Analyst March 1983 VoZ. 108 $$. 329-334 329 Quantitative Determination of Carbonyl Compounds in Rendering Emissions by Reversed-phase High-performance Liquid Chromatography of the 2,4- Di nitrop henyl hyd razones Herman R. Van Langenhove Marc Van Acker and Niceas M. Schamp Laboratoviunz VOOY Ovganische Scheikundr Faculteit van de I_airdboitz~~zuetenscha~p~~~~ I~ijksiiItivevsiteit-Gent, Coupure L i n k s 653 B-9000 Gent Belgium A simple and rapid method for the determination of volatile carbonyl com-pounds in air has been developed. The method is applied to the quantitation of C,-C aldehydes and acetone in rendering emissions. Carbonyl compounds are sampled by absorption in a 2 x lW4 M 2,4-dinitrophenylhydrazine solution a t pH 1 and the resulting hydrazones are extracted with 2,2,4-trimetliylpentane concentrated and analysed by HPLC on a 10-pni RP-C, column with a water - acetonitrile gradient as the eluent.The hydrazones are then spectrophotonietrically detected a t 356 nni . The micro-scale conversion of carbonyls into 2,4-dinitrophenylhydrazones is investigated the separation of hydrazones is improved and the sampling conditions are tested in order to achieve quantitative sampling a t an air flow-rate of 1 1 min-l. Quantitation is possible for concentrations as low as 15 p.p.b. (formaldehyde) and 2 p.p.b. (nonanal). The over-all coefficient of variation (taken over sampling conversion and analyses) is less than loo/. Keywords Carbonyl quantitation ; 2,4-dinitropIienyllzydrazine derivatisntion ; reversed-phase HPLC analysis ; rendering plant enaissions The lower aliphatic aldehydes formaldehyde acetaldehyde and acrylaldehyde are well known in atmospheric pollution chemistry as important intermediates in photochemical smog forma-tion.Higher aldehydes on the other hand are more associated with flavour chemistry be-cause they are secondary oxidation products of fatty acids. Aliphatic aldehydes as a whole are emitted by animal rendering plants and these compounds together with lower aliphatic acids amines and sulphur compounds contribute to the odour nuisance commonly accompany-ing rendering activities In order to evaluate the relative importance of different groups of malodorants a method for the quantitation of aldehydes a t odour threshold levels (parts per lo9 p.p.b.) has been developed.The derivatisation of aldehydes into 2,4-dinitroplienylhydrazones followed by either gas chromatographic or high-performance liquid chromatographic (HPLC) analysis of the deriva-tives is a widely proposed method.1-6 Derivatives of isomeric carbonyl compounds have been separated by gas chromatography on capillary columns coated with OV-17,OV-101 and SF 96.' However double peaks due to syn- and anti-isomeric forms may hamper the determination and quantitation of compounds in unknown mixtures by altering the retention data.lV2 These workers summarised different normal and reversed-phase HPLC analyses of hydrazones that were in use up to 1979. They also reported the separation of 2,4-dinitrophenylhydrazones of carbonyls up to C6 by reversed-phase HPLC with a LiChrosorb RP-18 column and acetonitrile - water as the mobile phase.Other reversed-phase separations have been reported with either isocratic4 or linear solvent programming e l ~ t i o n . ~ The latter technique allows the separation of C,-C, linear aliphatic aldehydes within 10 min. The micro-scale conversion of propanal with 2,4-dinitrophenylhydrazine was investigated by Selini.6 According to this worker a two-phase system aqueous reagent -2,2,4-trimethylpentane was needed to achieve quantitative conversion. In this paper results of studying the micro-scale reactivity of C,-C9 carbonyls and 2,4-dinitrophenylhydrazine (2,4-DNPH) the trapping efficiency of aldehydes in 2,4-DNPH reagent