March, 19531 EXTRACT AND MEAT STOCKS 135 The Determination of Carbonyl Compounds by Semicarbazide and Hydroxylamine With Special Reference to Fatty-Acid Oxidation Products BY A. J. FEUELL AND J. H. SKELLON (Presented at the meeting of the Society on Wednesday, October l s t , 1952) A volumetric method of determining aldehydes and ketones with semi- carbazide has been devised. It is suitable for carbonyl compounds that readily precipitate semicarbazones, but it is not applicable t o oxidised fatty-acids and esters. For estimating the carbonyl groups in the products of oxidation of fatty acids and esters by gaseous oxygen, a modified hydroxylamine method is described. The modified method is especially useful for coloured samples. DURING investigations into the oxidation of monoethenoid fatty acids and esters the need arose for a convenient method of determining carbonyl compounds of the keto- or ketol-acid type, which are known to be formed during these 0xidations.lJ,3~~,~ In an earlier study of an analogous problem, Marks and MorrelP concluded that the phenylhydrazine method of Maclean7 was often unreliable, but the method of Ellis8 was suitable.Ellis’s method involves the use of rather specialised apparatus, which is not always convenient when, as here, carbonyl determinations are required only occasionally and irregularly. For such occasional deter- minations a straightforward volumetric method seemed to be preferable. Volumetric methods based on hydroxylamine are in common use,9 and a selective list of references to the various modifications has been given by Maltby and Primavesi.lO Xever- theless, it seemed that there would be advantages in devising an alternative method that would permit the isolation of some derivative of the carbonyl compound simultaneously with its quantitative determination, as an aid to characterisation.No volumetric methods based on 2 :4-dinitrophenylhydrazine could be traced, although several gravimetric and colorimetric136 FEUELL AND SKELLON THE DETERMINATION OF CARBONYL [Vol. 78 procedures have been described.ll ,12 ,13 ,14 ,15 Of the other common carbonyl reagents, semi- carbazide seemed worth investigating, as volumetric methods for its estimation are a ~ a i 1 a b l e . l ~ ~ ~ ~ J8 J ~ ~ J ~ ~ ,22 One of the simplest of these methods is that of Smith and which involves direct titration with standard iodate.Smith and Wheat identified the carbonyl compounds by finding the equivalent weights of the semicarbazones. For this purpose they prepared the derivative and titrated a weighed amount of it, but did not attempt to determine the percentage of carbonyl group in compounds or unknown mixtures. USE OF SEMICARBAZIDE In the first part of our work an attempt was made to extend the method of Smith and Wheat22 to the direct determination of the carbonyl group in a given material. But an important modification was made in the titration procedure. Smith and Wheat used the iodine monochloride method,23 but it has been stated2* that Lang's iodine ~yanidemethod~*,~~126 is preferable for hydrazine and its congeners, although its use for semicarbazide has apparently not been reported.Experiment showed that semicarbazide could be titrated quantitatively and rapidly by Lang's method. DETERMINATION OF SEMICARBAZIDE- A 0-20 to 0.25-g sample of semicarbazide hydrochloride was dissolved in 25 to 30 ml of water, 20 ml of 5 N hydrochloric acid and 5 ml of 10 per cent. potassium cyanide solution were added and the solution was titrated with 0.05211 potassium iodate solution. When the solution became light brown (at about 80 per cent. of the complete titre), starch was added and titration was continued until the starch colour disappeared for at least 30 seconds. (1 ml of 0.05 M potassium iodate solution = 0.005577 g of semicarbazide hydrochloride.) As some hydrocyanic acid is evolved during the reaction, it is safer to titrate in a Biichner flask, with the side-arm connected by a flexible tube to a soiirce of gentle suction.METHOD FOR PURE CARBONYL COMPOUNDS- Reagent-Dissolve 3 g of semicarbazide hydrochloride and 3 g of sodium acetate crystals in 100 ml of water. This solution is fairly stable, but a slight loss of strength is unimportant, because a blank determination is always