . . . 268 Time, in hours 2 2 2 16 67 16 67 Percentage conversion 04 61 48 63 83 20 9.9 T H E 2,4-DINITROPHENYLHYDRAZINE - 66 PER CENT. PHOSPHORIC ACID COLIJMN- This column was made up as described22 and gave a quantitative conversion of benzo- phenone (3-9 pmoles and 0.39 pmoles) in 10 ml of cyclohexane into its 2,4dinitrophenyl- hydrazone at a flow-rate of 22 ml per hour. Higher rates of flow gave lower yields, only 19 per cent. conversion being obtained at 97 ml per hour. A second column, 1.0 x 18.5 cm, was made with 10.0 g of stationary phase, i.e., at least 900 pmoles of 2,4-dinitrophenylhydrazine after washing.This column gave 73 per cent. conversion of 3.9 pmoles of benzophenone into its 2,4-dinitrophenylhydrazone at 8 ml per hour. Nonan-2-one (6-0 pmoles) was quantitatively converted at 20 ml per hour, which is the flow- rate laid down by Begemann and de Jongell with a column of similar size containing 2,4-dinitrophenylhydrazine - 2~ hydrochloric acid. The reaction is reversible. A solution of butanone 2,4-dinitrophenylhydrazone (8.25 pmoles) in cyclohexane was decolourised completely by passage over a 1.0 Y 9 6 c m column of Celite impregnated with 66 per cent. phosphoric acid. Octanol 2,4-dinitrophenyl- hydrazone (6.9 pmoles) was 33 per cent. hydrolysed at a flow-rate of 7 ml per hour, and decan-%one 2,4-dinitrophenylhydrazone (6.0 pmoles) was 26 per cent. hydroly-sed at 8 ml per hour.The foregoing result suggested that, although a short chain carbonyl compound would react much more rapidly than benzophenone it might not react completely. Indeed, when a solution of 41 pmoles of acetone was percolated over the 2,4-dinitrophenylhydrazine column, the effluent had no more colour than a blank treated in the same way. It seems probable, therefore, that the values given by Schwartz and Parkes22 for the aliphatic monocarbonyl contents of various organic solvents do not include acetone, and it is not possible to say from the details given in their paper whether or not acetone is removed from solvents which have been percolated over a 2,4-dinitrophenylhydrazine - 66 per cent. phosphoric acid column. THE 2,4-DINITROPHENYLHYDRAZINE - 2 N HYDROCHLORIC ACID COLUMN- Gaddis et a1.N prepared 2,4-dinitrophenylhydrazones in 2,4-dinitrophenylhydrazine - 2 N hydrochloric acid, followed by extraction with carbon tetrachloride and with benzene ; both extracts were then washed with 2 N hydrochloric acid and with water.It appears from a302 PARSONS : QUANTITATIVE MICROANALYSIS OF CARBONYL COMPOUNDS [,4 nalyst, J701. 91 subsequent paper by Gaddis and Ellis35 that the 2,4-dinitrophenylhydrazones of formaldehyde, acetaldehyde, acetone and butanone are extracted by the benzene but not by the carbon tetrachloride, and these authors commented that the quantitative aspects of the procedure required further study. Begemann and de Jongll purified light petroleum, first by the method described by Van der Ven and de J~nge,~O and then by passage over a 2,4-dinitrophenylhydrazine - 2 N hydrochloric acid column, followed by distillation.Begemann and de Jong stated that a very small amount of acetone was still present. We have found that this column gives a 95 per cent. yield of 2,4-dinitrophenylhydrazone with acetone, and an 84 per cent. yield with diacetone alcohol. The column procedure is thus a considerable improvement over the simple extraction procedures, both for slow reactions and for reactions giving unfavourable equilibria. The use of benzene as the mobile phase would be unfavourable for slow reactions and would tend to strip the reagent from the column. However, its use by Gaddis and Ellis34 suggested how the 2,4-dinitrophenylhydrazine - 66 per cent. phosphoric acid column, which has several advantages over the 2,4-dinitrophenylhydrazine - 2 N hydrochloric acid column, might be modified to give an improved yield with acetone.METHOD REAGENTS- Cyclohexane and light petroZezlm-Cyclohexane for ultraviolet spectroscopy (as supplied by British Drug Houses Ltd. or Hopkin and Williams Ltd.) is used without further purification. For more exact work, light petroleum (b.p. 40" to 60" C, aromatic free, Carless, Capel and Leonard Ltd.) is rendered carbonyl free in the following manner. 