REACTIONS IN PHOSGENE SOLUTION. 11. FORMATION OF

---

REACTIONS I N PHOSGENE SOLUTION. 11. FORMATION O F CHLORALUMINATES.* BY ALBERT F. 0. UEKRIANN AND KENIiETH GAGOS

In a recent communication' one of us reported that potassium is attacked by phosgene containing dissolved aluminium chloride. The reaction in the case of potassium is too slow for convenient study. The present investigation was undertaken to make a survey of the behavior of other metals towards phosgene containing aluminium chloride, and to find, if possiblc, examples of the reaction lending themselves more readily to the study of the progress of the reaction. In the first, series of experiments, magnesium, calcium, zinc, cadmium and tin were treated in separate, sealed Faraday tubes with liquid phosgene containing dissolved aluminium chloride. Calcium and magnesium reacted very rapidly, the tube became very hot, and a gas was given off, which, when collected and analyzed, proved to be carbon monoxide. Cadmium, zinc and tin reacted more slowly, in the order named, and all three much more slowly than potassium . I n the case of the magnesium, 110 solid renct)ion product separated from the solution; when the solution was cooled in ice, two liquid layers separated, and on further cooling, the denser layer became syrupy. An attempt to study the phase relations of the system soon brought out the fact that more time would be required to complete the study than waszvailable for the investigation; hence the study of this system was abandoned in favor of the system resulting from the action on calcium. Cadmium, zinc and tin each yielded a solid reaction product, small in amount, forming an incrustation about the metal, and presumably slowing down the reaction. No effort has been made t o speed up these reactions by raising the temperature of the system studied, except as mentioned below. I n the case of calcium, the violence of the reaction diminishes after a time; and if the tube be allowed to cool over night, a crystalline product separates, which is neither calcium chloride, because this is entirely insoluble, nor aluminium chloride, since this is very soluble. However, since carbon monoxide is one product of the reaction, it seems reasonable to suppose that calcium chloride is also a product. The crystals, then, are evidently those of a calcium chloraluminate, or a double salt of calcium chloride with aluminium chloride; such a salt was prepared by Baud2 by fusing the anhydrous salts together. Before determining this point, it seemed desirable to know the exact function of the aluminium chloride in bringing about the reaction. With this point in mind, the following experiments were performed. Samples of calcium

* From a thesis presented by Kenneth Gagos in partial fulfilment of the requirements for the degree Master of Arts. Germann: J. Phys. Chem., 28, ooo (1924). * E . Baud: Ann. chim. phys., (8), 1, 51 (1904).

966

ALBERT F. 0. GERMANN AND KENNETH GAGOS

weighing 0.25 gm. were treated in separate tubes with I O cc. of liquid phosgene containing varying weighed amounts of aluminium chloride; the reaction was in each case allowed to run to completion, and any residue of calcium was washed with phosgene and weighed. It was thus found that the extent of the reaction is directly proportional to the amount of aluminium chloride present, provided that the calcium and phosgene are in excess,

In other words, the aluminium chloride does not act catalytically in bringing about the reaction; but calcium reacts with phosgene to form calcium chloride, which deposits on the calcium as a protective coating, and arrests further action; when aluminium chloride is present in the solution, it reacts with this surface layer of calcium chloride, forming the soluble double salt, and the reaction proceeds until the free aluminium chloride is exhausted. The same mechanism would apply to magnesium and the other metals. This interpretation of the reaction is entirely justified by the results of the following experiment, in which magnesium was selected because the product of its reaction with phosgene and aluminium chloride yields a two liquid layer system with phosgene: A gram of anhydrous magnesium chloride was sealed up in a tube with 5 cc. of liquid phosgene and 0.5 gm. of aluminium chloride, and laid aside. When next examined, the magne6um chloride had partly dissolved : cooling the solution brought about separation into two liquid layers, exactly as when metallic magnesium was used. Hence it may be concluded that anhydrous metallic chlorides may react with aluminium chloride in phosgene solution, to form double salts. Baud' isolated one double salt of aluminium chloride with calcium chloride, r.5CaCl2.zA1Cl3; two with barium chloride, BaC12.zAICI3, and 1.5BaClz.zA1C13,the second resulting when the first was heated to an elevated temperature, with loss by volatilization of one third of the aluminium chloride. Attempts to prepare CaCI2.zA1Cl3failed, one third of the aluminium chloride used always subliming, presumably because dissociation takes place at the melting point of the compound, according to the equation : 3(CaC12.2A1Cla)-+-

2

A1C13+z(~.5CaC12.zAIC13).