made. The usual titre is about 6 ml of 0.05 ill iodate per millilitre of reagent. Procedure-Dissolve 0.2 g of carbonyl compound (for an expected carbonyl content of 20 per cent. ; otherwise in proportion) in 10 ml of water or other solvent (see below, p. 137), add exactly 10ml of reagent and set aside. Filter off the precipitated semicarbazone and TABLE I 1':FFECTS OF TIME AND EXCESS OF REAGENT ON FORMATION OF SEMICAKBAZONES Salicylaldehyde Benzaldehj-de A v 7 -7 /-------------h---. Time allowed >CO found >CO found >CO found >CO found >CO found for precipitation, with E = 1.26, with E = 1-41, with E = 1.21, u-ith IZ = 1.,58), with E == 2.42, 10 22.1 22.2 25.5 2 5 6 - 20 22.2 22.2 25.4 25.5 - 30 22.1 22.4 25.4 25.6 6'4.5 60 22.1 22.4 25.5 25.7 - O/ /O 0 minutes % /O % i0 O? >CO calculated, "/b .. 22.9 26.4 SOTE--E is the molar ratio of semicarbazide hydrochloride to aldehyde. wash the flask and precipitate with two separate 5-ml portions of water, passing the rinsings through the filter and adding them to the filtrate. To the filtrates, whose combined volume is 301111, add 5 iV hydrochloric acid and potassium cyanide, and titrate as above.For the blank, omit the carbonyl compound and dispense with the filtration, but add 10ml of water before adding the acid and cyanide. Each millilitre difference between titrations is equi\-alent to 0.0014 g of ;-CO. The choice of solvent and the precipitation period are discussed below. Results-Preliminary tests were made with freshly-distilled salicylaldehyde and benz- aldehyde. The first factors studied were the length of time necessary for precipitation andMarch, 19531 COMPOUNDS BY SEMICARBAZIDE AND HYDROXYLAMINE 137 the effect of excess of reagent. These compounds were always dissolved in 50 per cent. acetic acid and, as an equal volume of aqueous reagent was added, the final acid content of the reaction mixture was about 25 per cent.; the significance of this is mentioned below.The percentage of >CO found is shown in Table I. I t is seen that with these substances useful results can be attained with a precipitation time of only 10 minutes and a 20 to 60 per cent. excess of reagent. APPLICATION TO KETOHYDROXYSTEARIC ACID- The method was next applied to a compound typical of those encountered in oxidation researches, namely, ketohydroxystearic acid. A pure specimen was prepared by King's method,l but the 9:lO and 10:9-isomers were not separated, as this was unnecessary for the purpose in hand. The purity of the specimen, found by titrating an alcoholic solution with standard alkali, was 99.2 per cent., corresponding to a carbonyl content of 8.84 per cent. (calculated, 8-91 per cent.).King1 showed that the semicarbazone of this acid could be precipitated from 60 per cent. alcohol, and these conditions were used for the determination. Procedure-Dissolve 0.4 g of the ketol acid in 16 ml of alcohol and add 10 ml of reageht. Precipitation of the semicarbazone begins within about half-an-hour and is completed overnight. Filter, rinse, and titrate the filtrate as previously described; make a blank determination simultaneously. Three separate determinations gave 8.75,9.06 and 8.87 per cent. for the carbonyl content. EFFECTS OF SOLVENT AND TEMPERATURE ON STABILITY OF SEMICARBAZIDE SOLUTIONS- Trials were made to determine the effect of temperature, as it seemed likely that it might occasionally be desirable to precipitate at other than room temperature.Erratic results under certain conditions were traced to unequal changes in titre between sample and blank, owing to decomposition of the semicarbazide. These changes were investigated systematically by setting aside blanks containing a fixed amount of the semicarbazide reagent in water or in one of three different strengths of acetic acid or alcohol for various times at three widely different temperatures. The results in Table II show the differences between initial and final titrations for pairs of similar blanks. It must be remembered that in practice decom- position will occur in both sample and blank; if these changes are equal there will be no error, but, owing to differences in concentration arising from the removal of semicarbazide in the precipitate, the amounts of decomposition will not generally be identical. The titration errors will, however, be less than the differences shown in Table 11, which can be regarded as the maximum.TABLE I1 EFFECTS OF TIME AND TEMPERATURE ON DECOMPOSITION OF SEMICARBAZIDE IN DIFFERENT SOLVENTS Titre differences after setting aside at Concen- I A 1 tration 0" C 16°C 16°C 16°C 45°C 45" C 45°C of for for for for for for for % v/v ml ml ml ml ml ml ml Solvent solvent, 8 hours, 2 hours, 4 hours, 8 hours, 2 hours, 4 hours, 8 hours, Water .. 