350 ml of fuming nitric acid, density 1-51, and 350 ml of sulphuric acid, density 1.84, are added, with stirring, to 3.5 litres of light petroleum, contained in a 5-litre flask fitted with a reflux condenser. The mixture becomes warm, and stirring is continued overnight. The mixed acid is removed by aspiration, and the organic phase is washed22 twice with 700 ml of water, nine times with 700 1111 of 20 per cent.potassium hydroxide and twice with 700 ml of water. The washed solvent is distilled over 350ml of refined coconut oil (Van Den Berghs and Jurgens Ltd., Purfleet), and percolated firstly through a 3-5 x 22-cm column of 100 g of alumina (P. Spence, grade H) and then through a second column of alumina activated by heating at 800" C for 4 hours (a personal communication from H. J. Duin, H. W. A. E. Groeneweld and H. Van der Wel). In order to obtain a stable product it was found to be essential to remove both the carbonyl compounds and their precursors in the foregoing manner and to store in brown bottles in the dark.Benzene-A mixture of 2-5 litres of thiophene-free benzene (Carless, Capel and Leonard Ltd.), 12.5 g of 2,4-dinitrophenylhydrazine and 2.5 g of trichloroacetic acid are refluxed for 4 hours under a Dean and Stark trap. The mixture is then distilled with constant stirring and use of a double splash head. If the mixture is not stirred, the reagent bakes on to the sides of the flask and is carried over into the receiver to give a coloured distillate. The distillation is repeated as necessary. 2,4-DinitropIze~zyZ~~~drazine--C:ommercial 2,4-dinitrophenylhydrazine and 100 ml per g of N hydrochloric acid are refluxed for 30 minutes, and the solution is then filtered. The filtrate is made alkaline with ammonia and cooled. The purified 2,4-dinitrophenylhydrazine is collected and re-crystallised from 200 ml per g of methanol to give leaflets of m.p.195" C. 98 per cent. nezitval alumina-A 2-ml portion of hydrochloric acid of density 1-16, and 98g of alumina (P. Spence, grade H) are shaken together for 30 minutes and then allowed t o stand overnight in a stoppered flask. 1.8 per cent. ethanoZic potassiz~m hydroxide-Five potassium hydroxide pellets are dis- solved in 25 ml of re-rectified ethanol by shaking mechanically in an Erlenmeyer flask. The solution is filtered through a Whatman No. 541 filter-paper and used within 2 hours. The alumina should be used within 1 week. PREPARATION OF 2,4-DINITROPHENYLHYDRAZONES- Impregnate 10-0 g of analytical-grade Celite (Johns - Manville) with a solution of 2,4-di- nitrophenylhydrazine (500 mg, 2-5 pmoles) in 6 ml of 88 to 93 per cent.orthophosphoric acid of density 1.75, diluted with 4ml of water. Transfer this stationary phase to a column of 2.1-cm internal diameter, fitted with a B24 socket and cone and a sintered-glass plate,May, 19661 PARSONS : QUANTITATIVE MICROANALYSIS O F CARBONYL COMPOUNDS 303 containing cyclohexane and tamp it down to a height of 9.5 cm.22 Wash with 50 ml of benzene to remove some of the reagent together with residual impurities. However, at least 2.2 pmoles of 2,4-dini t rophen ylhydrazine will remain. Fit this column to a swan-necked column36 containing 35 g of 98 per cent. neutral alumina. Apply the sample (containing between 1 and 100 pmoles of carbonyl compounds) in 5 to 10 ml per g of cyclohexane to the celite