1410 ATKINSON HEYCOCK AND POPE THE PREPARATION AND CLV.- Th e Preparation und Physical Fy opert ies of Carbonyt Chloride. By RALPH HALL ATKINSON CHARLES THO~AS HEYCOCK and WILLIAM JACKSON POPE. ALTHOUGH carbonyl chloride COCl, was discovered by John Davy in 1812 (Phil. Trans. 1812 102 144) and has found extensive scientific and technical applications very little information is contained in the literature concerning its preparation and physical properties; Davy obtained the substance by effecting the combination of carbonic oxide and chlorine under the influence of sunlight and this method of preparation was elaborated by Wilm and Wischin (AnnaZen 1868 147 150). Schutzenberger (BUZZ. SOC. chim. 1869 [ii] 12 198) and Armstrong (Proc. Roy. SOC., 1870 18 504) prepared it by the action of sulphur trioxide on carbon tetrachloride ; the latter method has been recently studied by Grignard and Urbain (Compt.rend. 1919 169 17) and used by Paternb and Mazzucchelli (Gazzetta 1920 50 i 30) in their determination of the physical properties of carbonyl chloride. On the introduction of carbonyl chloride as a weapon of chemica PHYSICATI PROPERTTES O F CARRONYT CHLORIDE. 1411 warfare it became necessary to ascertain which method of pre-paration was most adaptable as a works process for manufacture, and one described by Paternb (Gazzetta 1878 8 233) presented itself as probably the best for this purpose. Paternb states that, on passing a rapid current of mixed carbonic oxide and chlorine a t the ordinary temperature over animal charcoal carbonyl chloride is produced very rapidly with considerable evolution of heat.Difficulties were encountered in carrying out the Paternb method of preparation; a number of samples of commercial animal and vegetable clzarcoalsi proved almost without effect as catalysts. Ultimately a charcoal was prepared which gave quite remarkable results in inducing the combination of carbonic oxide and chlorine ; this was made in the following manner. Fresh ox bones were crushed embedded in sand in a clay crucible and burnt in a muffle furnace; the residual charcoal was then well extracted with hot hydrochloric acid washed with water and heated as before in sand. Finally the bone charcoal was kept a t a red heat for some time in a current of dry chlorine. The charcoal prepared in this way was crushed into small frag-ments and after sifting off the dust 10 grams were filled into a U-tube immersed in a water-bath; a rapid current of mixed carbonic oxide and chlorine in which the former was in rather the greater volume was then passed over the charcoal.Carbonyl chloride was produced freely when the water-jacket was kept a t the ordinary temperature but on raising the bath temperature to between 40" and 50° combination proceeded a t a greater rate than that a t which it was possible to supply the mixed gases. Using the bone charcoal catalyst in the manner described and liquefying the carbonyl chloride by means of a freezing mixture about 10 kilograms of the compound were produced without any sign that the activity of the 10 grams of charcoal had suffered diminution.It is thus shown that Paternb's method for preparing phosgene is correctly stated but that certain precautions are necessary in the preparation of the animal charcoal used. Purther investiga-tfions showed that certain kinds of vegetable charcoal act even more vigorously in inducing the condensation of carbonic oxide and chlorine and the highly activated wood charcoal used in the Army box respirator proved more efficient as a catalyst than the bone charcoal described above. A U-tube containing 10 grams of the box-respirator charcoal which had been heated in a current of chlorine gas proved extremely efficient when the water-jacket was maintained a t 14O and one such tube kept a t this temperature yielded more than 20 kilograms of carbonyl chloride with no sign that the catalytic activity had diminished.I n the wor 1412 ATKINSON IIEYCOCK AND POPE THE PREPARATION subsequently described in this paper this active vegetable charcoal was used. The carbonic oxide required for laboratory work 011 the pre-paration of carbonyl chloride is conveniently prepared by passing carbon dioxide over red-hot