100 0.05 0.00 0.10 0.15 0.05 0.15 0.15 Aceticacid . . 75 0.45 0.30 1.10 1-75 7-35 12.55 - 50 0.15 0.30 0.86 1.60 5-90 10.40 - 25 0.05 0.20 0.25 0.60 2-85 6.25 - Alcohol . . 75 0.50 0-55 0.50 0.55 1-15 1-25 2-15 60 0.35 0.25 0.30 0.40 0.90 0.75 1-05 25 0.05 0.10 0.16 0.20 0-40 0.30 0.40 It is clear that prolonged heating is undesirable when acetic acid is used as the solvent; but, in determinations made as above, in which the final concentration of acid in blank and sample solutions is about 25 per cent.and the proctdure is completed in 30 minutes or less, decomposition causes only a slight error. Water or dilute alcohol can be safely used for fairly long precipitation periods at room temperature, and at low temperatures any of the solvent mixtures appears to be suitable.138 FEUELL AND SKELLON THE DETERMINATION OF CARBONYL [Vol. 78 APPLICATION TO OXIDISED MATEKIALS- Monoethenoid fatty-acids and esters, after catalytic autoxidation for several hours at temperatures above 100" C, devclop reducing properties, which are apparently due to the formation of various types of carbonyl compound^.^ ,3 The application of the semicarbazide method to oxidised oils possessing marked reducing properties was disappointing, as semi- carbazones could not be precipitated; the method consequently failed.It was difficult to find a suitable solvent; the oxidised products are not soluble in 50 per cent. acetic acid, and, in view of the results shown above, a solution of any greater strength could not be used. Although soluble in 95 per cent. alcohol, the oils are precipitated as soon as the added water reduces the alcohol content to less than about 85 per cent.; hence the 60 per cent. alcohol conditions suitable for a pure ketol acid are inapplicable. Pyridine, which Hopper27 showed to be a useful solvent for preparing semicarbazones, proved unsuccessful. Moreover, its basicity caused a further complication, which would have made its use doubtful in any event ; for although additional concentrated hydrochloric acid was added before titration, a buffering effect seemed to retard the iodate reaction, and the end-point, which showed con- siderable drift, was reached slowly.Immiscible solvents such as chloroform were also investigated, but, even after shaking and setting aside for long periods, semicarbazones were not precipitated. U S E OF HYDKOXYLAMINE In view of the inapplicability of the semicarbazide method to oxidised materials it was decided to use hydroxylamine, and the recent method of Maltby and Primavesilo was selected.Careful matching of the bromophenol blue indicator at the end-point is essential in this method, but it was at once found that the procedure as published was not reliable for oxidised fatty compounds, since these were generally reddish-yellow or brown; they changed the tint of the indicator so much that it could not be properly matched against the standard. In addition, matching was found to be rather difficult in artificial light. As a result of further work, Maltby and Primavesi's method was modified in two ways, as follows- (1) By the use of two reagent solutions instead of a single one, so that interference from colour inherent in the sample is eliminated and, to a lesser extent, variations from neutrality in the sample are simultaneously compensated. Both these improve- ments can be achieved by using only about 25 per cent.