In extracting the bases from the distillate, a small amount of acidsoluble phenols and hydrocarbons is carried over to the acid solution, They also absorb ultraviolet energy a t 250 to 260 mp and will eventually interfere with the determination of the bases. This can be overcome by washing the acid extract with iso-octane and by holding all phenols in the alkaline solution during the h a 1 extraction of the bases with iso-octane. The four synthetic mixtures, which were contaminated purposely with hydrocarbons from tar distillate, demonstrated the removal of the hydrocarbons by washing the acid extract with iso-octane. Although the tar bases in distillates are mainly pyridines and quinolines, they may contain other nitrogenous compounds, such as pyrroles, indoles, and anilines (6, 8). However, pyrroles and indoles are very weak bases: Their pKb values are so close to 14 that they are not extractable with dilute mineral acids. Anilines are basic and are extractable with dilute mineral acids. Some aniline derivatives, such as toluidines, have been identified in tar bases. Their uItraviolet spectra (1, 4) show the highest absorbance from 235 to 250 mp and the lowest from 260 to 270 mp. Fortunately, their presence is usually small in comparison with pyridine bases (8).
Table 11. Determination of Total Pyridines and Quinolines in Tar Distillates
Distillate Lignite Bituminous iar coal tar Distillatitn end point, C. 106.0 Pressure, mm. Hg 0.2 Total pyridines in sample, wt. yo 1.1 Total quinolines in sample, wt. % 0.13
106.0
in the hexane phase can be analyzed by this procedure directly, and the bases in the methanol phase can be purified by distillation and extraction (2). Pyridine bases in samples from petroleum refining can probably be determined by this method.
0.2 LITERATURE CITED
1.1 0.52
The results with distillates from two different tars are presented in Table 11. The spectrum of tar bases from the bituminous coal tar showed a small additional peak a t 252 mp, This is due to a small amount of aniline derivatives (3). A crude tar base, consisting of neutral oil, tar bases, and small amounts of benzene and tar acids, was recovered as a by-product in the liquid-liquid countercurrent extraction of tar acids from the bituminous coal-tar distillate using aqueous methanol (9). Analysis of this crude tar base mas in good agreement with the results in Table 11. Therefore, this method can be applied to tar refinery streams, such as the extract and raffinate, from the countercurrent liquid-liquid extraction of tar distillates with aqueous methanol and hexane (9). The bases
(1) American Petroleum Institute Research Project 44, Catalog of Ultraviolet Spectral Data, 1957.
(2) Chang, T . 4 . L., Karr, Clarence, Jr., ANAL.CHEX.29, 1617-19 (1957). (3) Chang, T.-C. L., garr, Clarence, Jr.,
unpublished material.
(4) Friedel, R. A., Orchin, Milton, “Ultraviolet :pectra of Aromatic Compounds, Riley, New York, 1951. (5) Hofmann, Edward, Arch. Hug. u. Bakteriol. 128, 169-78 (1942). (6) Kruber, 0. von, Racithel, A., Grigoleit, G., Erdol u. Kohle 8 , 637-43 (1955). (7) LeRosen, H. D., Wiley, J. T., A x . 4 ~ . CHEM.21, 1175-7 (1949). (€4) LORTV.H. H.. “Chemistrv of Coal Utfihation,”’ pp. 467-f3, Wiley, New York, 1947. (9) Neuworth, M. B., Hofmann, Vera,
.,
Kelly, T. E., Znd. Eng. Chem. 43, 1689-94 (1951).
RECEIVED for review September 19, 1957. Accepted January 27, 1958. Division of Gas and Fuel Chemistry, Symposium on Modern Techniques in Research on Coal and Related Products, 132nd Meeting, ACS, New York, N. Y., September 1957.
Spectrophotometric Determination of Carbonyl Oxygen FRED
H. LOHMAN
Research and Development Department, Miami Valley laboratories, Procter & Gamble Co., Cincinnati, Ohio
b Carbonyl compounds in the range from 3 to 3 0 0 p.p.m. of carbonyl oxygen are determined b y a method based on condensation of the carbonyl compounds with 2,4-dinitrophenylhydrazine to form the highly colored 2,4-dinitrophenyIhydrazones, separation of the yellow product from excess reagent by extraction with hexane, and measurement of the absorbance of the resulting solution a t 340 mp. The procedure represents an improvement over existing methods because reagent interference is eliminated, the spectra of the pure, neutral 2,4-dinitrophenylhydrazones are very reproducible, and the color is completely stable.