coke contained in iron tubes heated in an ordinary combustion furnace; the resultant gas after passing over slaked lime and drying over sulphuric acid is practically pure carbonic oxide. I n industrial operations it is to be expected, however that the available carbonic oxide will contain hydrogen. It thus becomes of importance to compare the reactivity of carbonic oxide and hydrogen with chlorine under the catalytic influence of charcoal.A series of experiments was therefore arranged in which varying proportions of hydrogen were mixed with the carbonic oxide and chlorine passed over the catalyst. It was found that the vegetable charcoal was so active a t 14O in inducing the combination of carbonic oxide and chlorine that no advantage accrued from heab ing it to 90° and that working a t temperatures below 70° no hydrogen chloride was produced; a t SOo a small proportion of hydrogen chloride was formed and this was greatly increased a t 90°. It is thus shown that an active form of charcoal will effect the complete conversion of a mixture of carbonic oxide and chlorine into carbonyl chloride a t a temperature 50° below that a t which it begins to exhibit catalytic activity towards a mixture of hydrogen and chlorine.This conclusion seems of impcrtance in connexion with the manufacture of carbonyl chloride from carbonic oxide containing hydrogen. Carbonyl chloride dissociates into carbonic oxide and chlorine a t moderately high temperatures; Bodenstein and Dunant (Zeitsch. physikal. Chem. 1908 61 437) state that under atmospheric pressure carbonyl chloride dissociates to the extent of 67 80 and 91 por cent. a t 503O 553O and 603O respectively and that dis-sociation is complete a t 800O. The results obtained by these observers indicate that carbonyl chloride is appreciably dissociated a t 300O. A little consideration will show that if carbonyl chloride free from chlorine is to be manufactured from a mixture of carbonic oxide and' chlorine containing an excess by volume of the former, the temperature of the catalyst must be maintained below that a t which carbonyl chloride suffers appreciable dissociation.Further, it may be anticipated that within the dissociation range of temperature the catalyst will act comparatively sluggishly ; the chlorine content of the produced carbonyl chloride should thus be higher than that' indicated by the equilibrium composition and th PHYSICAL PROPERTIES OF CARBONYL CHLORIDE. 141.3 catalyst would become feebler with prolonged use. The following experiments were made for the purpose of further elucidating this question. A mixture of chlorine and carbon monoxide containing 3.5 per cent. excess by volume of the latter and1 standing over brine saturated with chlorine was dried over sulphuric acid and calcium chloride and passed over the absorbent charcoal the latter being contained in a glass tube 45 cm.in length and 1-9 cm. in diameter, a t different temperatures. For temperatures from 20° to 150° the tube was heated in an oil-bath; for temperatures from 150° to 520° the tube was wrapped in three layers of copper gauze encased in heavy iron gas pipe and heated in a gas furnace the gas pressure being carefully regulated. A quill tube was fitted into the axis of the catalyser tube so that the temperature a t any point could be measured by a copper-constantan thermo-couple. Control experiments showed that on sliding tlie thermo-couple along tshe tube the maximum temperature variation from place to place and for a period of several hours was not more than loo when no chemical action was occurring.With this arrangement of apparatus it proved easy to obtain carbonyl chloride free from chlorine a t temperatures between 50° and 200° and to collect 90 to 97 per cent. of the theoretical yield of phosgene. The important point was established however that the major part of the Combination took place at the position in the tube first exposed to contact of the mixed gases. This was further demonstrated by keeping the first or entrance half of the tube a t 18O and the second or exit half of the tube a t looo; the main disengagement of heat occurred a t the junction between the cool and the hot charcoal. By sliding the thermo-couplb along the