(by weight) more of sample than is required for a determination by the original method. (2) By making the matching procedure equally easy in either daylight or artificial light. The method devised is as follows. METHOD- Dissolve 20 g of hydroxylamine hydrochloride in 100 ml of water, make up to 1 litre with 95 per cent. alcohol, and add 25 ml of a 0.2 per cent. alcoholic solution of bromophenol blue. Heat on a water-bath for 30 minutes to react with any carbonyl compounds in the alcohol, cool, and adjust the colour as seen in the bulk to the neutral dichroic green-red with 2 N alkali. Prepare as for reagent A but use 5 g of sodium or ammonium acetate instead of the hydroxylamine; this imparts a slight buffering action.This solution need not be heated. Adjust its colour with 2 N acid until it matches that of reagent A. The exact tint is not critical within reasonable limits, and it is immaterial whether reagent B is adjusted to match reagent A or vice veysa. The solutions keep well, but should be re-matched before they are used. Prepare 4 volumes of reagent A for each volume of reagent B. Matching-Match in paired boiling or Nessler tubes instead of in conical flasks as in the original method, using about 20 ml of each solution. Hold the tubes vertically side-by-side and view horizontally against a piece of opal glass illuminated from behind by a fairly strong light. In the small thickness of liquid viewed, the indicator colour is a distinct green in contrast to the dichroic green-red displayed by a larger bulk in a flask, and slight differences in tint are more easily perceptible.By day, reflected sunlight from walls or a strong north light are equally suitable for illuminating the opal glass screen, whilst at other times a 60, 75 or 100-watt bulb placed about a foot behind the screen is adequate. Moreover, slightly turbid solutions, which are difficult to compare, Hydroxylamine reagents-Reagent A. Reagent B. This technique is both convenient and accurate.March, 19531 COMPOUNDS BY SEMICARBAZIDE AND HYDROXYLAMINE 139 even in tubes, against ordinary backgrounds, can be matched against the opal screen, the turbidity apparently disappearing. Procedure-Dissolve the sample (corresponding to 0.002 to 0.004 gram-molecules of carbonyl compound) in 25 ml of water or alcohol, and add 20 ml of this solution to a flask containing 60ml of reagent A.Add the remaining 5ml to 15ml of reagent B in one of a pair of tubes. The colour of the indicator in both reagents is thus equally affected by dilution and inherent sample colour. Set the sample aside for 5 to 10 minutes and then titrate it with 0.2 N alkali to match reagent B, a proportionate amount of water (one-quarter of the titration volume) being added to reagent B. With a little practice it is possible to titrate the liquid in the flask to a point just on the acid side of neutrality, as seen by the colour in bulk. Pour some of the mixture into the other tube and compare the two as described. If the liquids are not matched return the contents of the tube to the flask and cautiously continue to titrate, transferring a suitable volume to the tube from time to time for exact comparison.When the solutions are matched, TABLE I11 RESULTS BY PROPOSED METHOD APPLIED TO OXIDISED MOKOETHENOID ESTERS >CO found at Sample f A \ 55O c, 85" C, 120" c, % % % Oxidation products of ethyl oleate . . .. 0.64 1.86 1.95 ,9 n-propyl oleate . . 0.78 1.91 1.88 ,7 n-butyl oleate .. 1.09 1.99 1.83 set them aside for a further 5 to 10 minutes and compare them again, as some compounds react slowly. Finally, loosely stopper the two flasks, place them on a water-bath for 20 minutes, cool and titrate if necessary to ensure final matching; this may be necessary for compounds that do not react in the cold. Water must always be added to reagent B in the same pro- portion, one-quarter of the titration volume, before final matching.A slight departure from neutrality, such as is given by many organic acid samples, causes no serious interference. Strongly acid or alkaline samples are dissolved in water or alcohol, a drop of bromophenol blue solution is added and the colour adjusted approximately to the neutral tint with alkali or acid. The volume is then made up to 25ml and the procedure carried out as described above. An improvement in technique,28 which allows more rapid manipulation, is effected by titrating the sample in a standard-joint flask having a long narrow neck equal in diameter to the tube containing reagent B. For matching, a stopper is inserted, the flask inverted, and the columns of liquid in the neck and tube are compared as usual.Results-Since hydroxylamine is well established for