column and elute with the same solvent at 20 ml per hour (total volume 175 ml), the 2,4-dinitrophenylhydrazones being retained on the columns.Up to 10 g of triglycerides and other non-polar materials may be present in the sample and are recovered in the eluate. Elute both columns with 175 ml of benzene and examine an aliquot in the spectrophoto- meter, either directly or after adding to an equal volume of 1.8 per cent. ethanolic potassium hydroxide and leaving to stand for exactly 10 minutes. Approximate values for the wave- length maxima and extinction coefficients under neutral and basic conditionslg 335 (A. M. Parson, unpublished results) are given in Table 1’1. TABLE 17 WAVELENGTH MAXIMA AND EXTINCTION COEFFICIENTS OF 2 ,4-DIN ITROPHENYLHY DRAZONES 3,4-Dinitrophenylhydrazone h ~ ~ l l c , mp fniax.xmax. mP Emax. KOH - C,H,OH, Allcan-2-ones . . . . . . 36 1 21,500 43 1 22,200 Alkanals . . . . . . 354 21,000 430 20,900 Alk-2-enals . . . . .. 372 27,900 460 30,000 Allra-2,4-dienals . . .. 388 36,000 480 41,000 When the various classes of carbonyl compounds are present together, their total concentrations may be obtained29 with c2H50H = 21,000, or the separate concentrations of alkanals, alk-2-enals and alka-2,4-dienals may be obtained by measuring the absorptions at 430, 460 and 480mp and applying the equations derived by Keith and Cnfor- tunately, the absorption maxima of alkan-%one 2,4-dinitrophenylhydrazones are too close to those of alkanal 2,4-dinitrophenylhydrazones to permit separate determination by this means.However, by taking additional readings after 2 hours, further information may be obtained on this point, because only acetone and alkanal 2,4-dinitrophenylhydrazones fade appreciably under these condition^.^^ 935 For detailed analysis, the remainder of the benzene eluate may be fractionated into classes by adsorption chromatography on dry paper,37 magnesium o ~ i d e , ~ ~ ? ~ ~ or 92 per cent. alkaline alumina.40 Separation according to chain length may then be achieved by partition chromatography on paper,41 942 thin layers of Kieselguhf13+ or column^.^^^^^ 947 Xumerous thin-layer adsorption systems have also been d e ~ c r i b e d . ~ ~ y ~ 9 ~ ~ 349 I t is necessary to repeat the whole procedure, with omission of the sample, in order to allow for residual carbonyl compounds in the solvents and for artifacts produced on the columns.I t is found, for example, that two pigments < 320 mp) are obtained which are less polar than ketone 2,4-dinitrophenylhydrazones on adsorption chromatography. Winter et al. have noted several such artifacts50 and have pointcd outs1 that 2,4-dinitroaniline could arise from the reaction between acetoin and 2,4-dinitrophenylhydrazine. For polar compounds when the 2,4-dinitrophenylhydrazine column is eluted with benzene, the 2,4-dinitrophenylhydrazones of polar carbonyl compounds are transferred to the 98 per cent. neutral alumina column, together with some unchanged reagent. The alumina column may subsequently be eluted with benzene containing between 1 and 10 per cent of ethanol and aliquots of the solutions examined in the spectrophotometer as before. The absorption maxima for acctoin5l and for the two 2,4-dinitrophenylhydrazones of d i a ~ e t y l ~ ~ are shown in Table VII.TABLE iTII WAVELENGTH MAXIblA AND EXTINCTION COEFFICIENTS OF DINITROPHENYLHYDRA4ZONES OF ACETOIN AND DIACETYL 2,4-Dinitrophenylhydrazone X ~ ~ ~ ~ f o n n , m p Ernax. XI,,,. mCc Emax. Acetoin . . . . .. 357 33,000 433 Diacetyl-mono . . .. 351 29,100 50 1 38,000 Diacetyl-bis . . .. .. 