tube the heat-ing effect of the reaction could be followed as the couple was passed along the tube from the front of the exit half to the front of the entrance half.It would be expected that on subjecting mixtures of carbon monoxide and chlorine to the action of the catalyst a certain maximum temperature would be attained which could not be exceeded-the equilibrium temperature a t which the reverse action absorbed the heat liberated. On passing the mixed gases a t the rates of 0.84 and 2.4 litres per minute the temperature of 464O was attained and remained constant; this is t(o be regarded as the maximum temperature which can be attained in the large-scale catalysing vessels used in the manufacture of carbonyl chloride. Jn the experiments just described it is noteworthy that a t the rate of flow of 0.84 litre per minute the layer of charcoal found to be at 464" was only about 3 mm.long whilst when the ga 1414 ATmNSON HEYCOCK AND POPE THE PREPBATION AND passed through a t 2.4 litres per minute the layer of catalyst main-tained a t 464O was some 10 mm. in thickness. Attempts to prepare carbonyl bromide with t,he aid of charcoal as a catalyst were unsuccessful. The Dissociation of C'arBoi:yl C'hloride. Since the combination of carbon monoxide and chlorine is highly exothermic CO + C1 = COC1 + 26C and the catalyst used is very active it seemed undesirable to attempt the investigation of the dissociation curve of carbonyl chloride by the study of the associa-tion; the foregoing experiments have shown the difficulty of main-taining the catalyst a t an even temperature during the formation of the chloride.A series of experiments was therefore made on the dissociation of carbonyl chloride for the purpose of determining the equilibrium between it and its dissociation products with temperature a t the atmospheric pressure. Since the dissociation is endothermic and must proceed to equilibrium in contact with the catalyst it should be possible to ensure that the catalyst will be kept a t an even temperature and will set up equilibrium a t that temperature. The carbonyl chloride used in all the further work described below was prepared from carbon monoxide and chlorine and pre-served in glass vessels in contact with mercury t o ensure its freedom from chlorine.The absorbent charcoal was maintained a t a constant tempera-ture as described above and a slow steady stream of dry carbonyl chloride was passed over it. The gaseous products were passed through a separating funnel A of 338 C.C. capacity entering a t the bottom and leaving through a side-arm in the shoulder; when all the air had been displaced from A the two taps were closed and a 20 per cent. potassium iodide solution was run in from a dropping funnel B stoppered into the neck of A . After well shaking the whole apparatus further quantities of potassium iodide solution were run in until action was a t an end. The solution was then washed out with water and the iodine titrated against thio-sulphate in order to determine the free chlorine present in the gas.The residual gas consisting of carbon monoxide with some carbonyl chloride and usually a little air was passed into a gas pipette and shaken with sodium hydroxide solution to remove the chloride. The carbon monoxide was determined by absorption in a hydro-chloric acid solution of cuprous chloride ; the residual air was measured and its volume deducted from the volume of the cylinder, A . Direct determinations of the chlorine carbon monoxide an PHYSICAL PROPERTIES OF CaRBONYL CHLORIDE. 1415 carbonyl chloride are thus obtained; the volumes of carbon mon-oxide and chlorine always corresponded closely and in table I, which suminarises the results the mean of the two slightly differing volumes is given. The degree of dissociation 11 of the carbonyl chloride is calcu-lated from the formula and the curve (Fig.1) is drawn by plotting the values of D against the temperature. TABLE I. Dissociation of Car bony1 Chloride. Percentage by weight. T O . l...... 101 2...... 102 3...... 151 4...... 153 5...... 208 6...... 211 7...... 237 8 . . . . . . 239.5 g...... 309 10 ...... 314 11 ...... 318 12 ...... 341 13 ...... 400 14 ...... 406 15 ...... 443 16 ...... 