servation of coupling of the proton on the a-carbon of aliphatic amines. Direct addition of excess trifluoroacetic acid to the amine under study turned out to be a simple, satisfactory procedure for preparing a solution of amine salt. Thus, the N M R spectrum of an amine can be run in a suitable solvent (carbon tetrachloride or deuteriochloroform), the solvent evaporated with a stream of nitrogen, and trifluoroacetic acid added directly to the residue in the N M R tube. If the trifluoroacetic acid is added all at once, the heat of salt formation is controlled by dilution. Examination of the amine salt spectra yielded useful information. The broad absorption of the protons on the nitrogen atom was found downfield in the aromatic region, and the number of protons could be obtained by integrating the spectrum. This permitted classification of the amine as primary, secondary, or tertiary. The protons on the acarbon could be recognized by the downfield shift (12) from their position in the free amine spectrum, and by their additional splitting. Splitting of the protons on the a-carbon(s) of amine salts is a function of the number of protons on the p-carbon(s) and those on the nitrogen atom. In the compounds examined, the J values were very similar and the S 1 rule was applicable. The splittings are shown in Figure 1 for some primary, secondary, and tertiary amine salts. Spin decoupling can be used very effectively in amine salts. By de-
+
coupling the N + H protons from the nitrogen atom (irradiating the nitrogen atom), splitting by the protons on the a-carbon(s) would become apparent. This would give an additional check on the number of a-protons in the event they are buried under other absorption. Decoupling of the a-protons by irradiating the N + H protons gives a simplified a-proton spectrum : this provides additional confirmation for the identity of the a-proton peaks, and gives the number of 0-protons. The a-proton splitting can also be simplified by irradiating the 0-protons; in simple cases, the resulting a-proton pattern will depend on the number of N + H protons thus providing another parameter for classification of amines. Spin decoupling was carried out on a primary amine salt (Figure 1, VIa and VIb) to illustrate its utility in these situations. Thus, irradiation of the N +H3 protons collapsed the a-proton sextet to a triplet (VIa), and irradiation of the 0-protons resulted in a quartet (VIb). Of course, complications are encountered as the number of a- and 0-carbons increases. I n some cases, double decoupling experiments may be useful in simplifying complex splitting. It is of interest to note that the geminal methyl groups in compound I1 are nonequivalent because of asymmetry a t the a-carbon. Reynolds and Schaefer (9) pointed out that methylene proton nonequivalence is observed in protonated benzylamines which lack a plane of symmetry along the CsH5CH2h- bond axis. Thus, the methylene
protons of protonated N-methyl-dibenzylamine shows the A B part of a n ABX s p e c t r u m 4 . e , eight peaks. Following submission of this paper, a paper appeared (6) in which the shift of N methyl groups in perdeuterioacetic acid and in trifluoroacetic acid was used for detection of N-methyl groups. A large number of shift values and J-values were given. LITERATURE CITED
(1) Acton, E. M.,, Stanford Research
Institute, unpublished work.
(2) Bhacca, N. S., Hollis, D. P., Johnson,
L. F., Pier, E. 4;, “High Resolution NMR Catalogue, Vol. 2, spectrum 489. Varian Associates, Palo Alto, Calif., 1963. ( 3 ) Bruce, J. M., Knowles, P., Proc. Chem.
SOC.1964, 294. (4) Chapman, 0. L., King, R. W., J . Am. Chem. SOC.86, 1256 (1964). ( 5 ) Freifelder, M., Matoon, R. W., Ng, Y. H., J . Org. Chem. 29, 3720 (1964’1. (6) Ma, J. C. N., Warnhoff, E. W., Can. J . Chem. 43, 1849 (1965). (7) . , McGreer. D. E.. Mocek. M. M., J . Chem. Educ. 40, 358 (1963). (8) Reynolds, W. F., Schaefer, T., Can. J . Chem. 41. 2339 f 1963’1. (9, Ibid., 4i,2119(1964). (10) Silverstein, R. RI., Bassler, G. C., “Spectrometric Identification of Organic Compounds,” p. 145, Wiley, New York. 1963. (11) Ibid., p. 137. (12) Sudmeier, J. L., Reilley, C. N., ANAL.CHEM.36, 1698 (1964). \ - - - - ,
WILLIAM R. ANDERSON, JR. ROBERTM.SILVERSTEIN Stanford Research Institute Menlo Park, Calif.