T
HE determination of small amounts of carbonyl compounds based on the formation of the 2,Cdinitrophenylhydrazones (DNPH) was first reported by Mathewson (6). His procedure for the determination of acetone is appli-
972
ANALYTICAL CHEMISTRY
cable only to water-soluble samples. Lappin and Clark (3) determined total carbonyl in mixtures of carbonyl compounds by measuring the red color formed when an alcoholic solution of the 2,4-dinitrophenylhydrazone is treated with alkali. As absorbance measurements are made in the presence of excess reagent 2,4-dinitrophenylhydrazine, a different spectrum (which probably is a combination of the individual spectra of reagent and the condensation product) from that of the pure 2,Pdinitrophenylhydrazones is obtained. The method has been extensively criticized by Mandelowits and Riley (4). Pool and Klose (6) modified the method of Lappin and Clark for the determination of monocarbonyl compounds in rancid foods; dicarbonyl bis2,4dinitrophenylhydrazones and excess reagent were eliminated by carrying out the condensation on an alumina column. Measurement of the absorbance of the 2,Pdinitrophenylhydrazones in al-
kaline solution has the inherent disadvantage that the molar absorptivity decreases rather rapidly with time; evidently the colored species formed is unstable. If the absorbance is measured a t the wave length of maximum absorption (435 mp), there is serious interference from excess reagent (4). At the secondary maximum (535 mp), reagent intereference can be avoided, but the method is only half as sensitive (Figure 1). Toren and Heinrich (8) determined a butadiene-furfural condensation product by extraction of the 2,4dinitrophenylhydrazone with iso-octane. These authors suggested the extraction as a possible general procedure for the separation of carbonyl 2,4dinitrophenylhydraaones from the excess reagent; however, the procedure had not been tested on other carbonyl compounds. In the method described, reagent interference is eliminated by quantitatively extracting the carbonyl 2,4-dinitro-
/ ,
Acetaldehyde
Acetone
1
,/ f H e x a l d e h y d e - 2 , 4 -
-\\
0 w
DNPH
H e x a l d e h y d e - 2 , 4 - DNPH in olkaline solution
2 0.5 m
111
-I
0
v)
m
a
0.0
\
/
I I I I I I J 600 400 50 0
Y
I
30 0
I 700
WAVE L E N G T H , m p
Figure 1.
CONC.
All absorbance measurements were made with a Cary Model 11s spectrophotometer, using the visible source and 1-cm. cells. Standard Materials. The preparation, purification, and identification of the carbonyl 2,Minitrophenylhydrazones employed in this work (except crotonaldehyde) have been reported by Steward, Huber, and Lutton (7). Each had been recrystallized to a constant melting point.
Aldehydes and Ketones Obtained from Matheron, Coleman & Bell
2-Butanone Butyraldehyde Heptanal Crotonaldehyde 2.4Pentanedione Cyclohexanone 5-Nonanoneo
Boiling Point, C. 79-80 73-75 54-56
(20 mm., 16.4% oxygenj 100-102 138-14 1 154-156 4.5-3.5b (12,1% oxygen)
Organic Chemicals. ' Eastman Melting point.
Reagents. 2,4-Dinitrophenylhydrazine, Eastman Kodak KO. 1866, was recrystallized from ethanol. Ethanol, anhydrous, was purified as described by Lappin and Clark (9). Hexane, American Mineral Spirits Co. (initial boiling point 146' C., dry point 157' C.), was used as supplied. All other chemicals were reagent grade or better.
2, 4-DNPH (pg./ml.)
Absorbance measured In hexane at 340 m p
2,4-Dinitrophenylhydrazine (reagent solution, 1 mg. per ml. in ethanol) and hexanaI-2,4-dinitrophenylhydrazone (1 0 y per ml.) In hexane and in alkoline ethanol
APPARATUS AND REAGENTS
ALDEHYDE
Figure 2. Beer's law graphs for four typical carbonyl 2,4-dinitrophenylhydrazones
Visible absorption spectra
phenylhydrazones into hexane. The absorbance of the stable yellow hexane solution is then measured spectrophotometrically at 340 mp (Figure 1).