449 17 ...... 460 18 ...... 486 19 ...... 505 20 ...... 505 21 ...... 506 -&...... 517 '? '> 7 coc1,. 99.54 99.50 99-50 99.46 99.16 99.26 98.68 98.86 94.38 94.51 94.37 92.48 78.62 82.56 68-89 66.60 64-05 58.32 66.51 46-77 72.48 70.54 Cl,. 0.32 0.36 0.36 0.38 0.60 0.53 0-94 0.82 4.02 3.94 4-03 5-38 15.34 12.50 22.3 1 23.90 25-78 29.89 24.02 38.17 19.74 21.13 - co.0.13 0.14 0.14 0.15 0.24 0.2 1 0.37 0.33 1.59 1.55 1.59 2.14 6.04 4.94 8-80 9.40 10.17 11.79 9.47 15-05 7-78 8.33 Degreo Volume per cent. of dis-7' - sociation, COCl,. CO & Cl,. D. 99.10 0.45 0.45 99.01 0.50 0-50 99.01 0.50 0.50 99-95 0.53 0.53 98-36 0.82 0.83 98.55 0.73 0.74 97-41 1.30 1.32 97.74 1.13 1.14 89.35 5.31 5.61 89.61 5.20 5.48 89.34 5.33 5-63 68.05 6.98 7.50 64.79 17.60 21.36 70.31 14.84 17.43 52-57 23.71 31-08 49.23 25-39 34.02 47.14 26.43 35.92 41.19 29.41 41.65 49.85 25-08 33-47 30.55 34.72 53-19 56-87 21.57 27-50 54.51 22-76 29.45 The method of experimentation which has been applied has the fault that it does not ensure the dissociation products being cooled so rapidly that their partial recombination is prevented.The effect of this is seen in the fact that certain points namely those from experiments numbered 19 21 and 22 lie off the general course of the graph; in these experiments the gas current was passed much more slowly than in the others. I n consequence,'it may be found later that the degree of dissociation for any particular temperature as now recorded is too low. F o r practical purposes it may be concluded that experiments 1 to 4 referring to temperatures of looo to 150° have give 1416 ATKINSON HEYCOCK AND POPE THE PREPARA9!'TON AND identical results; the results of experiments 5 and 6 made a t t,emperatures just above 200° show a distinct increase in the degree of dissociation indicated a t looo and 150°.This increase is still more marked a t 2 3 7 O and 239O. The accuracy of the suggestion made above that carbonyl chloride free from chlorine can only be produced a t temperatures YlG. 1 . Dissociation of cnrbonyl chloride. 600' 600 400 c, 3 '$ 300 P ry" 200 100 10 20 30 40 50 Pcrccntage degrcc of dissociation. below 200° must be regarded as fully substantiated by the above experimental results. It is noteworthy that dissociation is indicated as occurring even a t 1000 and 150° by the results now recorded. It. will be seen, however that no difference is observable between the degree of dissociation indicated a t 100' and 1 5 0 O ; the indication may there-fore be attributed to some other cause than that of dissociation, possibly t o the slight reaction between potassium iodide in the acid solution and carbonyl chloride referred to below.Such a reaction would not disturb the equal volume ratio found betwee PHYSICAL PROPERTIES OF CARBONYL CHLORIDE. 1417 the carbon monoxide and chlorine and it is recorded by Bessoa (Compt. rend. 1896 122 140) that carbonyl chloride dissolves hydriodic acid with liberation of iodine. Whilst Bodenstein and Dunant (Zoc. cit.) found that carbonyl chloride is dissociated to the extent of 67 per cent a t 503O the smoothed curve (Fig. 1) expressing the Cambridge results indicates the dissociation to be 55 to 56 per cent. a t this temperature. The former workers analysed the mixture of gases by bubbling it through aqueous potassium iodide solution and it is known that pure carbonyl chloride liberates a small amount of iodine in such a solution and that the quantity of iodine formed increases on keeping (DelBpine and Ville BUZZ.SOC. chim. 1920 [iv] 27 283). It is significant that Bodenstein and Dunant obtained higher values for the dissociation in those experiments which lasted the longest (45 80 and 120 minutes) and that the dissociation-temperature curve plotted from their results slopes a t an improbable angle. The heat of formation of gaseous carbonyl chloride calculated from the degree of dissociation is +25*4 C. a t 475O and +23*4 C. a t 425O according to the Cambridge results and + 19.2 C. a t 528O and +26-6 C.a t 578O calculated from the results of Bodenstein and Dunant. Thomsen (“ Thermochemkhe Untersuchungen,” 1882, 11 364; Ber. 1883 16 2619) gives the value +26*14 C. a t the ordinary temperature and criticises adversely the value + 