PURIFICATION OF
DIMETHYLANILINE Use of Aqueous Formic Acid to Remove Aniline and Monomethylaniline €RANK 0.RITTER Washington Square College, New York University, New York, N. Y.
reacts with 85 to 90 per cent formic acid in a manner entirely analogous to aniline. Distillation of equivalent weights of the amine and acid leaves t,he high-boiling n-methylforinanilide as a residue. It was also determined thst the direct formylat.ion of both aniline and monomethylaniline takes place in the presence of a large excess of dimethylaniline. However, as was expected, loiiger contact of the reactants was found necessary. Because of these promising preliminary observations, it was considered desirable to determine the limits of the method by making a series of controlled runs with careful analyses of the resulting products. Analyses were made conveniently and accurately by use of the Zerewitinoff-Grignard method (3) as applied to amines by Hihbert and Sudborough ( 2 ) . As far as can be ascertained, the application of this met.lrod to anilinemonomettrylaniline-dimetliylaniline mixtures has not been reported. There is one obvious advantage in the use of formic acid as compared to the use of acid anhydrides or acid chlorides in this purification. Since formylation takes place in the presence of water, it is not necessary to have dry crudes in order to prevent loss of the acylating agent by hydrolysis.
LTHOUGH the nsc of acetic anhydride is one of the standard procedures for the removal of aniline and mononiethylaniline from dimethylaniline, the possibilities in the use of the usual formylating agent-formic acid-have not been investigated in this connection. Formylation of aniline takes place readily even with an aqueous acid. Wallach and Wusten (g) prepared formanilide in tliis manner by refluxing equivalent weights of aniline and the acid and then distilling the mixture. This procedure left the high-boiling formanilide behind, while both the .n.ater of the aqueous acid and that formed in the reaction n-cre drivcn over into the distillate. If it could be shown that monomethylaniline formylates under the same condit.ions as does aniline, the reaction might be useful as a method of separation of the primary and secondary amine from the tertiary. The present investigation has shown that methylaniline
Aqueous formic acid can be used to remove aniline and monomethylaniline from dimethylaniline to give good yields of purified products that are as good as or better than the usual commercial grades. Repeated distillation of the crude with the acid produces a good yield of a purified product that is better than some C. P. grades, even when the original impurity approaches 18 per cent. Steam distillation produces good yields of purified products comparable to the usual commercial grades if the original crude is not excessively contaminated. The method in general has the advantage that it can be applied to wet crudes.
General Experimental Procedure Known mixtures were made by adding aniline and monomethylaniline to dimethylaniline. These were then treated with a little more than the necessary amount of 85.3 per cent formic acid necessary to act with the added impuritics. Tbe new mixtures were thcn refluxed for varying lengths of timo. In some experiments the separation of anilides and dimethylaniline xas accomplished by direct distillation. In others, water \%-as added to the mixture after refluxing, and the whole was directly heated to effectst.eam distillation.
Analytical Procedure Before applying the Zereii-itinoff-GrignRrd method to the purified samples, it was necessary to hydrolyze residual anilides to the corresponding amines. This procedure rendered all impurities sensitive to the reagent. Hydrolysi,s was accomplished by refluxing the oils for 30 minutes to a n 33
INDUSTRIAL AND ENGINEERING CHEMISTRY
34
hour with enough concentrated hydrochloric acid to hold them in solution. The solutions were then made alkaline with sodium hydroxide solution to recover the hydrolyzed sample. The analytical method also required the removal of all traces of water. The hydrolyzed samples were therefore allowed to stand over solid sodium hydroxide for 24 hours, They were then decanted from the solid and further dried by agitating with a small amount of solid sodium peroxide a t room temperature. After contact with the peroxide for 30 minutes, the oils were filtered and ready for analysis. Sodium peroxide was used as a drying agent because of its ready availability and efficiency. It was used only after analyses had shown that tightly stoppered samples of the hydrolyzed oils suffered no change on standing in contact with the solid for several days. Analyses were made in a convenient adaptation of the original Zerewitinoff apparatus ( 3 ) . Figure 1 shows the essential parts. The confining liquid was mercury, Samples were weighed out in a short length of common test tube, which remained in the reaction chamber during the run. The sample was dissolved by adding 3 cc. of dry isoamyl ether. Three cubic centimeters of methyl magnesium iodide in isoamyl ether solution (made according to Zerewitinoff's directions, 3) was introduced into the side arm. Air in the apparatus was displaced by means of dry nitrogen. This precaution was found necessary because the last traces of methane were liberated only after vigorous agitation of the reactants. This agitation caused absorption of oxygen by the Grignard reagent if air was present, and the final readings therefore were not significant. After mixing the Grignard reagent and sample in the usual way, the reaction chamber was well agitated until all traces of methane had been evolved. Final readings were taken after the apparatus had attained thermal equilibrium. All analyses were made a t room temperature. Therefore aniline and monomethylaniline each liberated one equivalent of methane, and no differentiation between the amines was possible. Impurities were always calculated as monomethylaniline. The percentages reported in this paper are therefore higher than they would be had each amine been determined separately and the percentages added together. MELTING POINTS.As a check on the Zerewitinoff method, the melting points of several purified samples were also taken. These were determined with a large-scale thermometer graduated in 0.5" C. A correction was applied to the observed readings for depth of immersion used in the tests. ACCURACYOF METHOD. Before using the Zerewitinoff method on unknown samples, it was necessary to investigate its reliability when applied to known mixtures. Inasmuch as the purest dimethylaniline obtainable showed some impurity when analyzed by this method , known mixtures could not be prepared by the complete use of pure reagents. Pure aniline and monomethylaniline were therefore added in known amounts to a previously analyzed dimethylaniline. The new mixtures were again analyzed. The results of three such analyses follow: Mixtiire
1 2 3
Calcd. Impurity as Monomethylaniline Per cent 4.57 2.30 6.25
Impurity b y Analysis Per cent 4.45 2.29 6.32
The accuracy of the method as applied was within 0.1 per cent. The check on duplicate analyses was of the same order. The pure monomethylaniline used to prepare the mixtures was made from a commercial sample. It was &st treated with an excess of formic acid to produce n-methylformanilide.
