Anal. Chem. 1996, 68, 3354-3358

Identification of Chemical Interferences in Aldehyde and Ketone Determination Using Dual-Wavelength Detection Wilhelm Po 1 tter and Uwe Karst*

Lehrstuhl fu¨ r Analytische Chemie, Anorganisch-Chemisches Institut, Westfa¨ lische Wilhelms-Universita¨ t Mu¨ nster, Wilhelm-Klemm-Strasse 8, 48149 Mu¨ nster, Germany

A method for the rapid and convenient identification of chemical interferences in the determination of aldehydes and ketones in air samples using the 2,4-dinitrophenylhydrazine (DNPH) method is described. The ratio of absorptions at 360 and 300 nm is characteristic for groups of related aldehydes and ketones as well as for the main interferents. It has been determined by UV/visible spectroscopy of pure standard compounds and confirmed by HPLC analysis with UV/visible detection of complex hydrazone mixtures. The application of this method has led to the identification of 2,4-dinitrochlorobenzene as another interfering compound after sampling of air with high nitrogen dioxide content when using hydrochloric acid as catalyst. The possibilities and limitations of the dual-wavelength detection with the DNPH method are discussed for mixtures of standards and real samples from car exhaust. The 2,4-dinitrophenylhydrazine (DNPH) method has been established as the most widely used procedure for the determination of aldehydes and ketones in air samples in recent years.1-3 The carbonyl compounds react with acidified DNPH in solid or liquid phase, forming the corresponding hydrazones4,5

These are separated by means of high-performance liquid chromatography (HPLC) and detected using UV/visible spectrophotometry at wavelengths between 340 and 380 nm, depending on the absorption maxima of the relevant hydrazones.6 The main advantage of the DNPH method is the ability to analyze various aldehydes and ketones simultaneously in a complex mixture. Sampling can be performed in liquid phase (1) Beasley, R. H.; Hoffmann, C. E.; Rueppel, M. L.; Worley, J. W. Anal. Chem. 1980, 52, 1110-1114. (2) Kuwata, K.; Uebori, M.; Yamasaki, H. J. Chromatogr. Sci. 1979, 17, 264268. (3) Grosjean, D.; Fung, K. Anal. Chem. 1982, 54, 1221-1224. (4) Brady, O. L. J. Chem. Soc. 1931, 756-759. (5) Allen, C. F. H. J. Am. Chem. Soc. 1930, 52, 2955-2959. (6) Rappoport, Z.; Sheradsky, T. J. Chem. Soc. (B) 1968, 277-291.

3354 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

using solutions of the derivatizing agent in impingers7 or with solid sorbents using DNPH-coated test tubes8 or passive sampling devices,9,10 respectively. Due to the importance of the method, it has been introduced as a standard procedure by several national standardization bodies and is under discussion to be introduced as a standard method within the European Union. More recent research has resulted in the identification of chemical interferences caused by the presence of ozone11-13 or nitrogen dioxide14 in air samples. 2,4-Dinitrophenyl azide (DNPA) has been recognized as the main component when sampling air with high nitrogen dioxide concentrations with DNPH in acidic media. Its chromatographic properties are similar to those of formaldehyde 2,4-dinitrophenylhydrazone:14,15