To determine the formula of the double salt, an excess of calcium was allowed to react with phosgene containing an amount of anhydrous aluminium chloride insufficient to react with all of the calcium as chloride, in a tube provided with a stopcock and stopcock clamp2. The reaction, at first violent, became sluggish as the aluminium chloride was used up; the tube, from which carbon monoxide was allowed to escape from time t o time, was therefore placed in a boiling water-bath for several hours, or until all action had apparently ceased. On cooling, beautiful amber colored crystals (the color probably due to the presence of iron, which was present as impurity in the aluminium chloride, and doubtless also in the calcium), apparently octahedral, separated. The solution was evaporated to dryness by evacuation, the temperature raised Ann. chim. phys. (8) 1, 51 (1904). J. chim. phys., 8, 503 (1910).

* Guye and Drouginine:

I

REACTIONS I N PHOSGENE SOLUTION

967

by immersion in a boiling water-bath, and evacuation continued as long as phosgene was evolved. This brought about efflorescence of the crystals, yielding a greyish white powder, which was subjected to analysis. The results for chlorine are known to be low, as solution of the anhydrous salt was carried out in a covered beaker, and the heat of solution was so great that some hydrogen chloride, resulting from hydrolysis, was lost by volatilization; greater precautions were t,aken with sample €3, but the odor of hydrogen chloride was very evident over the beaker. Sample A Sample B Weight of sample 0.5674 gm. 0.5332 gm. Weight of silver chloride I .6495 gm. I .0747 gm. Weight of chlorine 0.4081 gm. 0.3896 gm. o/c of chlorine 71.92 % 73.06 %, Weight of alumina Weight of aluminium yo of aluminium

0.1641 gm. 0.0873 gm. 15.39 %

0.1536 gm. 0.0814 gm. 15.27 7%

Weight of calcium oxide Weight of calcium yc of calcium

0.0834 gm. 0.0596 gm. 10.50 yc

0.0781 gm. 0.0558 gm. 10.47 %

The formula CaC12.2A1C13 represents this composition very nearly, as is shown in the following comparison:Found A Found R Calczl:Eated Chlorine 71.92 % 73 5% 74.87 % Aluminium 15.39 % 15.27 % 14.57 % Calcium 10.47 % 10.58 % 10.50 % Iron, present as impurity in the aluminium chloride used, was not determined separately, but appears in the analysis as aluminium. I n spite of the precautions taken to ensure complete reaction, aluminium chloride still appears to be present in slight excess. The calcium salt which Baud was unable to prepare by fusion of the anhydrous salts, has thus been prepared as a phosgenate in phosgene solution. When we attempted to determine the melting point of the dephosgenated salt a portion of the aluminium chloride present sublimed, the salt decomposing according to the equation on page 966, and yielding the compound prepared by Baud. Composition of the phnsgenate. The method used by Baud' to determine the composition of the phosgenates of aluminium chloride is unsatisfactory. It was decided instead to determine the pressure concentration diagram at room temperature; the temperature chosen was 1g.5'C. Figure I represents diagrammatically the arrangement of apparatus. Fz is a Faraday tube containing the sample, fastened to the apparatus by means of the universal flat joint and clamp Jz. The tube dips into a thermostat, the Compt. rend., 140, 1666 (1905).

968

ALBERT F. 0. GERMANN AND KENNETH GAGOS

temperature of which was maintained at 1g.5'C. The manometer MI, open to the air on one side, communicates with the Faraday tube, and va,por tensions are read on it. A special form of mercury pump1 of small capacity, P2, communicates with a second manometer, M2, and through a PZOS tube and a stopcock, SU, with the system already described. Concentration changes in the solution in the Faraday tube are brought about by withdrawing any desired amount of phosgene vapor from the system by opening the stopcock S14 leading to the evacuated mercury pump. The gas drawn off is measured in the gas burette A, communicating with Pzthrough the capillary tube V,,

Fra.

I

where it is collected by operation of the mercury pump; atmospheric pressure for the reading is attained in the burette by opening stopcock 513, and manipulating the small mercury bulb B3. After measuring, the phosgene gas is discarded through the operation of a water suction pump, connected to WB, by opening stopcock S12 for a moment. Manometer M z serves a double purpose : with the mercury pump entirely evacuated, the barometric pressure is read upon it; and in the withdrawal of a portion of phosgene from the solution in Fz,the amount withdrawn is proportional to the change in pressure as registered on M2, since volume and temperature are very nearly constant; the change in pressure required to yield I cc. of gas when collected in A is determined a t the beginning of operations by a direct measurement. Samples of the double chloride were prepared in the manner already described (see page 966). Carbon monoxide, which appears to be appreciably soluble in the solution, was completely removed by prolonged boiling of the solution. All phosgene used was carefully purified (see below). Thus prepared, the tube, containing a quantity of the double salt in solution in excess of phosgene, was weighed; it was then clamped in position in the ther1

A modification of the pump described by Cermann and Cardaso; J. chim.phys.,

10, 3c6 (1912).

