thermal expansion, partial sample ejection, and occasional interruption of the arc. Ammonium chloride and graphite have dissimilar effects as shown by Table IV; relatively smaller proportions of graphite cause early volatilization. The slow calcium fluoride emission is pronounced in the 1018 mixture; the larger sample available with the 1012 mixture produces a better ratio. With ammonium chloride present, the best ratio is 1810, indicating an advantage of ammonium chloride if the required sensitivity permits such a drastic sample dilution. The inadequacy of a 30-second exposure is due to nonquantitative calcium fluoride emission during this time. The background improvement counteracts this for some analytical applications. The 1012 mixture arced for 3 minutes is the most dependable arrangement, as shown by Tables I11 and IV. The steadiness of the arc is a strong argument in its favor. Moderate variations in sodium, potassium, or barium content did not affect the fluorine results. Wainer and Dubois (8) presented briefly a method for enamel frits. They placed a calcium carbonate-sample mixture on the broken top of a flat electrode. This reduces the globule difficulties, but results in low sensitivity and high background. Additives for internal standardixation impose additional complications. Monnot ( 7 ) uses the unusual device of
reading a second-order silicon line, 2435.16 or 2987.65 A., against calcium fluoride 5291.00 A. Gillis, Eeckhout, and Kemp (6) use copper 5105.54 A. for slag analysis. Both claim better accuracy than when calcium fluoride is read against background, but this has not been studied to any extent for frits. AVERAGE-TIME SETTINGS
Ahrens ( I ) found that an exposure time of 1.5 seconds a t 7 amperes was adequate for phosphate rock containing a few tenths per cent fluorine. This is definitely misleading in the solution of the frit problem. For frits, the calcium fluoride emission occurred throughout a 3-minute exposure period, but this time can be altered considerably by changing the ratio of additives to sample. By burning to completion several sample and additive combinations, 8 or 10 amperes produced a favorable burn. S o serious attempt was made to increase the amperage, and a t a lower amperage the burning is too slow. With the more favorable mixtures, the calcium fluoride 5291.00-A. band in the second 30-second period is only 10 to 20% of the intensity of the first 30-seeond exposure period. This effect was studied with several mixtures. -4sample is arced a t 8 aniperes, jogged after 30 seconds, and burned an additional 2 minutes (approximately to completion for the mixtures studied). Table IV shows for
six sample mixtures the per cent transmittance of the calcium fluoride 5291.00-4.band head and background, and a corrected line intensity. The ratios of the lines for the first 30 seconds to the next 2 minutes are tabulated. Background is appreciable in the interval from 30 to 150 seconds, which makes it desirable to complete the arcing in 30 seconds. However, considerable fluorine is observed after the first 30 seconds. With the 1210 mixture, the major part comes over later, as shown by the line ratio of 0.94. LITERATURE CITED
.ihrens, L. H., “Quantitative Spectrochemical Analysis of Silicates,” pp. 93-8, .Iddison-Wesley, Cambridge,
Mass., 1955. 12j Ahrens, L. H., “Spectrochemical .\rialvsis,” pp. 145-51, 173-6, Addisonh-esley, Cambridge, Mass., 1950. (3) Castro, R., Loude, R., Spectrochint. Acta 4, 496 (1952). Crosswhite, H. M ,Zbid., 4,122 (1950).
Gillis, J., Eeckhout, J., Kemp, S . , Rev. universelle mines 8 , 284 (1952).
Harrison, G. R . , Lord, R. C., Loofbourow, ,J R., “Practical Spectroscopy, p. 446, Prentice-Hall, Sew York, 1948. IIonnot, G. A , Spectrochim. dcta 6, 153 (1954).
Kainer, E., Dubois, E. M., Bull. Am. Ceram. SOC.20, 4 (1941).
RECEIVEDfor review March 25, 1957. -4ccepted March 24, 1958. Pittsburgh Conference on -4nalytical Chemistry and -4pplied Spectroscopy, March 1957.