The DNPH method has found frequent use for the determination of aldehydes and ketones in car exhaust16,17 and cigarette smoke.18,19 These matrices are very likely to contain significant amounts of nitrogen oxides. Problems caused by coelution of the formaldehyde 2,4-dinitrophenylhydrazone and DNPA can be overcome chromatographically up to a certain extent using suitable stationary and mobile phases.14 Other polar aldehyde and ketone hydrazones (e.g., those of carbonyls with additional hydroxy groups or carboxylic acid functions) may elute with even lower retention times than the (7) Grosjean, D. Environ. Sci. Technol. 1982, 16, 254-262. (8) Binding, N.; Thiewens, S.; Witting, U. Staub-Reinhalt. Luft 1986, 46, 444446. (9) Levin, J. O.; Andersson, K.; Lindahl, R.; Nilson, C.-A. Anal. Chem. 1985, 57, 1032-1035. (10) Levin, J. O.; Lindahl, R. Analyst 1994, 119, 79-83. (11) Arnts, R. R.; Tejada, S. B. Environ. Sci. Technol. 1989, 23, 1428-1430. (12) Smith, D. F.; Kleindienst, T. E.; Hudgens, E. E. J. Chromatogr. 1989, 483, 431-436. (13) Rodier, D. R.; Nondek, L.; Birks, J. W. Environ. Sci. Technol. 1993, 27, 2814-2820. (14) Karst, U.; Binding, N.; Cammann, K.; Witting, U. Fresenius J. Anal. Chem. 1993, 345, 48-52. (15) Karst, U.; Gro ¨mping, A. H. J.; Binding, N.; Witting, U.; Cammann, K. Am. Environ. Lab. 1995, 3, 32-33. (16) Swarin, S. J.; Lipari, F. J. Liq. Chromatogr. 1983, 6, 425-44. (17) Lipari, F.; Swarin, S. J. J. Chromatogr. 1982, 247 (2), 297-306. (18) Risner, C. H.; Martin, P. J. Chromatogr. Sci. 1994, 32, 76-82. (19) Mansfield, C. T.; Hodge, B. T.; Hege, R. B.; Hamlin, W. C. J. Chromatogr. Sci. 1977, 15, 301-302. S0003-2700(96)00319-8 CCC: $12.00

© 1996 American Chemical Society

formaldehyde derivative.20 Therefore, a simple method for the identification of coelutions would be a very valuable tool for quality assurance within the DNPH method. EXPERIMENTAL SECTION Chemicals. All chemicals (aldehydes, ketones, DNCB, and DNPH) were purchased from Aldrich Chemie (Steinheim, Germany) in the highest quality available. Hydrochloric acid and sulfuric acid were Merck (Darmstadt, Germany) analytical grade. Chromosorb P/AW 60/80 mesh was purchased from Merck. The thin-layer chromatography (TLC) stationary phase and the organic solvents (acetonitrile gradient grade, ethanol) were also from Merck. Synthesis. All hydrazones described here were prepared according to a procedure based on the work of Behforouz et al.21 For the synthesis of the 1-buten-3-one (methyl vinyl ketone) derivative, the solvent ethanol must be replaced by diglyme22 to avoid the addition of ethanol to the double bond. The method of Behforouz et al. is advantageous compared to older methods4,5 due to the removal of excess acid from the product. This results in an increased stability of the more delicate hydrazones, e.g., hydrazones of unsaturated aldehydes and ketones, compared to the products obtained with the previously described procedures. The products were characterized by UV/visible, NMR (1H and 13C), and IR spectroscopy, mass spectrometry, melting point analysis, and elemental analysis. Purity of the products was controlled by TLC (stationary phase, silica gel; mobile phase, toluene/acetonitrile 7:3 (v/v)) and by HPLC (for conditions, see below). The products were purified by recrystallization from ethanol if necessary. The data of all products coincide well with literature data. A DNPA standard was synthesized according to the method of Bailey and Case23 and characterized by mass spectrometry, UV/ visible, and IR spectroscopy. The product is very hygroscopic, although no hydrolysis could be observed. This may be a source of uncertainty in the determination of its molar absorptivity (see below). Photometer. The Lambda 2 UV/visible spectrometer (Perkin Elmer, Norwalk, CT) with PECSS V.4.1 software (Perkin Elmer) was used. UV Absorption Measurements. All UV/visible measurements were performed within concentration ranges from 2.3 × 10-5 to 5.4 × 10-5 mol/L of the hydrazones in acetonitrile. The spectra were recorded in the range from 200 to 400 nm. Air Sampling. Sampling tubes and stoppers were prepared according to the dimensions published by Binding et al.8 Using a personal sampling pump, an air stream of 1.0 L/min was drawn through the tube. A trap filled with solid sodium hydroxide and calcium chloride prevented water and hydrochloric acid from vaporizing into the pump. Sampling was performed in a distance of 10 cm from the exhaust pipe of the car. Sample Preparation. The used sampling tubes were eluted with 5 mL of acetonitrile for 10 min under frequent shaking and left to rest for another 20 min. After centrifugation, 10 µL of the yellow solution was used for HPLC analysis. (20) Edelkraut, F.; Brockmann, U. Chromatographia 1990, 30, 432-435. (21) Behforouz, M.; Bolan, J. L.; Flynt, M. S. J. Org.Chem. 1985, 50, 11861189. (22) Shine, H. J. J. Org. Chem. 1959, 24, 1790-1791. (23) Bailey, A. S.; Case, J. R. Tetrahedron 1958, 3, 113-131.