VOL. 28, NO. 1
All traces of diniethylaniline were then removed from the latter by steam distillation. A small portion of the resulting residue was tested for the presence'of formanilide by shaking with sodium hypochlorite solution. Formanilide causes the development of a brown color with this reagent even when present in traces. The present sample remained colorless and was therefore free from formanilide. Hydrolysis of the pure n-methylformanilide produced pure methylaniline.
Detailed Description of Runs RUNSINVOLVING DIRECTDISTILLATION. (1) A mixture containing 100 grams commercial dimethylaniline (2.1 per cent original impurity), 10 grams aniline, and 10 grams commercial monomethylaniline was treated with 16.5 grams of 85.3 per cent formic acid (1.3 theory). After refluxing 3 hours, the mixture was distilled through a 6-inch (15.2-cm.) Hempel column. All distillate coming over through 1.5" C. above the boiling point of the main fraction was collected. A portion of the distillate was hydrolyzed, dried, and analyzed as already described: Yield of purified dimethylaniline % Impurity i n hydrolyzed oil (as donomethylaniline), % Melting point of hydrolyaed oil, C.
90 1.6 1.7
(2) A mixture of 100 grams commercial dimethylaniline (2.1 per cent original impurity), 10 grams aniline, and 10 grams commercial monomethylaniline was treated with 15.5 grams of 85.3 per cent formic acid (1.3 theory). After refluxing 1hour, the mixture was distilled through the Hempel column as in the previous run. The distillate containing unreacted and excess acid was returned to a clean flask and redistilled as before. This process was repeated. A portion of the thrice-distilled oil was hydrolyzed, dried, and analyzed: Yield of purified dimethylaniline, % Impurity in hydrolyzed oil (as monomethylaniline), % Melting point of hydrolyzed oil, C.
85 0.7 2.1
RUNSINVOLVIXG STEAM DISTILLATION. (1) A mixture of 100 grams commercial dimethylaniline (2.1 per cent original impurity), 10 grams aniline, and 10 grams commercial monomethylaniline was treated with 15.5 grams of 85.3 per cent formic acid and refluxed for 3 hours. Water was added to the mixture, and the whole was heated in a large Claisen flask. Water was added from time to time t o keep the volume constant. The oil coming over with steam was caught in five fractions. Each fraction was hydrolyzed, dried, and analyzed separately. The results are shown in Table I. IMPURITY WITH TABLEI. VARIATIONOF PERCENTAGE YIELDIN STEAM DISTILLATION Progressive Impurity in Vol. of Impurity i n Progressive Accumulated Fraction Fraction Fraction Yielda Fractions b cc. Per cent Per cent Per cent Run 1 (Original Impurity Approximately 18%) 1 21 1.6 21 1.5 2 20 2.0 41 1.7 3 21 2.4 62 1.9 4 21 3.7 83 2.4 5 9 8.7 92 3.0 Run 2 (Original Impurity Approximately 11%) 1 80 1.6 80 1.6 2 11 3.8 91 1.9 a Calculated by adding the volume of all fractions through the one in question and assuming the specific gravity of all fractions equal to unity. weight df pure dimethylaniline i n original mixture assumed equal to 106 grams. b Calculated by determining the actual weight of impurity in each fraction, summing u these weights through the fraction in question, a?d setting the sum over tEe weight of the combined fractions; speclfic gravity of all fractions assumed equal to unity.
The figures show that, as steam distillation of the homogeneous mixture in the still progresses, the result.ing distillate becomes increasingly richer in anilides. By starting with a crude containing less original impurity, one would expect a greater yield of purified product in each of the percentage ranges. This expectation is realized in the next run. (2) A mixture of 100 grams commercial dimethylaniline (2.1 per cent original impurity), 5 grams aniline, and 5.7 grams mmmercial monomethylaniline was treated with 8.0 grams of 85.3
INDUSTRIAL AND ENGINEERING CHEMISTRY
JANUARY, 1936
periquidcentwasformic acid and refluxed 4.5 hours. The resulting steam-distilled exactly as in the previous run. The
distillate was caught in two fractions. Each fraction was hydrolyaed, dried, and analyzed separately. The results are given in Table 1. DETERMINATION OF STANDARDS OF PURITY.In order to compare the products purified by this method with commercial dimethylanilines obtainable on the open market, the Zerewitinoff method of analysis was applied to two such samples. Analysis was also made of a so-called c. P. grade, the specifications Of Which read (IFree from Monomethy'aniline." All samples were thoroughly dried by the procedure already described:
35 Impurity
Melting Point
c.