Coulometric Titration of Hydrogen Peroxide with Electrogenerated Iodine SIR: Coulometrically generated cerium(1V) has been used to titrate 50 peq. of hydrogen peroxide with an error of 2y0 (10, 1 1 ) and generated manganese(II1) has been employed in the titration of 0.7 to 5.5 mg. of hydrogen peroxide with average errors of 0.3 to 0.4% (12). The present paper describes the coulometric determination of hydrogen peroxide by iodometry. Following reaction with iodide in the presence of a molybdenum catalyst, excess thiosulfate is added and the excess is determined by titrating with electrolytically generated iodine. The thiosulfate titer is determined coulometrically under the same conditions. A pH range of 0 to 7 . 5 has been investigated. The method has been applied to the determination of 2.6 mg. (150 peq.) to 0.9 pg. (0.05 peq.) of hydrogen peroxide with an average relative error ranging from less than 0.1 to 4%. 1418
ANALYTICAL CHEMISTRY
Swift and coworkers (8, 9) and Bard and Lingane (1) have described coulometric iodometry, and pardue (6) combined precision null-point potentiometry with electrolytic generation of iodine for microrange iodometry. They determined 36 to 3.6 pg. of hydrogen peroxide over a p H range of 2 to 4.5 with an error of 0.05 pg. at all concentrations. EXPERIMENTAL
Reagent grade chemicals were used without further purification. All solutions were prepared using freshly boiled, distilled, and deionized water. Solutions of 0.05N hydrogen peroxide were standardized by titrating a 10-ml. aliquot iodometrically with standard sodium thiosulfate (5). A 10-ml. buret was used for hydrogen peroxide titrations. The sodium thiosulfate was standardized coulometrically at p H 3.0 (9)*
The generating solution consisted of
0.1,M potassium iodide which was made O.OO1lvin sodium carbonate to decrease air oxidation of the iodide. The phosphate buffers employed (ionic strength = 0.2M) were prepared as previously described (4). All pipets and burets were calibrated. Micropipets were made from polyethylene as described by Mattenheimer ( 7 ) and they were calibrated coulometrically (3)with precision of 0.2y0or better. Coulometric titrations with generating currents of 4.825 ma. or -greater were made with a Sargent coulometric current source, Model IV. For currents less than these, a ChrisFeld Microcoulometric Quantalyzer, Model 6 (ChrisFeld Precision Instruments Corp., Beltsville, Md.) was used. The generating anode and cathode were platinum foils of 1.5 sq. cm. and 0.8 sq. cm., respectively. The cathode compartment was a glass tube fitted with a sintered glass frit end. A 3%
agar-saturated potassium chloride plug was placed in the bottom of the tube to pre-Jent' diffusion of hydrogen into the test solution. The buffer under investigation served as catholyte. The indicating electrodes were two platinum foils ( 2 sq. cm.) with. 135 mv. impressed between them. I n all titrations, the impressed potential was supplied by a Sargent Model XV or Model F S polarograph. All titrations were performed by recording automatically the indicating current on the polarograph. The divisions of the chart p p e r were calibrated in terms of seconds. For titrations of 5 peq. or greater of hydrogen peroxide, t'he indicating current was recorded on t'he Model XV polarography. For titrations of 0.5 peq. or less of hydrogen peroxide, the indicating current' was