969

REACTIONS IN PHOSGENE SOLUTION

mostat, and its vapor tension measured; a measuyed volume of phosgene vapor was then withdrawn, as described, and the vapor tension of the remaining solution measured; and so on to exhaustion of the phosgene; the tube was then weighed again, the loss in weight representing the total weight of phosgene evolved; the tube was finally opened, the dephosgenated sa,lt removed for analysis, and the tube and excess of calcium weighed; thus the weight of dephosgenated salt was arrived at.

FIG.2 System: CaC12,2 A1 Cl8-COCl2

FIG.3 System: CaC12,2 AI C18-COC1~

The pressure concentration diagrams for samples numbered 17 and 18 are given in Figure8 2 and 3. I n these curves, the upper, nearly horizontal portion represents the vapor tension of the solution of the double chloride; in the presence of crystals, this should be constant, and hence the curve should be perfectly horizontal; but as already pointed out, (page 967), aluminium cliloride is probably present in slight excess, due to incompletc reaction of phosgene with calcium; and hence the vapor tension diminishes as the concentration of this impurity increases. Figure 3 shows one point, at 944.5 mm., considerably below the others on this portion of the curve; when this measurement was made, the solution contained no crystals; but during the subsequent evaporation of nearly 400 cc. of phosgene vapor from the solution, crystals suddenly appeared, and the vapor tension rose to 1013 mm.; the lower point corresponds to a supersaturated solution of the double chloride in phosgene. This is evidence of rather high solubility, However, no solubility determinations have been made. The second, the descending portion, of the curves, represents the drying of the crystals; in other words, the evolution of adsorbed phosgene from the surface of the crystals; t,he slope of the curve here probably depends on the size of the crystals, or, what amounts to the s h e thing, on the surface area per unit of mass. The final portion of the curves is horizontal, and represents the constant vapor tension during dephosgenation of the crystals; this amounts to about 2 5 mm. a t 19.5OC. The mass of phosgene evolved during this interval was chemically combined with the double chloride R T ; phosgene of crystallization. The calculation follows :-

970

ALBERT F. 0. GERMANN AND K E N N E T H GAGOS

Weight of aluminium chloride, Weight calcium taken, Excess of calcium, Weight calcium reacted, Weight dephosgenated salt, Volume of combined phosgene, Total volume phosgene, Weight of total phosgene, Moles combined phosgene, Moles CaCl2.2A1Cl3, Ratio salt t o phosgene,

Sample No. 17. Sample No. 18 3 , I 286 gm. 2 .2783 gm. 0.5463 gm. 0.5378 gm. 0.1159 gm. 0 . 2 1 2 5 gm. 0.4304gm. 0.3253 gm. 4.2790gm. 3.0282 am. 5 2 I cc. 375 cc. 2 4 2 1 CC. I 0 2 1 cc. I O .2309 gm. 4 . 1 8 6 0 gm. 0 . 0 2 2 2 5 mole, o ,01553 mole. 0 . 0 1 1 3 2 mole, 0.008011mole. I :I

,965

1:1.939

The ratio of salt t o phosgene is thus, in round numbers, ula of the crystals is therefore CaCI2.zA1C13.zC0Cl2.

I :2,

and the form-

Puri$cution of Phosgene. Figure 4 is a diagrammatic sketch of an allglass apparatus, as used in this laboratory for the purification of phosgene. The main supply tank (not shown) is a IOO lb. steel container, in which phosgene was supplied by Edgewood Arsenal. The supply valve of this is con-

FIG.4

nected to a 1/4"pipe line, having outlet valves at convenient points in the laboratory; another outlet valve, the exhaust valve, opens from the 1/4" pipe directly into a 3/4" waste pipe, which opens into a 2 " standpipe installed as a phosgene waste, extending above the highest part of the laboratory roof. The phosgene pressure in the 1/4"supply pipe, after shutting off t.he valve a t the main supply tank, is released by opening this exhaust valve for a moment. The glass apparatus is connected t o onc of the brass supply valves a t Wd; De Khotinsky cement is used to make a gas-tight joint with the brass. On being admitted to the apparatus, the gas traverses successively a tube of anhydrous calcium chloride, a sulfuric acid wash bottle, a tube containing