% Commercial sample 1 Commercial sample 2 c. P . grade
2.1 1.8 1 .o
1:'i 1.8
Literature Cited (1) Hibbert, H . , and Sudborough, J. J.,J. Chem. Sor. (London) Proc., 20, 165 (1904). (2)-Wallach, O., and Wusten, M., Ber., 16, 145 (1883). (3) Zerewitinoff, Th.,Ibid.,40, 2023 (1907). RECEIVED March 20, 1935. Presented before the Division of Industrial and Engineering Chemistry at the 89th Meeting of the American Chemical Society, New York, N . Y., April22 to 26,1936.
Dicarboxylic Acid Esters of Tetrahydrofurfuryl Alcohol J. N. BORGLIN Hercules Powder Company, Wilmington, Del.
AKETTI ( 3 ) previously described the preparation of the acetate, propionate, butyrate, valerate, benzoate, and furoate of this alcohol. Hewlett (1) prepared the chloroacetate, iodoacetate, oxalate, and salicylate. The oxalate is the only dicarboxylic acid ester reported. The following paragraphs will describe the preparation and properties of dicarboxylic acid esters of tetrahydrofurfuryl alcohol and phthalic anhydride and also terpinene maleic anhydride. The esters of other dicarboxylic acids, such as succinic, sebacic, maleic, etc., may be similarly prepared. The preparation of terpinene maleic anhydride has been described by Peterson and Littman (8).
2
Preparation FROM TERPINENE MALEICANHYDRIDE.Four hundred and fifty parts by weight of terpinene maleic anhydride and 95 parts by weight of tetrahydrofurfuryl alcohol, refined to remove furfuryl alcoho1,l were heated 36 hours a t 170' to 190' C. under a blanket of carbon dioxide. The excess alcohol was removed by reduced pressure distillation, whereby 711 parts or an 89 per cent yield of a semi-solid ester resulted. This ester was equivalent to a WW grade of rosin in color and had an acid number of 66.5. The half-ester has an acid number of 167. The acid ester obtained may be further esterified by means of an aliphatic alcohol. Thus, when 100 parts by weight of the ester were heated 16 hours a t 160' C. with 100 parts by weight of butyl alcohol and 10 parts by weight of toluene, and the volatile constituent was removed by reduced pressure distillation, 103 parts by weight of a pale-colored ester were recovered which analyzed as follows : Acid No. Refractive index
(ZOO
C.)
45 1.5023
1 The furfuryl alaohol was removed b y treating the tetrahydrofurfuryl alcohol with sulfur dioxide, allowed t o stand several hours, filtered, and redistilled under reduced pressure. The furfuryl alcohol may also be removed by contacting with fuller's earth at elevated temperature and distillation under reduced pressure.
In these experiments purified tetrahydrofurfuryl alcohol was used. If, however, the alcohol used contains furfuryl alcohol, the resulting esters are quite dark. Catalytic hydrogenation of the latter products results in esters of satisfactory color. FROM MALEICANHYDRIDE.Ninety-eight parts by weight of maleic anhydride and 200 parts by weight of tetrahydrofurfuryl alcohol were refluxed 48 hours at 170' t o 190' C. with 20 parts by weight of toluene to remove the water of reaction. This toluene was removed as needed. The excess alcohol and toluene were removed by reduced pressure distillation, whereby a pale-colored liquid ester was recovered and which analyzed as follows: Acid No. Refractive index (20' C.) Sp. gr. (15.6°/15.60 C.)
69
1.4897 1.1931
The half-ester of maleic anhydride has an acid number of 282. FROMPHTHALIC ANHYDRIDE.One hundred parts by weight of phthalic anhydride and 150 parts by weight of tetrahydrofurfuryl alcohol were heated 48 hours a t 170' to 190' C. with 20 parts by weight of toluene. The excess alcohol and toluene were removed by reduced pressure distillation. The resulting residue was a pale-colored liquid ester which analyzed as follows: Acid No. Refractive index ( Z O O C.) Sp. gr. (15.6°/15.60 C.)
64.5 1.6260 1.2165
The half-ester of phthalic anhydride has an acid number of 226. For the preparations described, the reactants were placed in a round-bottom flask to which was connected a 12-inch unpacked Liebig column; the latter was connected to a water condenser. This arrangement permitted the water of reaction to distill off,, leaving anhydrous reactants in the flask. If the water of reaction is not removed, the rate of esterification is appreciably reduced. These esters have rather high acid numbers which can be reduced by more extended heating or by the use of a higher temperature. Also, esters of lower acid number can be pre-