and analyses of the derivatives obtained by HPLC are described. Results of the quantitation of carbonyl compounds in rendering emissons are also reported.Kuwata et aZ.3 reviewed the advantages of HPLC for hydrazone separation 330 Reagents The carbonyl compounds investigated were obtained from various commercial sources. Owing to trimer formation the purification of the aldehydes is essential if reliable results are to be obtained. Formaldehyde acetaldehyde and propanal were purified by distillation. Higher aldehydes were purified by preparative gas chromatography. The 2,4-dinitrophenyl-hydrazine reagent consists of 0.125 g of 2,4-DNPH in 100 ml of 6 N hydrochloric acid. The reagent was extracted twice with 50 ml of carbonyl-free 2,2,4-trirnethylpentane and the purified reagent was kept covered with a layer of the same solvent. The 2,2,4-trimethyl-pentane (1 1) was refluxed with 0.5 g of 2,4-DNPH and 50 ml of 6 N hydrochloric acid for 2 h and was then distilled from the reaction mixture.Hydrazone standards were prepared and purified by use of standard methods.' Apparatus and Chromatographic Conditions A Varian 8500 liquid chromatograph and a Varian spectrophotometer (UV-VIS Model 635) operating at a wavelength of 356 nm were used to perform the analyses. Samples were injected by use of an injection valve (Valco CV-6-UVPa N60) together with a 10-pl loop. The column used was a pre-packed 25 cm x 4 mm i.d. 10-pm LiChrosorb RP-18 column (E. Merck Darmstadt F.R. Germany). The column temperature was held at 40 "C with a water-jacket. The mobile phase was acetonitrile - water at a flow-rate of 1.5 nil min-l.Solvent concentrations are indicated on the chromatograms. Micro - scale Reaction The micro-scale conversion of C,-C9 linear aliphatic aldehydes into 2,4-dinitrophenyl hydra-zones was performed in an aqueous reagent (25 ml of distilled water and 0.8 ml of 2,4-DNPH reagent) and in a two-phase system (aqueous reagent and 15 ml of 2,2,4-trimethylpentane). The aldehyde concentrations ranged from 8.0 x M for nonanal. After being stirred for 1 h reaction mixtures were extracted with 2,2,4-triniethyl-pentane (2 x 15 ml). The solvent was then evaporated the hydrazones were re-dissolved in 1 ml of acetonitrile and analysed by HPLC. In a second experiment propanal hexanal and nonanal standards were diluted 1 + 2 1 + 4 and 1 + 9 and the reaction was carried out in the aqueous reagents.Sampling Efficiency In order to determine the sampling efficiency air concentrations of 0.36 p.p.m. (mol/mol) of formaldehyde 0.93 p.p.m. of propanal 0.56 p.p.m. of hexanal and 0.39 p.p.m. of nonanal were generated with a motor-driven syringe. Two bubblers containing different amounts of 2 x M 2,4-DNPH solution were connected in series and the polluted air was sampled at a rate of 1 1 min-l for 30 min. After sampling the reagent solutions were combined and stirred with a magnetic stirrer for 1 h. The hydrazones were next extracted with 50 ml of carbonyl-free 2,2,4-trimethylpentane while the bubblers were rinsed with a further 15 ml of carbonyl-free 2,2,4-trimethylpentane. The solvent fractions were then combined and the analysis was completed as described above.Rendering Emission Sampling Samples were taken in a Belgian rendering plant with an annual capacity of 10000 t of raw material. Gases and vapours released during cooking pass consecutively through a grease trap a surface condenser and two water scrubbers. The air sampling flow-rate was 1 1 min-l; samples were taken during 15-min sampling periods. VAN LANGENHOVE et a2. CARRONYL QUANTITATION Experimental A d y s t VoZ. 108 M for formaldehyde to 1.8 x The material is processed in batch cookers under vacuum. Samples were taken at the last scrubber outlet. Results and Discussion HPLC Separation and