recorded on the Model F S polarograph which has a chart slieed 10 times faster than the former polarograph. During generation and current measurement, the solution was stirred with a magnetic st,irring bar. The titration vessel was either a 100-ml. beaker or a 50-ml. weighing bottle. Procedure. For titrations a t 4.825 ma. or greater, 20 i d . of the potassium iodide-generating solution plus 20 ml. of the buffer or hydrochloric acid solution were added to a 100-ml. beaker. For titrations a t currents less than this, 8 nil. of potassium iodide plus 8 nil. of buffer were added to a 50-1111. \veighin,g bottle. In all titrations, a pretitration procedure was used, except where noted. A small, unmeasured amount of sodium thiosulfate solution (about 30 to 100% of the sample size) was added to the cell and then was titrated to a slight excess of iodine; the excess iodine was noted. Then the sample of thiosulfate or hydrogen peroxide plus thiosulfate was added and again titrated to a current rise. The excess iodine in the previous titration was added to the end point value. Three or more samples could be titrated in succession. Sodium thiosulfate, 0.075, was standardized by pipetting a 3-ml. aliquot into the pretitrated solution and then titrating as described. above a t a generating current of 48.25 ma. Hydrogen peroxide, 0.05S, wvas determined coulometrically by adding a 3-ml. aliquot plus the molybdenum catalyst' to the pretitrated solution followed by 3 ml. of the standard 0.07S sodium thiosulfate. The excess thiosulfate was titrated with generated iodine at 48.25 ma. The titration .value of the 3 ml. of thiosulfate minus the net' end point value for the excess thiosulfate gave the titration value of the hydrogen peroxide sample. All other titrations were performed in the same manner, unless ot,herivise noted. T h a t is, the sodium thiosulfate was standardized by taking an aliquot equal to that to be added to the sample and titrating a t the same current as used for the sample titrations. Then the hydrogen peroxide sample was titrated by taking approximately equal volumes of the peroside and the thiosulfate solutions with normality ratios
Table 1. Effect of pH on Coulometric Determination of Hydrogen Peroxide. Solutions of pH 2 or greater were buffered with orthophosphate. Generating current was 48.25 ma. NO.
Mod,
PH ml. ca.O(l.OMHC1) 0.2 ca. 0.3 (0.5144 HCl) 0.2 1 . 3 (0.06M HC1) 0 . 2 2.0 0.2 3.0 0.2
H~OZ, req. titra- hIean error, Av. dev., Taken Found tions 70 % 153.3 152.8"; 161.2b 4 -0.20"; 5.04b 0.63a; 0.59*
154.1 153.8"; 158,5c 4 -0.19"; 2.8@ 0.42a; 0.44C 153.8 154.8 3 0.65 0.25 153.8 154.2 3 0.26 0.02 154.1 154.1 3 0 0.24 4.5 0 . 4 153.8 153.7 5 -0.06 0.29 5.6 0 . 4 153.8 153.4 3 -0.26 0.33 6.2 1 . 0 153.8 152.9 4 -0,58 0.37 7.0 2 . 0 153.8 150.6 3 -2.08 0.40 7.5 4 . 0 153.8 149.2 3 -2.99 1.07 a i Y Na2S203determined at pH 3.0. * N Na2S203determined in 1.0144 HC1. c Na.&Oa determined in 0.5144 HC1. 3 7 , ammonium molybdate. Table II.
Titration of Microgram Amounts of Hydrogen Peroxide.
pH 3.0, pretitration procedure used except where noted
a
H202, req. No. Mean Av. dev., Taken Found titrations error, 70 7c hlilliamperes 5.094 5,103 3 0.14 0.61 4.826 0.5284 0.529 4 0.10 1.2 4.825 0,0519 0.049E? 2 -4.0 5.0 0.0965 0,0495 0.0481* 2 -2.8 5.5 0.0965 N o pretitration. Blank correction made. * Platinum wire indicator electrodes.