Figure 1. UV/visible spectra of the 2,4-dinitrophenylhydrazones of acetaldehyde (s), butynal (-‚-), crotonaldehyde (‚‚‚), and p-tolualdehyde (- -).

HPLC Instrumentation. The high-performance liquid chromatograph consisted of the following components: two LC-10AS pumps (Shimadzu, Duisburg, Germany), SPD-10AV detector (Shimadzu), SIL-10A autosampler (Shimadzu), Class LC 10 Version 1.4 software (Shimadzu), and CBM-10A controller unit (Shimadzu). Injection volume was 10 µL. The column material for column A was Merck LiChroSpher RP-18 (Merck, Darmstadt, Germany) in ChromCart cartridges (Macherey-Nagel, Du¨ren, Germany): particle size, 5 µm; pore size, 100 Å; column dimensions, 250 mm × 3 mm; guard column, 8 mm × 3 mm. Total dead volume of the system was ∼0.7 mL, including the 0.5 mL volume of the mixing chamber Model SUS (Shimadzu). The column material for column B was Deltabond AK (Keystone, Bellefonte, PA): column dimensions, 150 mm × 4.6 mm; particle size, 5 µm; pore size, 300 Å. Chromatographic data were obtained using column and elution conditions A unless otherwise specified. HPLC Analysis. Different elution conditions were chosen for both columns. For separation, binary gradients consisting of acetonitrile and water were selected with the following profiles: column A time (min) c(CH3CN) (%) flow (mL/min)

0 49 0.62

1 49

6.5 65

8 80

0 30 1.5

1 30

8.5 60

10.5 80

11.5 80

14.5 49

19.5 (stop) 49

column B time (min) c(CH3CN) (%) flow (mL/min)

12 80

12.5 30

14 (stop) 30

RESULTS UV/visible spectra of some aldehyde 2,4-dinitrophenylhydrazones from four different groups of aldehydes (aliphatic, aromatic, those with conjugated double bonds, and those with conjugated triple bonds) are presented in Figure 1. It should be noted that each of these aldehyde hydrazones is representative of its group of aldehyde and ketone hydrazones in terms of the UV/visible absorption maxima and minima and the molar absorptions (compare to the spectroscopic data of related hydrazones in Table 1). It is evident that acetaldehyde 2,4-dinitrophenylhydrazone and butynal 2,4-dinitrophenylhydrazone exhibit their absorption minima Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3355

Table 1. UV/Visible Spectroscopic and Chromatographic Data for a Series of Hydrazones , ×103 2,4-dinitrophenylhydrazone

360 nm

300 nm

(360 nm)/ (300 nm)

A(360 nm)/ A(300 nm)

λmax

(λmax), ×103

formaldehyde acetaldehyde propanal acetone acrolein butanal isobutanal 2-butanone methacrolein 1-buten-3-one cyclopropanecarboxaldehyde cyclobutanone crotonaldehyde 1-butyn-3-one 2-butynal pentanal hexanal cyclohexanone benzaldehyde salicylaldeyde p-tolualdehyde DNCB DNPA

17.6 17.6 20.1 19.6 25.5 20.9 20.1 19.1 23.4 17.2 19.7 20.8 22.0 18.4 22.0 20.6 20.7 20.8 17.6 17.2 19.3 0.3 1.1

3.5 2.1 2.1 1.8 3.9 2.2 2.1 1.7 3.6 2.4 1.5 2.0 4.4 2.2 2.6 2.1 2.1 1.7 5.1 6.3 7.0 1.8 9.8

5.1 9.2 9.5 11.0 6.5 9.7 9.4 11.3 6.6 7.3 12.9 10.5 5.0 8.5 8.5 9.8 10.0 11.9 3.5 2.7 2.7 0.1 0.1

5.9 8.4 9.3 9.4 5.6 9.8 9.7 10.2 5.4 5.8 11.2 9.7 3.9 7.6 7.7 9.9 10.0 11.2 2.9 2.6 2.4 0.1 0.1