REACTION IN PHOSGENE SOLUTION

97=

antimony powder supported on glass wool, another tube containing alternate layers of phosphorous pentoxide and glass wool, and finally a tube containing a concent'rated solution of aluminium chloride in liquid phosgene, TI; the bulb above this tube must be large enough to prevent the liquid sucking back into the phosphorous pentoxide tube, and is usually made with a capacity of 500 cc., while the total capacity of the tube TI is no more than 2 5 0 cc. T1is a very useful part of the apparatus, as it provides an auxiliary storage tank for phosgene, and makes the system extremely elastic. The great solubility of aluminium chlopide in phosgene and the existence of several phosgenates of this salt combine to yield a solution whose vapor tension is much lower than that of pure phosgene. Baud1 reports that the pentaphosgenate, zA1C13.5COClz,has a vapor tension of one atmosphere2 a t 3ooC., whereas pure phosgene has the same vapor tension a t 8°C. Hence under the pressure a t which phosgene is supplied from t'he main supply tank, phosgene readily condenses in TI, and a bath of ice and water is sufficient to absorb the heat of condensation. Impurities such as HC1, CO, COz, for which no chemical absorbent is provided in the gas train, are partially removed here, after closing the supply valve, by boiling off, opening stopcocks Sz, Ss and S?for the purpose. The gases escape by bubbling through the mercury manometer VP,and are carried off through the tube W2, which communicates with the exhaust pipe for waste phosgene. Further purification is carried out in Tz and T3; a sample is distilled into Tz from T1, using liquid ammonia as refrigerant. The vapor tension at zero a t this stage is ordinarily much in excess of 5 5 7 mm., the average of the values obtained by Paternh and Mazz~cchelli~, of the somewhat higher values obtained by the Chemical Warfare Service4, whose value is given by Schaufelberger as 563 mm., and by Atkinson, Heycock and Popes who give the value 568 mm. Volatile gases are pumped off by the action of the water suction pump, which communicates with the apparatus a t W5, through the stopcock S6; the evolved gases are drawn through bottles containing sodium hydroxide before being discharged by the pump. The vapor tension of pure phosgene a t the boiling point of ammonia ( - 3 3 ' ) is given by Atkinson, Heycock and Pope as 124 mm.; the loss of phosgene is therefore not great. The vapor tension of the residual phosgene is measured from time to time, by surrounding the liquid with an ice bath, and making an approximate reading on themanometer, V z . When the vapor tension has been reduced to a vdue approximating 560 mm., three or four fractional distillations are carried out b e b e e n Tz and TB,both the first and last fractions being discarded. The liquid phosgene may be kept for some days in Tz or T3, by closing S7 or S8,and providing each of the stop cocks guarding the storage trube with clamps.

* Compt. rend.

140, 1688 (1905). We have made an initial study of t,he pressure-concentration diagram of this system, but, as the results are not in ent>ireagreement with those obtained by Baud, the problem is being subjected t o closer scrut,iny. a Gaze. chim. ita]., 50 I, 30 (1920). Rchaufelberger : Thesis, Stanford University, (1920) ; based on Edgewood Arsenal Chemical Laboratory Report No. 223. J. Chern. Soc., 117, 1410(1920).

972

ALBERT F. 0. GERMANN AND KENNETH GAGOS

The Faraday tube or other trial tube F1may be attached t o the apparatus at JI, by means of the universal flat joint and clamp. Any part of the apparatus may be evacuated with the water suction pump, by operating stopcock Sg, or with the mercury pump by operating stopcock S1. Phosgene discarded through the mercury pump is discharged through W1 into the exhaust pipe for phosgene. Ordinary rubber grease is very soluble in phosgene; this fa,ct is unimportant when pressures are low; but whenever the pressure within the apparatus approaches or exceeds one atmosphere, the grease is more or less rapidly dissolved away. Stopcocks lubricated by dusting with phosphorous pentoxide, and exposing to the atmosphere until moist, work very well; they last much longer when access of air is prevented by a ring of rubber grease at the top and bottom'. Conclusion. I n the presence of aluminium chloride, phosgene reacts with a number of metals, forming double chlorides. Magnesium and calcium especially react readily, the first yielding a two liquid layer system when cooled, the second a crystallizable product that is very soluble. The crystals were found t o have the formula CaClz.zA1C13.zCOClz. The pressure concentration diagram was determined for this system, and the decomposition pressure of the phosgenate a t this temperature found to be 2 5 mm. approximately, A method of purifying technical phosgene has been described, Acknowledgement. We wish t o take this opportunity of expressing our appreciation t o the Chemical Warfare Service for the generous supply of phosgene contributed for this research. Stanford University, Calffornia.

Guye and Drouginine: J. chim. phys., 8, 503 (19x0).