Determination of Simple Aliphatic Nitriles by Reaction with Alkaline Hydrogen Peroxide DOROTHY H. WHITEHURST and JAMES E. JOHNSON Development Deportment, Union Carbide Chemicals Co., Division o f Union Carbide Corp., South Charleston, W. Va.
b A chemical method for the determination of nitriles is based on the reaction of nitriles with alkaline peroxide. By concentrating the reaction mixture, the nitrile is completely saponified to the salt of the corresponding acid. One mole of base is consumed in the over-all reaction. The method has been applied to the determination of a number of simple aliphatic amines. Where it is applicable, concentrations ranging from a few parts per million to 100% can be determined. The standard deviation of the method is 0.5% for high purity samples and 0.15 p.p.m. in the range of 5 p.p.m. of the nitriles in water. 1332
ANALYTICAL CHEMISTRY
M
OST of the analytical methods reported for nitriles depend on a determination of nitrogen content. These methods are of little value for mixtures of nitrogen-bearing compounds. Siggia and Stahl (a) were able to reduce certain nitriles to amines Land t o determine the amine formed. This method has its limitations because the reduction of a number of important nitriles, such as acetonitrile and succinonitrile, is not complete. In the method presented here, the reaction of a nitrile Kith alkaline hydrogen peroxide to form the amide (1) has been used as the basis of a n analytical procedure. In the initial reaction of nitrile with
excess hydrogen peroxide and potassium hydroxide some of the amide is simultaneously converted to the corresponding acid salt. By concentrating the alkaline reagent the remaining amide can be converted completely to the acid salt. The excess potassium hydroxide is theh titrated with standard sulfuric acid using phenolphthalein indicator. The difference between a blank and sample titration is a measure of the nitrile present, APPARATUS A N D REAGENTS
Potassium hydroxide, 0.2N and 1.ON squeous solutions.
Hydrogen peroxide, 3.0 and 30.0% aqueous solutions, reagent grade. Sulfuric acid, 0.LV and 0.5.V standard aqueous solutions. Phenolphthalein, 1 .O% methanolic solution. Flasks, 300-ni1., alkali-resistant, Corning Glass Works Catalog S o . 75000. Glass columns, 40 cm. long, 10 mni. in diameter, equipped with a 24 ’40 ground-glass joint at one end.
Table I.
Purity of Nitriles -4v.
Compound .Icetonitrile Propionitrile Butyronitrile Succinonitrile
Purity, No. of Wt. yo Detns. 100.5 99.7 100.0 100.7
34 13 28 17
Std. Dev. 0.5 0.3 0.7 0.5
PROCEDURE
Determination of Purity of Nitriles. Pipet 50 ml. of 1 . O S potassium hydroxide into each of t x o 300-ml. glassstoppered, alkali-resistant flasks. Into each flask pipet 100 ml. of the 3% hydrogen peroxide. Reserve one of the flasks as a blank. Into the other flask introduce a n amount of sample t h a t contains from 6 t o 10 nieq. of nitrile. Allow the sample and blank to remain a t room temperature for 5 minutes with occasional swirling. To each flask add a few glass beads (boiling stones must not be used or erratic results will be obtained because of inconsistent recovery of the potassium hydroxide) and attach each t o a 40-cm. glass column. Grease each joint with silicone stopcock grease. Apply heat and allow the sample and blank to evaporate to a volume of approximately 10 ml. Do not allow the flasks to evaporate to dryness. Cool and wash each glass column with 100 ml. of distilled water collecting the washings in their respective flasks. Drain the column, remove the flasks, and pipet exactly50nil. of 0.5n’ sulfuric acid into each flask. Add 6 to 8 drops of phenolphthalein indicator to each flask and titrate with standard 0.LiLtrsulfuric acid to the disappearance of the pink color. The difference between a blank and sample titration is a measure of nitrile. Determination of Low Concentrations of Nitriles in Water. Pipet exactly 25 ml. of 0.2N potassium hydroxide into each of two 300-ml. glass-stoppered, alkali-resistant flasks. Into each flask pipet 20 ml. of the 307, hydrogen peroxide. T o one flask add 200 ml. of the n a t e r sample, using a suitable graduate for the transfer. To the other flask add 200 ml. of distilled m-ater, and reserve for a blank determination. Allom- the sample and blank to remain a t room temperature for 5 minutes with occasional swirling. To each flask add a fex glass beads, as in determination of purity. and attach each to a 40-cm. glass column. Greaqe each joint with silicone stopcock grease. Apply heat and allow the sample and blank to evaporate t o a volume of 2 ml. or less. Do not allow the flasks t o evaporate to dryness. Cool and wash each condenser with 50 nil. of distilled water, collecting the washings in their respective flasks. Drain the condensers and remove the flasks. Add 6 to 8 drops of phenolphthalein indicator to each flask and titrate with standard 0.1S sulfuric acid to the disappearance of the pink color. The difference between a blank and sample titration iq a measure of nitrile.