OF
EXPERIMENTAL WORK
Determination of Molar Absorptivities. Stock solutions of the standard aldehyde and ketone 2,4-dinitrophenylhydrazones were prepared by weighing approximately 3 mg. t o the nearest 0.01 mg., dissolving in hexane, and diluting t o 100 ml. with the same solvent. Three quantitative dilutions were made from each, and their absorbance was measured at 340 mp against a blank of hexane. The wave length of maximum absorption for each compound was also measured on one of these dilutions. The graph of Beer's law for four typical standard substances is shown in Figure 2. Molar absorptivities were calculated from the slope of the appropriate line and the molecular weight of the compound involved. The 2,4-dinitrophenylhydrazones, their melting points, molar absorptivities at 340 mp, and wave length of maximum absorption are listed in Table I. With the exception of crotonaldehyde and formaldehyde the molar absorptivities are in very good agreement with
each other, as are the values of. , , ,A The rather large uncertainty expressed for the mean molar absorptivity (Eaao) is largely a reflection of the variation in Amax for crotonaldehyde and formaldehyde rather than a real variation in color intensity. The impurity of crotonaldehyde2,4-dinitrophenylhydrazone probably contributes to the low value of E- for that substance. As the method was developed for the analysis of oxidized fats which contain unsaturated as well as saturated aldehydes (I, e), the uncertainty expressed for EM@ is inherent because the proportions of such a mixture are totally unknown. Analyses of Standard Carbonyl Solutions. Standard solutions of seven carbonyl compounds in mineral oil were prepared by weight t o contain approximately 1 mg. of aldehyde or ketone per gram of oil. The solutions were stored in glass-stoppered weighing bottles or glass-stoppered weighing burets. The solutions were analyzed according to the following procedure. PROCEDURE. Pipet 10 ml. of 2,4dinitrophenylhydrazine solution (1 mg. per ml. in purified ethanol) and 5 ml. of hexane into 8 50-ml. glass-stoppered
Table I.
Molar Absorptivities and Absorption Maxima of Some Carbonyl 2,4Dinitrophenylhydrazones Carbonyl 2,4DNPH M.P., O C. E,L./Mole Cm., 340 Mp Arn*x, h.Ip Acetaldehyde 161.2 22,100 340 340 Propionaldehyde 152.4 22,500 Butyraldehyde 121-2 22,000 340 Hexanal 107-8 23,200 340 Heptanal 105-6 23,300 340 342 Octanal 106-7 23,500 Nonanal 105-6 23,200 340 Decanal 106-7 23,200 341 Acetone 125-6 21,800 346 Crotonaldehydea 173- 185 17,200 358 Formaldehyde 165-166 18,900 33 1 Average 21,900 =k 1,400 a Derivative not recrystallized.
VOL. 30, NO. 5, MAY 1958
973
conical flask, add 1 drop of concentrated hydrochloric acid, and weigh the flask and contents accurately. Transfer approximately 0.25 to 0.5 gram of sample solution to the flask, quickly restopper, and reweigh. Obtain the sample weight by the dxerence in weights. Place the mixture in a water bath at 50" C. for 1 hour, then transfer the solution quantitatively to a small separatory funnel containing 15 ml. of methanol and 10 ml. of 1% sodium bicarbonate solution. Add 10 ml. of hexane, shake, allow the phases to separate, and catch the alcohol-water phase in the original sample flask. Collect the hexane layer in a 150-ml. beaker. Again transfer the alcohol-water solution to the separatory funnel and repeat the extraction four times with 15-ml. portions of hexane. Dry the combined extracts with a little sodium sulfate, transfer to a 100-ml. volumetric flask, and dilute to the mark. Read the absorbance a t 340 mp in a 1-cm. cell using hexane as the reference solution (zero absorbance). Divide the net sample absorbance by the average molar absorptivity to obtain the molar concentration of carbonyl oxygen. Reagent Blank. Absorbance values for a set of reagent solutions and solvents, determined according t o the above procedure, were constant within f 0.02 absorbance unit when the reagent solution was used within 4 days. The average blank absorbance (for a given set of reagents) was subtracted from total sample absorbances to obtain the net absorbance of the sample. A new blank should be determined for each new set of reagents. RESULTS
ferent types of carbonyl compounds are shown in Table 11. The recoveries listed for each carbonyl sample are the results of consecutive analyses. Results for crotonaldehyde are low because of its much smaller molar absorptivity a t 340 mp as compared with the average used in all of the calculations. Either 2,4pentanedione failed to form a colored derivative under the conditions chosen, or the derivative was not extracted by hexane.