393 47,000 556 54,000 KOH - C,H,OH, 435 40,000 - - (shoulder)304 PARSONS : QUANTITATIVE MICROANALYSIS OF CARBONYL COMPOUNDS [Analyst, Vol. 91 The reagent 2,4-dinitrophenylhydrazine has A::: - CzH50H = 360 mp, E = 1,860. There are, however, more satisfactory methods for determining acetoin and diacetyl which do not involve preparation of 2,4-dinitrophenylhydrazones.53 755 RESULTS Application of the method to model compounds gave good yields of products; even acetone gave 84 per cent.ANALYSIS OF BUTTER FAT The method also worked well with the molecular distillates of 200 g of butter fat obtained with a 2-inch wiped wall molecular still (Edwards High I'acuum Ltd.), 10-20 p (b.p. 160" C), fitted with liquid nitrogen cooled traps. The values (Table VIII) agreed closely with those obtained after subsequent chromatography. A 10-g sample of New Zealand butter fat was also examined directly, but in this case we were unable to separate the classes by chromatography. Schwartz et al. claim56 that this separation can be performed after a preliminary fraction on a partition column but, as can be seen from the last column in Table VIII, the quantity of involatile carbonyl compounds in this fat is so large that it would place great demands on any technique to require it to carry out a separation at this stage.TABLE 1~111 CARBOKYL CONTEST OF BUTTER FAT AND BUTTER FAT VOLATILES New Zealand Danish New Zealand volatiles, volatiles, butter fat, pmoles per kg pmoles per kg pmoles per kg Alkan-2-ones and alkanals . . 2.2 8-1 730 A1 k-2-enals . . . . . . 1.0 1.0 130 Alka-2,4-dienals . . . . . . 0.4 0.6 30 More recently, Schwartz and co-workers5' have shown that keto-glycerides can be removed from butter fat by adsorption chromatography, and that these glycerides amount to no less than 0.045 per cent. by weight (550 pmoles per kg as keto-tripalmitin).We have removed polar material from 20 g of Kew Zealand butter fat in this way, and have converted the remaining carbonyl compounds to 2,4-dinitrophenylhydrazones by the method described above. The carbonyl compounds still amounted to over 120pmoles per kg and, judging by their behaviour on subsequent chromatography, consisted largely of keto-glyceride 2,4-dinitrophenylhydrazones. The polar fraction was also examined in the same way, except that a strongly acid (2 N perchloric acid) 2,4-dinitrophenylhydrazine column was used in order to decompose acid labile carbonyl precursors.5s The 2,4-dinitrophenylhydrazones gave two bands on mag- nesium oxide,38 939 in similar positions to those given by pentanal 2,4-dinitrophenylhydrazone and by dec-2-enal 2,4-dinitrophenylhydrazone. The spectral properties given in Table IX, however, indicate an alkanone and an alkanal fraction (Amax.= 430 mp in each instance, and the latter fading rapidly) amounting to 165 pmoles per kg and 36 pmoles per kg, respectively. TABLE TX WAVELENGTH MAXIhfA AND FADING OF L)INI?'ROPHENYLHYL)RXZONES OF POL-iR CARBONYL COMPOUNDS FROM BUTTER FAT Fading, Fraction GZ:'& Amax., ml* per cent. 1 362 430 2 0 540 3.4 2 360 430 15 635 24 KOH - C,H,OH It therefore appears that, despite the use of a strongly acid reaction column, the carbonyl compounds still contained polar groups and were probably keto-glycerides and aldehydo- glycerides, respectively. The author thanks Dr. I. D. Morton for his interest and encouragement, and Mr.D. J. Moore and others for technical assistance.May, 19661 PARSONS : QUANTITATIVE MICROANALYSIS OF CARBONYL COMPOUNDS 305 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. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 40. 50. 51. 52. 53. 54. 55. 56. 57. 