DISCUSSION
Reaction Conditions. I n the determination of purity of nitriles, which uses 100 ml. of hydrogen peroxide, complete conversion of the nitrile to the amide and acid salt occurs in 5 minutes a t room temperature for each compound listed in Table I. If the volume of peroxide is reduced t o 50 ml. the reaction of acetonitrile, propionitrile, and succinonitrile is complete b u t butyronitrile results are approximately 3% low. For the determination of low concentration of nitriles in water, it was necessary to add 30’% peroxide to a 200-ml. water sample to produce approximately a 301, solution of hydrogen peroxide, thereby simulating conditions used for purity determinations. Because only part of the amide formed from the nitrile is converted to the acid salt in the 5-minute room temperature reaction, it is necessary to saponify the remaining amide by a second reaction. I n the case of acetonitrile, approximately 70YGof the amide is converted to potassium acetate in the first reaction. By concentrating the potassium hydroxide until it is a t least 2N, complete saponification of the amides formed from acetonitrile, butyronitrile, propionitrile, and succinonitrile was obtained. I n the case of acetonitrile, when the potassium hydroxide concentration was adjusted to 1N by evaporating to a volume of 50 ml., approximately 97% purity was obtained. However, when evaporation was continued until the final volume was between 10 and 25 ml., an average purity of 100.5% was obtained for the same material. I n determining low concentrations of nitriles in water, it is necessary to evaporate to approximately 2 nil. to obtain a 2,V solution. Potassium hydroxide was best concentrated by evaporating the water through a 40-em. X IO-mm. glass column equipped with a 24/40 groundglass joint a t one end. This type of evaporation prevents loss due to spattering and entrainment. INTERFERENCES
,411 compounds n-hich are oxidized to an acid under the conditions of the reaction will interfere. Certain compounds-e.g., acetaldehyde and formaldehyde-will oxidize quantitatively,
and a correction can be made if they are present in a sample. Methanol, ethanol, and %propanol interfere only slightly and can be tolerated in small quantities. Most esters and amides will react quantitatively and can be determined independently. A correction can be applied for free acid in the sample. Amines such as ethanolamine and 2ethylhexylamine will not interfere if they form azeotropes or steam distill with water during the evaporation step. Results as high as 120% were obtained for purities of benzonitrile, acrylonitrile, ethylene cyanohydrin, and 3-methoxypropionitrile. In each case the high results could possibly be explained either by side reactions or uncontrolled oxidation a t points of unsaturation. Approximately 2 moles of potassium hydroxide were consumed for each mole of lactonitrile; however, quantitative purity values could not be obtained under the conditions of the reaction. RESULTS
Table I shows data obtained by analyzing four common saturated nitriles for determination of purity. Each liquid nitrile was redistilled and assumed to be loo%, because only a small hearts fraction, with a narrow boiling point range, mas used for the analyses. Known mixtures of aceto-, butyro-, propio-, and succinonitrile in water were prepared in 5 to 1000-p.p.m. range; a standard deviation of 0.15 p.p.m. was obtained for acetonitrile in the 5-p.p.m. range. Table I1 shows the data obtained when these mixtures were analyzed to determine the low concentration of the nitriles. Table II. Low Concentrations of Nitriles P.P.M. Compound Sdded Found Acetonitrile 1001 988 502 500 50 58 25 24 5.6 5.0” Propionitrile 6.2 8.8 Butyronitrile 5.6 4 1 Succinonitrile 5.3 5.7 a Average of 10 or more determinations, std. dev. 0.15 p.p.m.
ACKNOWLEDGMENT
The authors would like to thank
E. J. Taylor, Jr., for many of the data included in this paper. LITERATURE CITED
(1) McMaster, L., Langreck, F. B., J . Am. Chem. SOC.39, 103-9 (1912). (2) Siggia, Sidney, Stahl, C. R., ANAL. CHEM.27, 550-2 (1958).
RECEIVEDfor review October 7, 1957. -1ccepted February 28, 1958. VOL. 30, NO. 8, AUGUST 1958
1333