Five hundred grams (9.35 moles) of technical ammonium chloride and 1330 g. (14.8 moles) of paraformaldehyde are mixed and allowed to react as in the procedure described for the preparation of trimethylamine hydrochloride (p. 531). To obtain from this crude reaction mixture a water or alcohol solution of trimethylamine, it is not satisfactory merely to treat with alkali and lead the trimethylamine through the solvent, since the gas is inefficiently absorbed; in fact, it is almost impossible to get a concentrated solution in this way without loss of material. The best way to prepare a solution is to liquefy the trimethylamine (b.p. 3.5°) and in this form run it into the solvent. For this purpose, the 5-l. flask in which the initial reaction is carried out is fitted with a rubber stopper holding a separatory funnel and a glass tube for leading off the trimethylamine (Note 1). This tube is run into one opening of a large U-tube or Wolff bottle holding soda-lime for drying the gas. The exit tube from this drying bottle has a glass tube leading to the top of an upright condenser which consists of a spiral tube cooled with a freezing mixture (Fig. 29) (Note 2).

Fig. 29.

The bottom of this condenser leads through a rubber stopper fitting tightly in a 2- or 3-l. flask, holding the solvent and immersed in an ice-salt mixture. An exit tube from this flask leads to a Wolff bottle kept at a low temperature and holding a little solvent, and through this solvent any unabsorbed trimethylamine must bubble. The exit tube of this last Wolff bottle is closed with a soda-lime tube if an anhydrous solvent is being used (Note 3).

The separatory funnel is filled with sodium hydroxide solution (1100 g. in 2 l. of water) which is then allowed to flow into the reaction mixture at such a speed that, after the air in the apparatus has been replaced by trimethylamine, a continuous stream of drops of trimethylamine flows from the condenser into the solvent and practically no gas goes through the end Wolff bottle. At the beginning, the evolution of trimethylamine is rapid, and hence the addition of the alkali must be very slow (Note 4); as soon as the reaction mixture is partly decomposed, however, the alkali may be added more rapidly. After it has all been added, the reaction mixture is heated ten to fifteen minutes to make sure that all the amine is evolved. The time for this alkali treatment is about three to four hours.

With such an experiment as is described above, using 1000 g. of absolute alcohol (p. 249) in the solvent flask and about 150 g. in the end Wolff bottle, 1469 g. of solution is obtained in the main flask and 174 g. in the Wolff bottle, thus indicating that a total of 491 g. (89 per cent of the theoretical amount) of trimethylamine is present. There is thus obtained a 32 per cent solution in the main flask, which has a specific gravity of about 0.760 at +5° (Note 5). By titration with standard hydrochloric acid (Note 6), the same amounts of trimethylamine are indicated as are found by weighing. In several experiments, the yields were consistently 85–90 per cent of the theoretical amount (Note 7).