of about 5:7-i.e., a 40% excess of thiosulfate was added. RESULTS AND DISCUSSION
Often, there was a minimum in the amperometric titration curve just prior to the final current rise. I n these cases a line was drawn at the minimum perpendicular to the final current rise and the intersection of this line with the extrapolated current rise was taken as the end point. The O.05N hydrogen peroxide solution \vas stable a t room temperature for 4 hours. Solutions of 0.005-V hydrogen peroxide were standardized by taking 3-ml. aliquots and titrating coulometrically a t pH 3.0 a i t h generating currents of 4.825 ma. This was satisfactory because reliable results could be obtained a t this pH (4.825 ma.) for as little as 0.5 peq. of hydrogen peroxide. .Is little as 0.5 peq. of hydrogen peroxide was taken for coulometric analysis by pipetting 10.00 pl. of the 0.05X solution. A sample of 0.05 peq. was taken for analysis by pipetting 10.00 pl. of the 0.005LVsolution. I n initial titrations of thiosulfate at 48.25 ma., the cathode compartment was not fitted with an agar-potassium chloride plug. Results were generally high and erratic because of diffusion of hydrogen into the test solution. This difficulty was circumvented by placing the plug in the cathode compartment. The effect of pH on the titration of hydrogen peroxide is summarized in Table I. The number of milliliters of molybdenum catalyst solution given for
each p H value is the amount added in the first or single titration. With successive titrations, no more catalyst was added with the sample in solutions of pH 0 to 4.5. From pH 4.5 to 7.5, one half of the amount of catalyst used in the first titration was added with each successive sample. The sodium thiosulfate, in solutions of p H 0 to 2.0, was added and titrated immediately after the hydrogen peroxide sample and catalyst were added. From p H 2.0 to 7.0, the hydrogen peroxide was allowed to react with the iodide in the presence of the catalyst for 1 minute before the thiosulfate was added. At p H 7.0 and 7 . 5 , a waiting time of 2 minutes was used. Results were consistent with observations of other investigators. At pH 5 to 7.5, reproducibility and accuracy were increased by addition of thiosulfate at the same time the sample was added to shift the equilibrium. Recoveries in acid solutions were within 0.2% if the thiosulfate titer was determined a t p H 3.0 (true value), even though the excess thiosulfate was titrated in hydrochloric acid. The rapid increase in positive errors with increasing acidity tends to diminish as the concentration of hydrogen peroxide is decreased. Titrations of dilute solutions in acid medium (pH < 2) are more satisfactory if solutions are deaereated (9). For maximum accuracy, a pH range of 2.5 to 4.5 is recommended. Considerations of the thiosulfateiodine reaction are described elsewhere (2, 9 ) . VOL. 37, NO. 1 1 , OCTOBER 1965
0
1419
The results of the titration of a range of microgram amounts of hydrogen peroxide a t p H 3.0 are summarized in Table 11. Down to submicrogram levels, the mean error was about 0.1% and the average deviation was about 1%. With samples smaller than 1 pg. (0.05 peq.) the error increased to about 4y0 with an average deviation of 5 to 6%. I n titrations a t low current, the end point break was rounded and results became less precise. The titration of 0.05 peq. was also performed using platinum wire indicating electrodes. The noise level was reduced but the current response was much smaller. The accuracy using the platinum wires was slightly better than xhen platinum foils were used.
ACKNOWLEDGMENT
The author thanks William C. Purdy for suggesting this problem. He is indebted to Richard E. Wolf for technical assistance. LITERATURE CITED
(1) Bard, A. J., Lingane, J. J., Anal. Chim. Acta 20,463 (1959). (2) Bradburv, J. H., Hamblv. A. N.. Australian ’ J . Sci. ’ Research“’A5. 541 (1952). ( 3 ) Christian, G. D., Microchem. J . 9, 16 f1965).
(4)Christian, G. D., Purdy, W. C., J . Electroanal. Chem. 3, 363 (1962).
15) Kolthoff, I. M., Sandel, E. B., “Text-
book,, of Quantitative Inorganic Analysis, 3rd ed., p. 600, MacMillan, New York, 1952. (6) Malmstadt, H. V., Pardue, H. L., ANAL.CHEM.32,1034 (1960).
(7) Mattenheimer, H., “Mikromethoden Fur Das Klinish-Chemisch Und Biochemische Laboratorium,” p. 9, Walter De Gruyter & Co., Berlin, 1961. (8) Ramsey, W. J., Farrington, P. S., Swift, E. H., ANAL.CHEU.22,332 (1950). (9) Rowley, K., Swift, E. H., Ibid., 26, 373 (1954). (10) Sakurai, H., Kagyo Kagaku Zasshi 64,2119 (1961). ( 1 1 ) Takahashi, T., Sakurai, H., Talanta 9.189 (1962). (12j Tutundzic, P. S., Paunovic, M. M., Anal. Chim. Acta 22,201 (1960).
GARYD. CHRISTIAN^ Division of Biochemistry Walter Reed Army Institute of Research Walter Reed Army Medical Center Washington, D. C. 20012 Present address, Department of Chemistry, University of Maryland, College Park, Md.