Calibration of the 2,4-Dinitrophenylhydrazones The separation of 2,4-dinitrophenylhydrazones of the linear aliphatic C,-C aldehydes 2-methylpropanal 2-methylbutanal 3-methylbutanal and acetone were investigated.Fig. 1 (a) and (b) show the separation obtained with 10- and 5-pm columns respectively. In both instances the derivatives of linear aliphatic aldehydes and acetone are completely separated March 1983 IN RENDERING EMISSIONS BY REVERSED-PHASE HPLC 33 1 i 100 a 2 7 5 1 50 25 I 2 5 /-.7 8 910 4 0 5 10 15 20 5 10 15 20 25 tlmin Fig. 1. Separation of 2,44nitrophenylhydrazones by reversed-phase HPLC using (a) a 10-pm and ( b ) a 5-pm RP-C, column. The water - acetonitrile gradient is indicated as percentage of acetonitrile. The compounds in the mixture are 2,4-dinitro-phenylhydrazones of 1 formaldehyde; 2 acetaldehyde; 3, acetone; 4 propanal; 5 butanal and 2-methylpropanal; 6 2-methylbutanal and 3-methylbutanal; 7 pentanal; 8 hexanal ; 9 heptanal; 10 octanal; and 11 nonanal.On the 10-pm column the same relative molecular mass derivatives of isomeric C and C carbonyl compounds overlapped. As is shown in Fig. 1 ( b ) partial separation of pentanal from 2-methyl-butanal and 3-methylbutanal could be achieved. Different solvent programmes with initial acetonitrile concentrations ranging from 40 to 60% were tested with no better result. On analysing standard hydrazone solutions of the different C and C aldehydes it was found that equal amounts (0.5 pg of derivative) of equal relative molecular mass derivatives gave equal peak heights the ratio of peak heights of derivatives of different aldehydes being as follows: 2-methylpropanal to butanal 1 .OO ; 2-methylbutanal to pentanal 1.02 ; 3-methylbutanal to pentanal 1.02.Therefore 2-methylpropanal and butanal were determined as the C aldehyde group 2-methylbutanal 3-methylbutanal and pentanal as the C aldehyde group. Deriva-tives of butanal and pentanal were the standards for the C and C aldehyde group. Calibration of the method was carried out by plotting peak heights (in millimetres) zwsus the amounts of aldehydes injected. Five-point calibration graphs ranging from 15 to 1.5 pg gave correlation coefficients greater than 0.999 for all of the aldehydes tested. Assuming an air sampling volume of 25 1 (at 25 "C) and 200 pl of acetonitrile in order to re-dissolve hydra-zones quantitation of aldehydes can be performed from as low as 15 p.p.b. (formaldehyde) to 2 p.p.b.(nonanal). TABLE I CONVERSION OF LINEAR ALIPHATIC ALDEHYDES INTO 2,4-DINITROPHENYLHYDKAZONES Conversion efficiency is expressed as a percentage relative to standards ($1 = 5 ) . In j cc ted / illdch yde nmol Formaldehyde . . . . 20.9 ilcetcldehydc . . . . 14.25 Propanal . . . . 11.12 Butanal . . 9.1 Pen tanal . . 7.6 Hexanal . . . . 6.7 Heptanal . . . . 6.0 Octanal . . . . 5.1 Nonanal . . . . 4.7 One- phase system 70.95 & 2.6 101 & 3.5 101 * 4.9 93 * 6.1 89 * 4.7 104 f 2.8 96 & 3.6 94 3.5 101 f 2.5 Two-phase system 74.7 j 7.6 99 * 4.2 105 5.2 102 f 2.4 95 f 3.6 93 f 3.7 55 f. 2.8 32 f. 6.2 6 f 5. 332 Analyst VoZ. 108 Micro-scale Conversion of Aldehydes Into 2,4-Dinitrophenylhydrazones Table I shows that a quantitative conversion of linear C,-C aldehydes is obtained in the one-phase system.Formaldehyde shows a conversion of 75%. In the two-phase system heptanal and higher aldehydes show a decreasing conversion due to the hydrophobic character of these compounds. Table I1 shows that no decrease in conversion efficiency is ascertained by lower-ing aldehyde concentrations. From this it can be concluded that C,-C9 aldehydes can be quantitatively converted in the aqueous reagent system. The efficiency of conversion for formaldehyde is 75%. VAN LANGENHOVE et aZ. CARBONYL QUANTITATION TABLE I1 CONVERSION OF PROPANAL HEXANAL AND NONANAL INTO THEIR 2,4-DINITROPHENYLHYDRA-ZONES AT LOWER ALDEHYDE CONCENTRATIONS Conversion efficiency is expressed as percentage relative to Conversion (%) at dilution Initial aldehyde A > standards.Aldehyde concentration/M 1 + 2 1 + 4 1 + 9 Hexanal . . 