DISCUSSION
Table 111. Reproducibility of Method as Applied to Soybean and Cottonseed Oils
Sample"
Heating Period a t 50" C., Hours
S. B. oil
1
Mg. 7% Carbonyl Oxygen*
2 2 4 4
13.9 13.2 11.9 13.3 12.8 13.9
2 2
13.0 13.7 15.4 14.8 15.2 15.3
1
a Analyses reported are on one sample each of two oils,'and hence do not reflect variations of composition. Actually composition of vegetable oils is constant for a given kind of oil. Mg./gram X 100.
Results for the analysis of seven difTable 11.
Oxygen Recoveries on Standard, Known Mixtures of Carbonyl Compounds in Mineral Oil
Carbonyl Function Butyraldehyde
*
Oxygen as Carbonyl Taken, mg. Found, mg.
0.231 0.226 0.224 0.220 0.274 0.282 0.134 0.138 0.148 0.149 0.0318 0.0302 %Butanone 0.0510 0.0495 0.0523 0,0550 0.0522 0.0532 HeptanaP 0.0874 0.0877 0.138 0.136 0.0472 0.0560 5-Nonanoneb 0.0518 0.0612 0.259 0.223 Crotonaldehyde 0.289 0.264 0.108 0.0933 0.0610 0.0517 0.0700 0.0575 0.0792 0.0797 Cyclohexanone 0.105 0.114 0.142 0.133 0.0583 0.0465 0.0984 0.0833 None recovered 2,PPentanedione 0.211 16.4y0 oxygen by analysis; theory, 14.0% oxygen. 12.1% oxygen by analysis; theory, 12,5yo oxygen.
974
ANALYTICAL CHEMISTRY
The method was applied to samples of partially oxidized cottonseed and soybean oil to show the applicability of the method to vegetable oil samples and to determine the effect of heating (50" C.) on such samples. Table I11 shows no significant increase in carbonyl content with increased time of heating and, therefore, no measurable carbonyl formation resulting from this mild heating.
The high specificity and wide applicability desired of a good functionalgroup reagent are combined in 2,4-dinitrophenylhydrazine. The extraction of the carbonyl derivatives into hexane should further increase the specificity of this reagent. The absorbance for reagent blanks was commonly around 0.3, even when purified ethanol was used to prepare the reagent solution. For this reason the reagent solutions should be pipetted carefully for each sample and blank. Methanol, rather than ethanol, was placed in the separatory funnel before extraction to decrease the solubility of the short-chain carbonyl 2,4-dinitrophenylhydraaones in the aqueous alcohol phase. The addition of more water might be more effective, but rather serious emulsion difficulties result when vegetable oil samples are analyzed. Following the suggestion of Toren and Heinrich (8),iso-octane was tried in place of hexane. Hexane gives a slightly more efficient extraction of the lower molecular weight carbonyl 2,4-dinitrophenylhydrazones. Since completion of this study, the method has been successfully applied to the determination of laurone in synthetic samples.
yo Recovered 102 102 97.4 96.4 99.5 106 103 95.1 102 100 98.4 119 118 86.0 91.4 86.8 84.8 82.2 101 108 93.6 80.0 84.6
LITERATURE CITED
(1) Coleman, J. E., Knight, H. B., Sm-ern, D., J. Am. Oil Chemists' SOC.32, 135 (1955). (2) Kawahara, F. W., Dutton, H. J., Ibid., 29, 372 (1952). (3) Lappin, G. R., Clark, L. C., ASAL. CHEM.23, 541 (1951). (4) Mandelowitz, A., Riley, J. P., Analyst 78, 704 (1953). (5) Mathewson, W. E., J. Am. Chem. SOC. 42, 1277 (1920). (6) Pool, M. F., Klose, A. A., J. Am. Oil Chemists' SOC.28, 215 (1951). (7) Stewart, C. B., Huber, W. F., Lutton, E. S., J . Am. Chem. SOC.73, 5903 (1951).
(8) Toren, P. E. Heinrich, B. J., ANAL. CHEM.27, 1986 (1955).
RECEIVEDfor review August 28, 1957. Accepted December 30, 1957. Division of Analytical Chemistry, 132nd Meeting, .4CS, Sew York, N. Y., September 1957.