18.8 C. previously obtained by Berthelot. Vayour Pressure of C’arbonyl Chloride between - looo and + looo. The boiling point of carbonyl chloride is given by Beckmann (Zeitsch. anorg. Chem. 1907 55 370) as 8*2O/756 mm.; no other data concerning the vapour pressure of this compound were avail-able before the recent publication of Paternb and Mazzucchelli (Zoc. cit.) who made a series of determinations a t temperatures between - 1 9 O and +24O. We have determined the vapour pressure at temperatures from -183O to + 1 8 O in one type of apparatus and by means of another have extended the results up to +looo.The apparatus used in the determination of the vapour pressure between - 1 8 3 O and + 1 8 O is depicted diagrammatically in Fig. 2. The distillation flask F of about 60 C.C. capacity is connected with a differential manometer HB placed in front of a silvered mirror on which a millimetre scale is engraved so as to avoid parallax. Through the neck of the flask passes a thistle funnel, S for the purpose of sealing the leads of the thermo-couple air-tight by means of Faraday wax at the point where the bulb joins VOL. CXVII. 3 1418 ATKINSON HEYCOCK AND POPE THE PREPARATION AND the stem. A joint thus made was found to be very satisfactory and quite unattacked by carbonyl chloride vapour.The thermo-couple used was of copper-constantan ; one junction was kept in ice and the other in the liquid carbonyl chloride. The couple was connected to a delicate millivoltmeter which could be read with accuracy to 0.01 millivolt. For the graduation of the couple the following temperatures were assumed : TABLE 11. Melting ice ................................. - 0.00 Millivolts. Steam a t 760 mm. pressure ............ 100" + 4-07 Freezing point of mercury ............... - 38 - 1.38 Solid carbon dioxide and ether ......... - 79 - 2-65 Boiling liquid oxygen ..................... - 183 -5.12 FIG. 2. The millivolts corresponding with each of these temperatures were read off on the instrument and the results plotted. The accuracy of the calibration was confirmed by determining with the aid of the couple the boiling point of a very pure sample of ethylene; this gave the value -104O which compares well with the standard value of -103.50 given by Wroblewski and Witkowski.After some preliminary trials the following procedure was adopted. About 50 C.C. of pure liquid carbonyl chloride was dis-tilled into the flask F ; the mercury was then run out of the mano PHYSICAL PROPERTIES O F CARBONYL CHLORIDE. 1419 meter and the whole apparatus exhausted while the contents of the flask were maintained a t -80'. The temperature of the carbanyl chloride was then allowed to rise until vapour was freely given off. This process was repeated several times during the three days occupied by the experiments so as to ensure the complete absence of air.A series of observations of the pressure and temperature of the carbonyl chloride was made a t the selected temperatures with the follc~wing results (table 111) : TABLE 111. Temperature. Water-bath .................................... + 100" , , .................................... , , .................................... Z K 5 Melting ice .................................... -Ice and salt mixture - 19 Freezing point of mercury - 39 Solid carbon dioxide and ether ............ - 79 Boiling liquid oxygen ........................ - 183 ........................ .................. Pressure mm. of Hg. 16.07 x 760 5.1 1 x 760 1105-5 568.3 236.0 89.5 4-0 0 mm. These results have been plotted on the curve (Fig.3) from which the vapour pressures of carbonyl chloride for each loo between +20° and -looo have been taken; these values are included in table V. After the vapour pressure of carbonyl chloride had been measured over the range -183O to + 1 8 O recourse was had to a different form of apparatus in order to extend the observations up to looo. A sealed tube containing carbonyl chloride and a special type of manometer was heated in a water-bath to the desired temperature and kept there for two to three hours. The glass manometer is shown in Fig. 4. The graduated tube AB was carefully calibrated by means of mercury. At the same time the volume of the manometer