26.5 x 100 111 104 Nonanal . . . . 18.6 x 105 103 110 Propanal . . 44 x 10-6 100 103 90 Sampling Efficiency The sampling efficiency of formaldehyde propanal hexanal and nonanal in the 2,4-dinitro-phenylhydrazine solution was tested using increasing amounts of reagent. Table 111 shows the best sampling results. It can be seen that a sampling unit of two bubblers containing 100 ml of reagent each was efficient for formaldehyde propanal and hexanal ; nonanal however was not quantitatively sampled Quantitative sampling of all four aldehydes was achieved by using two bubblers containing 200 ml of reagent each. Therefore the latter conditions were used during rendering emission sampling.TABLE I11 HYDRAZINE REAGENT AT A SAMPLING FLOW-RATE OF 1 1 min-1 SAMPLING EFFICIENCY OF PROPANAL HEXANAL AND NOSANAL IN THE 2,4-DINITROPHEXYL-Conversion efficiency is expressed as percentage relative to standards ( 1 2 = 5). Sampling efficiency using two bubblers containing n r v Aldehyde 100 nil of reagent each 200 nil of reagent each Formaldehyde . . . . 97 f 6.4 104 f 5.4 Hexanal . . . . 103 f 2.S 104 7.0 Nonanal . . 90 f 2.3 103 f 7.5 Propanal . . 100 f 5.4 100 .& 8.7 Quantitation c3f Carbonyls in Rendering Emissions Table IV shows the results of the quantitative determination of aldehydes in rendering emissions taking iI.to account the 75% efficiency of conversion of formaldehyde.Fig. 2 shows a typical chromatog;s,;. Samples were taken on 1 3 . 2 . 1 9 8 1 between 4 p.m. and 8 p m . The sampling time w~~ 15 min and it took another 15 min to prepare for sampling. During sampling 10 cookcis were in use. Each cooker contains 1500-3000 kg of raw material which is heated for 1- C h. Malodorants are liberated by the thermal breakdown of cell structures and the chemkal decomposition of animal matter. Amounts of malodorants vary during the heating proczss and depend on the nature and freshness of the raw material which results in a typical emission pattern for each batch process. The normal activity of the plant involves processing different materials at the same tinie. Therefore the concentrations mentioned in Table IV do not show the evolution of amounts of carbonyls emitted during the processing o March 1983 I N RENDERING EMISSIONS BY REVERSED-PHASE HPLC 333 100 50 25 I I I L 0 5 10 15 2C tlmin Fig.2. Chromatograxn of the analyses of 2,4-dinitrophenylhych-a-zones of carbonyl compounds samplecl a t the rendering plant. The water - acetonitrile gradient is indicated as percentage of aceto-nitrile. Compounds are 2,4-dinitro-phenylhydrazones of 1 formalde-hyde; 2 acetaldehyde; 3 acetone; 4 propanal; 5 C,-aldehydes; 6, C,-aldehydes ; 7 hexanal ; 8 hep-tanal; 9 octanal; 10 nonanal. one material. The variations in the carbonyl concentrations of the rendering emissions shown in Table IV are caused by the coincidence of malodorant producing stages in the different batches.This may explain the relatively large concentration differences between different samples. In order to find out if there is some relationship between the individual carbonyl concentra-tions the Spearman rank correlation coefficients8 for all aldehydes except formaldehyde which shows little variation were calculated. Hexanal heptanal The results are shown in Table V. TABLE IV QUANTITATIVE DETERMINATION OF CARBONYL COMPOUNDS IN RENDERING EMISSION Concentrations are in p.p.m. Sample number Aldehyde Formaldehyde Acetaldehyde Acetone . . Propanal C,-aldeh ydes C,-aldehydes Hexanal . . Heptanal Octanal . . Nonanal . . I 1 1 2 3 4 5 6 7 . . 