S pe ct ro phot o met ric Dete r min a ti o n of LevuIinic Aci d with Hy d razi ne SIR: Phenylhydrazine and its 2,4dinitro analog are not suitable for the gravimetric determination of levulinic acid because of the appreciable solubility of the hydrazones and because the reagents tend to precipitate gummy materials from impure levulinic acid solutions. From a spectrometric standpoint, the absorbance of the reagent would be expected to interfere with that of the derivative. Hydrazine itself has a low molar absorptivity so that if it formed a strongly absorbing derivative quantitatively, it would be most suitable. Hydrazine was reacted with levulinic acid, and the result exhibits a strong maximum at 242 mw. On the basis of this finding, it was possible to determine extremely small amounts of levulinic acid with good precision and without preliminary isolation. EXPERIMENTAL
Apparatus. T h e absorbances were measured with a Beckman DK-2 spectrophotometer using 1-cm. cells. Reagents. HYDRAZINE MONOHYDROCHLORIDE, 50% SOLUTION. This was prepared from t h e commercially available salt or by neutralizing hydrazine hydrate with the appropriate amount of hydrochloric acid, with cooling. LEVULINICACID. Molten levulinic acid was cooled and allowed to crystallize slowly until about one half of it was solid. The melt was separated and the process repeated on the crystals. The recrystallizations were then carried out in carbon tetrachloride containing 7% chloroform, using Korite at first. The resultant colorless hygroscopic crystals were considered pure and used as a standard. 1420
ANALYTICAL CHEMISTRY
4,5 - DIHYDRO - 6 - METHYL - 3(2H) This was prepared according to Poppenberg (4) and recrystallized from benzene. The long colorless needles melted at 105’ C. and became opaque on exposure to air because of the formation of the monohydrate. Procedure. A sample containing 10-200 mg. of levulinic acid in a 20 x 150 mm. test tube is treated with 1.0 ml. of 5070 aqueous hydrazine monohydrochloride and enough water t o make about 10 ml. T h e mixture is heated in a boiling water bath for 30 minutes and then diluted to at least 1 liter. Further dilution may be necessary t o bring t h e absorbance within t h e optimum range for the spectrophotometer. T h e p H of the final solution is adjusted to 1-2 with concentrated hydrochloric acid. A blank is prepared containing an equal amount of sample and hydrochloric acid but no hydrazine. It is diluted the same as the sample, and the absorbance, A , of the sample, when run against the blank, is determined at 242 mp. Then: PYRIDAZINONE.
0.00156 X A = levulinic acid, grams/100 ml. of final solution RESULTS A N D DISCUSSION
The outstanding feature of this method is that the levulinic acid does not need to be isolated or partly purified before being determined. The use of the increase in absorbance over that of the blank, and the high dilution employed, allows the determination to be carried out on extremely dark, impure solutions. Reagent Absorption. T h e molar absorptivity of levulinic acid at 242 mp is only 20, so its presence in the blank does not interfere. The molar
absorptivity of hydrazine a t this wavelength is 43, but in strongly acid solution (pH 1-2), it falls to about unity. Because of this dependency on pH, and to minimize the absorption due to the large excess of hydrazine used, the absorption measurements are made on strongly acid solutions. Even so, with the amount of hydrazine used, it is necessary to dilute the reaction mixture to at least 1 liter to reduce the absorption due to the hydrazine salt to less than 1-270, which is about the limit of accuracy of our instrument. These precautions are particularly important when working in the range of 1-20 mg. of levulinic acid, where the hydrazine excess is about 100-fold. Probably the method could be extended to lower levels of levulinic acid by using double cells to keep hydrazine and levulinic acid separate in the blank, but this approach was not pursued. The absorption curve of the pure reaction product of levulinic acid and hydrazine was scarcely affected a t all by strongly acidifying its aqueous solution. Mineral Acid Concentration. With hydrazine alone, the reaction did not proceed to completion. I n t h e presence of mineral acid it did, but low values were also obtained with too much acid. A satisfactory amount of hydrochloric acid corresponded to t h a t in hydrazine monohydrochloride, which was used. The p H a t the beginning of the reaction could vary from 3 to 7 without significantly affecting the results. Time and Temperature. T h e reaction proceeded to completion in 30 minutes in a boiling water bath. Longer time (up to 1 hour) and lower temperature (70’ C.) did not seem to be harmful, b u t the boiling water bath was most convenient. Very little reaction occurred at room temperature in 30 minutes.