349 360 360 364 369 361 361 365 370 371 368 363 376 363 360 361 361 367 380 382 383 247 302

19.4 19.6 20.1 19.9 26.9 21.0 20.1 19.5 25.2 18.7 20.6 20.9 25.7 18.5 22.0 20.5 20.7 21.6 21.6 23.1 25.8 10.9 9.9

Table 2. Characteristic Absorption Ratio Ranges for Different Groups of Hydrazones aldehyde/ketone formaldehyde other aliphatic aldehydes/ketones aldehydes/ketones with conjugated triple bonds aldehydes/ketones with conjugated double bonds aromatic aldehydes/ketones 2,4-dinitrophenyl azide (DNPA) 2,4-dinitrochlorobenzene (DNCB)

Figure 2. UV/visible spectra of formaldehyde 2,4-dinitrophenylhydrazone (- -), DNPA (s), and DNCB (‚‚‚).

and maxima at lower wavelengths than crotonaldehyde 2,4dinitrophenylhydrazone and p-tolualdehyde 2,4-dinitrophenylhydrazone. As the most important aldehydes and ketones to be determined with the DNPH method are the aliphatic compounds, the detection wavelengths of 360 and 300 nm have been chosen to be a good compromise for the absorption maxima and minima of these compounds. Therefore, aromatic and unsaturated aldehyde hydrazones will be detected at shorter wavelengths than their maxima and minima. This results in lower ratios of the peak areas at the detection wavelengths. Throughout the whole text, the peak area ratio is the peak area at λ ) 360 nm, divided by the peak area at λ ) 300 nm. In Figure 2, the UV/visible spectra of formaldehyde 2,4dinitrophenylhydrazone and the interfering compounds DNPA and DNCB are presented. Compared to the spectra of other aliphatic hydrazones, the formaldehyde 2,4-dinitrophenylhydrazone spectrum shows a small shift to lower wavelengths for both the absorption maximum and minimum. DNPA exhibits its maximum at λ ) 302 nm and only a weak absorption at λ ) 360 nm. Therefore, the selected wavelengths are well suited to detect an interference from DNPA when using HPLC, due to an absorption 3356 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

A(360 nm)/ A(300 nm) 5.5 8-11.5 7.5-9 5-7.5 2.5-3.5 0.12 0.12

ratio of far less than 1 at the detection wavelengths selected above. DNCB shows a broad absorption band at even lower wavelengths than DNPA, with a maximum at λ ) 247 nm. The absorption at λ ) 300 nm is already very low (compare to the molar absorptivity listed in Table 1), and it is even lower at λ ) 360 nm. The absorption ratio for DNCB at the selected detection wavelengths is far below 1, as was the case with DNPA. Interferences from both compounds can therefore be detected using the dualwavelength approach. The absorption ratios obtained for the wavelengths mentioned above in UV/visible spectroscopy should correlate with the data of the peak area ratios at these wavelengths for pure peaks in HPLC. In Table 1, the UV/visible spectroscopic data given by photometric and HPLC measurements for a series of hydrazones and the two interfering compounds are presented. It is evident that the ratios of the molar absorptivities differ significantly for various groups of hydrazones and the interfering compounds. These characteristic absorption ratio ranges for different groups of hydrazones are summarized in Table 2. The values obtained by UV/visible spectrophotometry and by HPLC coincide well in most cases. The variations between both values have been observed to be up to 30% in some cases. Three factors may have caused this: (a) Measurements at the flanks of the peaks for those compounds with absorption maxima and minima which are significantly different from the selected wavelengths for ratio

Figure 3. Chromatogram of a series of aldehyde and ketone 2,4dinitrophenylhydrazones, DNPH, and DNPA at the detection wavelengths λ ) 360 nm (s) and λ ) 300 nm (‚‚‚). The concentrations range from 1.4 × 10-4 to 2.6 × 10-4 mol/L.