was measured by weighing the amount of mercury which it contained. After the manometer had been thoroughly cleaned the tube AB was silvered internally.The slight difference in the weight of the manometer before and after silvering proved that the amount of silver deposited was so small that its eflect could be neglected. The manometer having been carefully cleaned and dried was filled with dry air and lowered into a hard glass tube containing about 10 C.C. of dry mercury (see Fig. 5). A wire F served to keep the manometer upright in the centre of the tube. The tube was drawn off and finally sealed after 30 C.C. of pure liquid carbonyl chloride had been introduced in the usual manner; due care was taken to ensure that the space above the liquid contained 3 s 1420 ATKINSON HEYCOCK AND POPE THE PREPARATION AND carbonyl chloride vapour only and that air was absent.The pressure tube placed inside an iron tube was kept a t 50° for two hours in a water-bath. As the pressure of the vapour increases it forces the mercury into the manometer thereby compressing the air. When FIG. 3. Vapour pressure of carbonyl chloride. 1 IT. 0.005 0.004 0.003 0.002 0.001 4- 40" + 20 0 - 20 Q u s E E - 40 F4 h - 60 - 80 - 100 _- . -- 120 0 200 400 600 800 1000 1200 14 t h e 1 2 n, Q 9 3 4 0 Pressure in mm. of mercury. mercury rises into the silvered part of the tube it dissolves the silver as far as the highest point it reaches. Thus a record is made of the volume of the compressed air at the temperature of the experiment. After the tube had cooled and a note been made of the point to which the mercury had dissolved the silver the experiment wa PHYSICAL PROPERTIES OF CARBONYL CHLORIDE.1421 repeated a t looo. The following data were obtained volume of manometer 4.675 C.C. ; barometer 758 mm. ; temperature 16.6'. Volume of compressed air at 50° = 1.038 c.c. and a t loo0'= 0.380 C.C. The following results were calculated after applying corrections for the pressure of the columns of carbonyl chloride and mercury: Vapour pressure of carbonyl chloride at 50" = 5.11 x 760 mm. Y 9 Y ? Y 9 ) 100" = 16.07 x 760mm. It was noticed that the surface of the mercury was slightly dirty FIG. 4. FIG. 5. FIa. 6. after these experiments; it is suggested that some chemical change occurs in accordance with the equation 2Hg + COCl = CO + Hg,Cb, and that the two values last given may be rather high by reason of the formation of a trace of carbonic oxide.From the vapour-pressure determinations now recorded the molecular heat of evaporation (without performance of external work) of carbonyl chloride between Oo and So is calculated as approximately 5500 calories. A molecular heat of evaporation of 5500 calories for carbonyl chloride gives a value of 2 8 O to 2 9 O for the molecular rise in boiling point of this solvent; Beckmann (Zeitsch. anorg. Ghent. 1907 55 371) has determined thi 1422 ATKINSON HEYCOCK AND POPE THE PREPARATION AND quantity experimentally as 29O. Since the lahter value is obtained as the mean of six values lying between 27-2O and 30*8O the agreement between his value and ours may be considered satisfactory.Our values for the vapour pressure of carbonyl chloride at the higher temperature are rather higher than the Italian ones whilst a t Oo and below the reverse is the case. That oiir values arc con-cordant among themselves is seen from the fact that the curve plotted between the reciprocal of the absolute temperature and the logarithm of the observed vapour pressure is pra,ctically a straight. line; this curve is indeed drawn as a straight line in Fig. 3. Since the vapour-pressure curve for carbonyl chloride only begins to fall rapidly at below -40° it is clear that considerable losses will occur in the preparation unless the effluent gas saturated with carbonyl chloride is cooled well below -4OO. The solubility of the chloride in a number of liquids from which it' might be sub-sequently recovered by evaporation was therefore ascertained in order to learn whether washing the effluent gas with a solvent would lead to economy.Solubility of Carbonyl Chloride in Or?qanic Liquids. Two or three C.C. of the solvent to be used are placed in a glass bulb of about 10 C.C. capacity immersed in