estimating the commoner aldehydes and ketones, trials with the modified method were mainly concerned with the effects of colour in the sample. Satisfactory results were obtained with salicylaldehyde and anisaldehyde each tinted with brown dye; mean percentages of >CO group found were, rspectively, 22.7 and 20.1 per cent. (calculated, 22.9 and 20.6 per cent.). Ketohydroxystearic acid, mixed with various amounts of brown dye in imitation of the oxidation products for which the method was ultimately to be used, was then tried. Even with highly coloured samples there were no difficulties with the modified method. The carbonyl content found in a series of runs with 0-5 to 0.6 g varied from 8.88 to 9-13 per cent. (calculated, 8.91 per cent.).An even more representative test was made by dissolving some of the ketol acid in oleic acid and again adding brown dye, thus closely simulating a possible oxidation product. The mixture contained 15-4 per cent. w/w of ketol acid (equivalent to 1-37 per cent. of >CO). Three separate determinations indicated 1.42 to 1-45 per cent. of >CO, which corresponded to 15.9 to 16.3 per cent. w/w of ketol acid. Application of the method to oxidised monoethenoid esters has enabled the influence of temperature and constitution on formation of carbonyl groups to be studied. Some typical results are indicated in Table 111, which shows the percentage of >CO found in three homologous esters autoxidised at different temperatures in presence of a uranium catalyst.Subsequent work on oxidised materials has shown that the method can be scaled down if necessary and applied to smaller samples with the use of 0.1 N alkali, a match to within140 FEUELL AND SKELLON Pol. 75 acid.28 1. 2. 3. 4. 6. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 26. 26. 27. 28. one drop still being readily attainable. Occasionally it is desirable to titrate with alcoholic alkali, as certain complex products, such as those from oxidised butyl oleate, are precipitated if appreciable amounts of aqueous alkali are added; a proportionate volume of alcohol instead of water is then added to reagent B before matching. The method has been found useful for studies of the carbonyl content of the various fractions obtained in the separation of the complex end-products of oxidised fatty-acid esters (being prepared for publication elsewhere).It has also been used to follow the steady decrease in carbonyl content occurring during the thermal catalytic autoxidation of ketohydroxystearic REFERENCES King, G., J . Chem. SOC., 1936, 1788. Skellon, J. H., Ibid., 1948, 343. Skellon, J. H., and Thruston, M. N., Ibid., 1949, 1626. Ellis, G. W., Ibid., 1950, 9. -, Biochem. J., 1950, 44, 129. Marks, S., and Xlorrell, R. S., Analyst, 1931, 56, 508. Maclean, I. S., J . Biol. Chem., 1913, 7, 611. Ellis, G. W., J . Chem. SOC., 1927, 848. Analytical Methods Committee, “The Determination of Aldehydes other than Citronellal,” A nulyst, Maltby, J. G., and Primavesi, G. R., Ibid., 1949, 74, 498. Iddles, H. A., and Jackson, C. E., Ind. Eng. Chem., Anal. Ed., 1934, 6, 454. Iddles, H. A., Low, A. W., Rosen, B. D., and Hart, R. T., Ibid., 1939, 11, 102. Houghton, R. E., Amer. J . Plzarm., 1934, 106, 62. Lappin, G. R., and Clark, L. C., Anal. Chem., 1951, 23, 541. Pool, M. F., and Klose, A. A., J . A.mer. Oil Chem. SOC., 1951, 28, 216. Hovorka, V., Coll. Trav. Chim.. Tchdcosl., 1931, 3, 285. Horlay, V., J . Pharm. Chim., 1936, 23, 199. Veibel, S., Ibid., 1936, 24, 499. Bartlett, P. D., J . Amer. Chem. SOC., 1932, 54, 285i3. Miller, C. 0.. and Furman, N. H., Ibid., 1937, 59, 161. Smith, S. G., J . Chem. SOC., 1937, 1925. Smith, G. B. L., and Wheat, T. G., Ind. Eng. Chem., Anal. Ed., 1939, 11, 200. Jamieson, G. S., “Volumetric Iodate Methods,” Chemical Catalog Co., New York, 1926. Bottger, W., and Oesper, R. E., “Newer Methods of Volumetric Chemical Analysis,’’ Chapman Mitchell, A. D., and Ward, A. M., “Modern Methods in Quantitative Chemical Analysis,” Longmans, Lang, R., 2. anorg. Chem., 1932, 122, 332. Hopper, I. V., J . Roy. Tech. Coll. Glasgow, 1929, 2, 52. Parkinson, T. L., private communication. 1934, 59, 105. & Hall Ltd., London, 1938. Green & Co., Ltd., London, 1932. ACTON TECHNICAL COLLEGE HIGH STREET, ACTON LONDON, W.3 August 22nd, 1962