68. REFERENCES Winter, M., Stoll, M., Warnhoff, E. W., Greuter, F., and Buchi, G., J . Fd Sci., 1963, 28, 554, Lea, C. H., and Swoboda, P. A. T., Chem. & Ind., 1958, 1289. Meijboom, P. W., J . Amer. Oi2 Chem. SOC., 1964, 41, 326. Hoffman, G., Ibid., 1961, 38, 1. Forss, D. A., Dunstone, E. A., and Stark, W., J . Dairy Res., 1960, 27, 211. Commonwealth of Australia, Commonwealth Scientific and Industrial Research Organisation, Forss, D.A, J . Dairy Sci., 1964, 47, 245. Braun, P. A, and Mosher, W. A., J , Amer. Chenz. SOC., 1958, 80, 3048. Chen, P. S., Analyt. Chem., 1959, 31, 296. Haverkamp Begemann, P., and de Jong, K., Recl Trav. Chim. Pays-Bas Betg., 1959, 78, 275. Dana Johnson, G., J . Amer. Chem. SOC., 1951, 73, 5888. Shriner, R. L., and Fuson, R. C., “Identification of Organic Compounds,” Third Edition, John Buss, C. D., and Mackinney, G., J . Amev. Oil Chem. Soc., 1955, 32, 487. Henick, A. S., Benca, 54. F., and Mitchell, J . H., Ibid., 1954, 31, 88 and 447. , Ibid., 1956, 33, 35. Skeriett, E. J., and Baker, E. A., Analyst, 1959, 84, 376. Keith, R. W., and Day, E. A., Ibid., 1963, 40, 121. Pool, M. F., and Klose, A. A., J . Amer. Oil Chem. Soc., 1951, 28, 215.Lea, C. H., and Jackson, H. A. F., Chem. & Ind., 1964, 1429. Forss, D. A., Pont, E. G., and Stark, W., J . Daivy Res., 1955, 22, 91. Schwartz, D. P., and Parks, 0. W., Analyt. Chem. 1961, 33, 1396. Schwartz, D. P., Haller, H. S., and Keeney, M., Ibid., 1963, 35, 2191. Horikx, &I. RI. J , Appl. Chem. Lond., 1964, 14, 50. Chang, S. S., and Watts, €3. M., “Flavour Chemistry Symposium,” Campbell Soup Co., Camden, Loftus Hills, G., and Thiel, C. C . , J . Dairy Res., 1946, 14, 340. RlcDowell, A. K. R., Ibid., 1964, 31, 221. Bennett, A., May, L. G., and Gregory, R., J . Lab. Clin. Med., 1951, 37, 613. Stitt, F., Seligman, R. B., Resnik, F. E., Gong, E., Pippen, E. L., and Forss, D. A., Spectrochim, Van der Ven, B., and de Jonge, A. P., Red Trav. Chim. Pays-Bas Belg., 1957, 76, 169.Mendelowitz, A., and Riley, J . I?., Analyst, 1953, 78, 704. Jones, L. A., Holnies, J. C., and Seligman, R. B., Analyt Chem., 1956, 28,191. Belcher, R., and Fleet, B., J . Chem. SOC., 1963, 5720. Gaddis, A. M., Ellis, R., and Currie, G. T., Fd Res., 1959, 24, 283. Gaddis, A. M., and Ellis, R., Ibid., 1959, 24, 392. Bush, I. E., “The Chromatography of Steroids,” Pergamon Press, 1961, p. 149, Gaddis, A. M., and Ellis, R., A n a l y f . Chem., 1959, 31 870. Schwartz, D. P., Parks, 0. W., Keeney, M., Ibid., 1962, 34, 669. Schwartz, D. P., and Parks, 0. W., Microchem. J . , 1963, 7, 403. Van der Yen, B., Haverkamp Begemann, P., and Schogt, J . C. M., J . Lipid Res., 1963, 4, 91. Horner, L., and Kirmse, W., Justus Liebigs A n n l n Chem., 1955, 597, 48. Klein, F., and de Jong, I<., Red Trav. Chim. Pays-Bas Belg., 1956, 75, 1285. Badings, €1. T., and M’assink, J . G., Xeth. Afilk G. Dairy J . , 1963, 17, 132. Urbach, G., J. Chromat., 1963, 12, 196. Kramer, P. J . G., and Van Duin, H., R e d Trav. Chim. Pays-Bas Belg., 1964, 73, 63. Corbin, E. )I., Schwartz, D. P., and Keeney, >I., J. Chvowiat., 1960, 3, 322. Freytag, W., Fefie Seifen, 1963, 65, 603. Radings, H. T., J . Amer. Oil Chem. SOC., 1959, 36, 648. Ilhont, J. H., and de Rooy, C., .4$zaZyst, 1961, 86, 74. Winter, M., and Sundt, E., Helv. Chim. A c f a , 1962 45, 2195. Winter, RI., and Enggist, P., J . Fd Sci., 1963, 28, 685. Jones, L. A,, and Kinney Hancock, C., J . An7er. Chein. Soc., 1960, 82, 105. Pien, J., Baisse, J . , and Martin, R., Lait, 1937, 17, 675. Kimphorst, L. C. E., Kruisheer, C. I., 2. Unters. Lebeizsmitfel, 1937, 73, I . Owades, J . L., and Jakavac, J. L4., Proc. Amer. SOC. Brew. Chem., 1963, 22; A n a l y f . Abstr., 1964, Schwartz, D. P., Haller, H. S., and Keeney, M., A n a l y f . Chem., 1963, 35, 2191. Parks, 0. W., Keeney, M., Katz I., and Schwartz, D. P., J . Lipid Res., 1964, 5, 232. Schogt, J. C. M., Haverkamp Begemann, P., and Koster, J., Ibid., 1960, 1, 232. -, Ibid., 1962, 39, 439. Division of Dairy Research, Annual Report 1961-1962, Melbourne, 1962, p. 10. Wiley and Sons Inc., 1948, p. 171. -- ___ -- New Jersey, 1961, p. 145. Acta, 1961, 17, 51. 3363. R w e i v d Novdmher 16th. 1965.