a constant-temperature bath and a slow current of the gas evolved from gently boiling carbonyl chloride is passed through the bulb until no further absorption occurs. Saturation was generally complete in about three hours and the contents of the bulb were then hermetically sealed. The weight of the saturated solution was given by direct weighing and the quantity of carbonyl chloride present determined by breaking the bulb in a closed bottle containing 50 or 100 C.C.of standard sodium hydroxide solution and titrating with standard acid. The number of grams of carbonyl chloride dissolved by 100 grams of solvent are notw stated. Toluene.-At 17*0° 23.5O 30-5O and 31-5O 244.7 124.2 79.38, and 74-48 respectively. Coal-tur Xy1ene.-At 12.307 16-4O 16*9O 23.8O and 29'8O : 457.3 225.6 217.9 103-4 and 71.24 respectively. Creosote Oil.-At 1 6 ~ 2 ~ 77.42. Petroleum boiling at 180-280°.-At 12*3O 15-8O 16*7O 22-4O, 23*7O 29*9O and 30° 263.8 163.1 143.4 79-5 71-2 49.2 and 48.6 respectively. Heavy Lubricating Oil.-At 1 5 * 6 O 23*5O and 31.0° '79.7 39.3, and 24.5 respectively PHYSICAL PROPERTIES 08 CARBONYL CHLORIDE. 1423 Nitrobenzene.-106.4 a t 16.8O. a-CRloronaphthalene.-lO4-5 a t 17.0O.Ch1orobenxene.-At 12.307 16'6O 16*7O 24*2" and 29-7O 422.1, 204.3 221.6 99.9 and 81-9 respectively. Acetylene Tetrachloride.-At 16.807 25.1° and 29.9O 149.7, 89.4 and 74.9 respectively. The solubility of carbonyl chloride differs widely according to the solvent. Toluene coal-tar xylene and chlorobenzene are by far the best solvents of those examined and in view of their elevated boiling points would appear to offer most advantages as scrubbing agents for efluent gases containing carbonyl chloride. Ordinary burning petroleum boiling a t 180° to 280O' is the next best solvent and heavy mineral lubricating oil and acetylene tetrachloride are not quite so good. MeZti?ag and Freezing Points of Carbonyl Chloride. The only recorded melting point for carbonyl chloride is that given by Erdmann (Annalen 1908 362 148) as -118O.Our determinations were made by immersing the bare junction of a standardised copper-constantan thermo-couple in pure liquid carbonyl chloride; on cooling in a bath of liquid air the carbonyl chloride solidified to a white crystalline mass and on allowing the temperature to rise slowly the temperature as recorded by a milli-voltmeter remained steady a t -126O throughout the melting of the substance. On cooling the liquid very slowly with constant shaking the carbonyl chloride commenced to crystallise at - 128O. Density of Liquid Car bony1 Chloride. The density of liquid carbonyl chloride was determined by Emmerling and Lengyel (Annalen Supp. 1870 7 106) as 1.432 a t 0°/40 and as 1.392 a t 18*6O/4O; Paternb and Mazzucchelli (loc.cit .) have quite recently made a series of careful determinations between - 15'4O and + 5 9 ~ 9 ~ . We have determined the density a t -104*Oo -79O O" and +49*90 with the aid of weight thermo-meters of transparent silica; two such thermometers A and B, were used. After filling the thermometers with mercury and repeatedly boiling out they were cooled for some hours in ice dried in a vacuum desiccator and weighed precautions being taken to collect the mercury overflow; thermometer A contained 80.1054 grams of mercury a t Oo whilst B contained 67.2640 grams. Assumin 1424 ATKINSON HEYCOCK AND POPE THE PREPARATION AND Regnault's value of 13.5955 grams for t3he weight of 1 C.C. of mercury at Oo the capacities of A and B are calculated as 5.8920 and 4.9475 C.C.respectively. Thermometer A was filled with mercury a t Oo heated in a steam hypsometer (Bar. 768 mm.) for an hour and then weighed; it contained 78.6835 grams of mercury a t 100-3O. From these values, the coefficient of apparent expansion of mercury in silica is calcu-lated as (1.4219) / (78.6835 x 100.3) = 0~000,180,17 ; assuming Regnault's value of 0*000,180,92 for the coefficient of absolute expansion of mercury between 00 and looo the coefficient of cubic expansion of transparent silica becomes 0*000,000,75 a quantity which is small enough to be neglected. I n making the density determinations a t Oo and lower tempera-tures the thermometer A was filled with liquid carbonyl chloride in the