0.2 0.2 0.19 0.19 0.21 0.19 0.19 2 1.87 2 2.27 3.8 2.13 1.92 0.16 0.16 0.14 0.08 0.22 0.2 0.13 0.21 0.25 0.16 0.22 0.35 0.24 0.29 .. 0.53 0.96 0.90 1.22 1.45 0.82 0.64 0.94 1.79 1.75 2.3 2.7 1.49 1.28 . . 0.44 0.48 0.34 0.56 0.69 0.57 0.34 . . 0.19 0.21 0.16 0.25 0.30 0.32 0.16 . . 0.17 0.21 0.16 0.23 0.27 0.36 0.14 . . 0.28 0.38 0.33 0.41 0.48 0.44 0.2 334 VAN LANGENHOVE VAN ACKER AND SCHAMP octanal and nonanal correlate significantly at the 1% level. At the 5% level these aldehydes, also correlate with propanal. The C,-aldehyde group correlates with the C,-aldehyde group at the 1% level. Neither the C,- nor the C,:group correlate with propanal hexanal heptanal or octanal. An explanation for the correlations may possibly be found in the reactions that lead to aldehyde formation. Linear aliphatic aldehydes are formed by the dismutation of fatty acid hydroperoxides with or without migration of double bonds.g During the hating of animal matter amino acids may react with a-dicarbonyl compounds to form an aldehyde having one carbon atom less than the original amino acid.This reaction called Strecker degradation, will give rise to the branched aldehydes 2-methylpropanal 2-methylbutanal and 3-methyl-butanal from valine isoleucine and leucine respectively.1° TABLE V SPEARMAN RANK CORRELATION COEFFICIENTS BETWEEN THE DIFFERENT CARBONYL CONCENTRATIONS 1 Acetaldehyde; 2 acetone; 3 propanal ; 4 C,-aldehydes ; 5 C,-aldehydes ; 6 hexanal 7 heptanal ; 8 octanal; 9 nonanal. 2 3 4 5 6 7 8 0.210 0.331 0.501 0.501 0.690 0.610 0.690 0.652 0.104 0.104 0.580 0.540 0.576 1.0003. 0.595 0.444 0.500 0.595 0.444 0.500 0.607 0.607 0.879* 0.803* 0.786* 0.9603.0.9543. 0.9923. 9 0.828t 0.557 0.786* 0.750* 0.750* 0.9493. 0.8983. 0.929t * Significant at 5% level (n = 7). t Significant at 1% level ( n = 7). Taking into account these two reactions generating aldehydes the Spearman correlation coefficients seem to indicate that the C,- and C,-aldehyde groups mainly consist of branched aldehydes otherwise a correlation of these groups with aldehydes formed by fatty acid oxida-tion would be expected. The literature results for odour detection values of individual aldehydes are not very con-sistentll in that the thresholds for aldehydes seem to vary between 1 and 50 p.p.b.l0?l2 In contrast the reported threshhold for acetone varies between 20 and 32 Comparing these data with the concentrations found in rendering emissions it can be concluded that acetone does not contribute to the odour problem.All of the aldehydes that were determined are present in supra-threshold concentrations. Therefore these compounds are at least partially responsible for the rendering odours. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. References Hoshika Y. and Takata Y. J. Chromatogr. 1976 120 379. Smith R. A. and Drummond I. Analyst 1979 104 875. Kuwata K. Uebori M. and Yamasaki Y. J . Chvomatogr. S c i . 1970 17 264. Fung K. and Grosjean D. Anal. Chem. 1981 53 168. Demko P. It. J . Chromatogr. 1979 179 361. Selim S. J . Chromatogr. 1977 136 271. Wild F. “Characterization of Organic Compounds,” Second Edition Cambridge l’niversity Press, Siegel S. “Nonparametric Statistics for the Behavioral Sciences,” McGraw-Hill New York 1956, Badings H. T. Neth. Milk Dairy J . 1970 24 147. Amoore J . E. Forrester L.J. and Pelosi P. Chem. Senses Flavor 1976 2 17. Fazzalari F. A. Editor “Compilation of Odor and Taste Threshold Values Data,” American Society Quadagni D. G. Buttery R. G. and Okono S. J . Sci. Food Agvic. 1963 14 761. Davis J . C. Chem. Eng. 1973 86. Kittel G. and Wendelstein P. G. J. Arch. Klin. Exp. Ohren-Nasen- Kehlkopfheilk 1971 199 683. Cambridge 1962 p. 112. p. 202. for Testing and Materials Baltimore MD 1978. Received January 4th 1982 Accepted September 23rd 198