determination. In this case, small variations of the absorption wavelengths will cause larger uncertainties in the absorption determination. (b) Measurements in different solvents for UV/visible spectrophotometry and HPLC, respectively. To obtain realistic chromatographic conditions, the HPLC mobile phase was chosen as a gradient with acetonitrile and water, leading to different solvent compositions for all hydrazones during one chromatogramm. The UV/visible solvent was pure acetonitrile. This may result in slight differences in both the extinction coefficient and the absorption maximum. (c) A propagation of errors leading to significant uncertainties. Considering that the differences between the molar absorptions of all hydrazones and the interfering compounds are very large, a coelution of only a few percent of an interfering compound with a hydrazone can be identified. This can be demonstrated by a simple calculation: a mixture of 95% (molar ratio) formaldehyde hydrazone (peak area ratio, 5.5) and only 5% of 2,4-dinitrophenyl azide (peak area ratio, 0.12) will lead to a total ratio of 4.5. This is significantly different from the values for the formaldehyde hydrazone, even when considering the potential uncertainties discussed above. In Figure 3, a chromatogram of a mixture of hydrazone standards, DNPH, and DNPA is presented at the detection wavelengths of λ ) 360 and 300 nm. The hydrazones of aliphatic, unsaturated and aromatic aldehydes and ketones can be classified easily on the basis of considerations about the peak area ratios stated above. The very low peak area ratio obtained for DNPA proves that this peak does not represent a hydrazone. This chromatogram, however, represents only a very small fraction of the hydrazones that might be present in a real sample chromatogram. Different kinds of interferences may occur in real samples: coelution of related aldehyde hydrazones may become a serious problem. In reversed-phase chromatography, the hydrazones are eluted in the order of decreasing polarity. In our case, 2,4-dinitrophenylhydrazones of a homologous series of aliphatic aldehydes will therefore elute in the order of increasing chain length of the aldehyde. Therefore, lower aldehydes, such as formaldehyde and acetaldehyde, should be separated well from other aliphatic hydrazones. Nevertheless, an increasing chain length of the aldehyde will also lead to a larger number of isomers of the carbonyl compound. Unsaturated aldehydes and ketones

Figure 4. Chromatogram of a series of 10 C4 aldehyde and ketone 2,4-dinitrophenylhydrazone standards at the detection wavelengths λ ) 360 nm (s) and λ ) 300 nm (‚‚‚). The concentrations range from 0.3 × 10-4 to 1.2 × 10-4 mol/L. 1, Cyclopropanecarboxaldehyde; 2, cyclobutanone; 3, 2-butynal, 1-butyn-3-one; 4, crotonaldehyde, 1-buten3-one (methyl vinyl ketone), 2-butanone; 5, methacrolein; 6, butanal, isobutanal, 2-butanone.

as well as cyclic carbonyls may occur, starting with C3 or C4 compounds, respectively. Propanal, acetone, and acrolein may occur as C3 compounds, but the separation of their hydrazones is still possible.24 The C4 compounds may be subject to larger problems. We have therefore synthesized a series of 10 C4 aldehyde and ketone hydrazones and separated them using our standard conditions for column A mentioned above. The corresponding chromatogram is presented in Figure 4. It is evident that the dual-wavelength detection adds only a little information in this case compared to a single-wavelength detection at λ ) 360 nm. This is because the separation of the hydrazones of related carbonyls (e.g., butanal, isobutanal, and 2-butanone, E-isomer) is not sufficient under these chromatographic conditions. Additional problems arise from the significantly different chromatographic properties of the hydrazone isomers of asymmetric ketones (E- and Z-isomers of 2-butanone). The chromatographic and the UV/visible spectroscopic properties of the 2,4-dinitrophenylhydrazones of butanal and isobutanal are almost identical. Even a mass spectrometric detection of these two compounds is not very promising, as they have similar mass spectra. The only significant difference in the mass spectra is the different intensities of the (M+ - 29) and the (M+ - 30) m/z peaks for these compounds. This is, however, not sufficient for a reliable detection of the individual components. Further problems may arise from the hydrazones of long-chain or aromatic aldehydes and ketones with additional polar functional groups. Note the retention time of salicylic aldehyde hydrazone in Figure 3, which is very similar to the retention time of the C4 carbonyl hydrazones in Figure 4. In this case, the dual-wavelength detection is very helpful to classify a hydrazone by tracing back to an aliphatic, aromatic, or unsaturated carbonyl compound with different UV/visible spectroscopic properties. The best application of the dual-wavelength detection, however, is the identification of chemical interferences resulting from the (24) Possanzini, M.; DiPalo, V. Chromatographia 1995, 40, 134-138.

Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3357

Figure 5. Chromatogram of an automobile exhaust sample with a volume of 2 L from a diesel-fueled car with dual-wavelength detection at λ ) 360 nm (s) and λ ) 300 nm (‚‚‚).

reaction of DNPH with oxidants. We have therefore investigated the determination of aldehydes and ketones in automobile exhaust as an example for a matrix with a very high content of nitrogen oxides. In Figure 5, the chromatogram of a real sample with a volume of 2 L from a diesel-fueled car is presented. With a retention time of 7.15 min, the DNPA peak can be identified. The wavelength ratio A(360 nm)/A(300 nm) of 0.14 (determined) compared to 0.12 (standard) proves that this peak is not a hydrazone. The UV/visible spectrum is almost identical with that of the DNPA standard. More information can be gained from the peak at 7.66 min. The formaldehyde 2,4-dinitrophenylhydrazone is expected to elute with this retention time. A comparison of the expected wavelength ratio A(360 nm)/A(300 nm) of 5.5 with the obtained value of 2.5 proves, however, that this peak is not pure. Therefore, we have used a second column with different separation properties to identify the impurity. HPLC fractions were collected and analyzed by gas chromatography/ mass spectrometry (GC/MS). Two compounds were detected, which could be identified as formaldehyde-2,4-dinitrophenylhydrazone and 2,4-dinitrochlorobenzene (DNCB) by their characteristic fragmentation and by comparison with spectra of pure standards. A chromatogram of the same sample obtained with column B and the corresponding elution conditions listed above is presented in Figure 6. Under these conditions, DNCB eluted after formaldehyde 2,4dinitrophenylhydrazone. Note that the peak maximum of the coeluting compounds formaldehyde 2,4-dinitrophenylhydrazone and DNPA depends on the applied detection wavelength due to small differences in the retention times of the hydrazone (4.33 min) and DNPA (4.25 min). Because of to the low molar absorptivity of DNCB at both detection wavelengths (see Table 1), very large amounts of DNCB must be present in the sample. To establish the origin of the DNCB peak, the synthesis of DNPA from DNPH and nitrite25 was (25) Clusius, K.; Schwarzenbach, K. Helv. Chim. Acta 1958, 41, 1413-1416.

3358 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

Figure 6. Chromatogram of an automobile exhaust sample with a volume of 2 L from a diesel-fueled car with dual-wavelength detection at λ ) 360 nm (s) and λ ) 300 nm (‚‚‚) using column B and the corresponding elution conditions listed above.

performed using hydrochloric acid and, alternatively, sulfuric acid to acidify DNPH to identify DNCB as a reaction byproduct of this reaction. As expected due to the absence of other potential chloride sources, the origin of DNCB could be unequivocally traced back to the presence of hydrochloric acid. Hydrochloric acid has been described by many authors as an acidifying agent when using the DNPH method.1-3,8,24 As the pure hydrazone standards are prepared in solutions with high sulfuric acid content, according to Behforouz et al., sulfuric acid should be a suitable substitute for hydrochloric acid when sampling aldehydes and ketones with the DNPH method in matrices with high nitrogen dioxide content. CONCLUSIONS Dual-wavelength detection is shown here to be a valuable tool for the identification of chemical interferences when determining aldehydes and ketones using the DNPH method. It can be applied with programmable dual-wavelength detectors as well as with diode array detectors. In the latter case, the dual-wavelength detection provides a fast screening possibility for interferences. Even variable-wavelength detectors can be used with this method, recording the same sample twice at different detection wavelengths. The method is particularly valuable for identifying chemical interferences of the reaction of DNPH with nitrogen oxides, as demonstrated for automobile exhaust samples from diesel-fueled cars. ACKNOWLEDGMENT Financial support by the DFG (Deutsche Forschungsgemeinschaft, Bonn, Germany) and the Fonds der Chemischen Industrie (Frankfurt, Germany) is gratefully acknowledged. Received for review April 1, 1996. Accepted June 27, 1996.X AC960319V X

Abstract published in Advance ACS Abstracts, August 1, 1996.