ordinary way and kept a t the steady low temperature for for half to one hour; no attempt was made to weigh the vessel on account of the difficulty of preventing moisture from the air con-densing on it during transference.The weight of carbonyl chloride was found by removing the tube from the cooling bath and lower-ing it into a bottle containing 200 C.C. of ice-cold standard ( 2 N ) sodium hydroxide which was then stoppered and clamped. The liquid in the thermometer slowly boiled off usually taking from one to two1 hours and when all had evaporated and reacted with the alkali the excess of the latter was determined by titration with an acid; the weight of carbonyl chloride was then calculated. I n making the determination at +49*9O the weight thermo-meter B was filled a t -79O and was then lowered nose down-wards into a stout combustion tube the lower end of which contained mercury.Cold carbonyl chloride was poured into the tube which was then drawn off and placed inside an iron tube (see Fig. 6) gradually heated to 4 9 ~ 9 ~ in a water-bath and kept a t that temperature for three hours. The tube was then slowly cooled finally in ice and salt and opened in the same way as in an ordinary pressure-tube experiment. The thermometer, which now contained mercury and carbonyl chloride was with-drawn from the tube drenched on the outside with cold ether, placed in the standard sodium hydroxide solution and allowed to become warm slowly; the titration was then performed as before. All errors due to the deformation of the silica tube by pressure were in this way eliminated as the pressure within and without the tube was practically the same.The density of carbonyl chloride (that is weight in grams of 1 c.c.) was determined a t -104O (boiling point of ethylene) -79O (solid carbon dioxide wetted with ether) Oo (ice) and +49*9 PHYSICAL PROPERTIES OF CARBONYL CHLORIDE. 1425 (thermo-regulator) ; the results of the actual experiments are given in tabular form (table IV). TABLE IV. Weight Weight of COCl, Temperature. thermometer. in grams. C A -104'0" t(Vo1. 5.892 c.c.)) 9'892 - 79 Y ? 9.519 - 79 Y Y 9-503 0 9 9 8463 0 8.443 6-502 i5 +49.9 { (Vol. 4.948 c.c.)} Density of COC1,. (Weight of 1 C.C. of liquid. ) 1-679 1-616 1.613 1.436 1.433 1.314 From these results a smooth curve was drawn and a table (table V) constructed giving the density of carbonyl chloride for each loo from -looo to +50°.TABLE V. Density and Vapour Pressure of Liquid Car bony1 Chloride. Tempera-ture. -110O - 100 - 90 - 80 - 70 - 60 - 50 - 40 - 30 Density. Grams per C.C. 1.685 1-663 1.640 1-617 1.594 1.572 1 $49 1,526 1.504 Vapour pressure. Mm. of mercury -4 11 24 47.5 85 141 Tempera- Density. ture. Grams per C.C. - 20 1-481 - 10 1.459 0 1.435 1.412 $-lo 1.388 +" + 30 1.363 + 40 1.338 + 50 1.314 Vapour pressure. Mm. of mercury. 226 361 568 844 1212 -I 5-11 x 760 The mean coefficient of cubical expansion of carbonyl chloride between -79O and +49*9O is calculated as (1.6145 - 1*3140)/(1*3140 x 129) =0*001,77. The values for the temperature ranges -1040 to -79O -79O to Oo and Oo to +49*9O are 0~001,60 0*001,59 and 0.001,84 respectively. The values for the density of liquid carbonyl chloride which we now give are slightly higher than those obtained by the Italian workers; the latter obtained their product by the action of sulphur trioxide on carbon tetrachloride and it is curious that Ernmerling and Lengyel who prepared their material as we did by the com-bination of carbonic oxide and chlorine obtained results for the densities which are almost identical with ours for the two tempera-3 F 1426 PYMAN AND RAVALD: tures a t which they worked. The discrepancies Letween the results obtained by Paterni and Mazzucchelli and by ourselves may possibly be traced to the presence of some characteristic impurity in the carbonyl chloride made by one or the other method. The work described in the present paper was carried out for the purposes of the Chemical Warfare Department and permission for its publication has been given by the General Staff. Our thanks are due to our assistant Mr. George Hall for much help in experimental work. THE CHEMICAL LABORATORY, UNIVERSITY OF CAMBRIDGE. [Received October lBth 1920.