February, 1941
INDUSTRIAL A N D ENGINEERING CHEMISTRY
sis, samples of pure 3-chloropropene and l-chloropene were prepared and subjected to hydrolysis. Samples of the pure compounds were weighed in bottles fitted with ground-glass stoppers to eliminate evaporation. The bottle and sample were placed in a tube containing 25 ml. of a standardized alcoholic base which had previously been cooled in ice, and the tube was sealed. The glass stopper was removed from the weighing bottle, and the contents of the tube and bottle were thoroughly mixed by shaking. The results in Table VI show that a t 70" to 75" C. l-chloropropenes are not hydrolyzed a t all, but that 3-chloropropene is completely hydrolyzed. This same technique was used in analyzing mixtures of chloropropenes obtained by dehydrochlorination of l,Zdichloropropane, and the results are reported in Table VII. Three different temperature ranges were used, 540-550 ", 635-640", and 750" C. At 540" C. the best yield of chloropropenes was obtained with an exposure time of 2.30 seconds. Either longer or shorter exposure times caused a decrease in the yield. When the temperature was increased to 635" C. a maximum yield of 76 per cent was obtained with an exposure time of 0.85 second. The ratio of chloropropenes did not seem to vary appreciably with temperature, and 3-chloropropene was approximately 60 per cent of the total chloro-olefins. Catalysts
181
caused a lower yield of desired products even a t low conversion.
Literature Cited (1) Cahours, Ann., 76, 283 (1850); Beilstein's Handbuch der organischenChemie,Vol.1,p. 108, Berlin, Julius Springer, 1918. (2) Dibble, J . Econ. Entomol., 26, 893 (1933). (3) Essex and Ward, U. S. Patent 1,477,047(Dec. 11, 1923). (4) Faraday, Ann., [2]18,53 (1821). (5) Geuther, Ibid., 107,212 (1858). (6) Goldschmidt, Ber., 14, 929 (1881). (7) Hartmann, Ibid., 24,1011-26 (1891). (8) Hass, McBee, Hinds, and Gluesenkamp, IND.ENG. CHEM., 28, 1178 (1936). (9) Hutson, J . Econ. Entomol., 26,291 (1932). (10) Klebanskii and Vol'kenshtein, J . Applied Chem. (U. S . S . R.), 8, 106-16 (1935). (11) Krafft and Merz, Ber., 8, 1296 (1875). (12) Munro and Fox, N. Dak. Agr. Expt. Sta., Bull. 278 (1934). (13) Prins, J . prakt. Chem., 89,418 (1914). (14) Reed, U. S. Patent 1,969,183(Aug. 7,1937). (16) Regnault, Ann., 33, 324 (1840). (16) Sabatier and Mailhe, Compt. rend., 138,409 (1904). THISpaper contains material abstracted from the Ph.D. theses of T. H. Chao and L. E. Thomas, and the M.S. thesis of Z. D. Welch.
CHLORINOLYSIS OF CHLOROPENTANES E. T. MCBEE, H. B. HASS, AND EARL PIERSON Purdue University and Purdue Research Foundation, Lafayette, Ind.
A high-temperature high-pressure chlorination of chloropentanes has resulted in the formation of hexachloroethane, carbon tetrachloride, and hexachlorobutadiene. The term "chlorinolysis" is proposed as being descriptive of the reaction. Explosions and carbonization are prevented by the use of excess chlorine. At 400' C. (752' F.) and a pressure of 1000 pounds per square inch the conversions to carbon tetrachloride and hexachloroethane are approximately 90 per cent, and the potential yield of these products approaches the theoretical value.
T
H E terms hydrogenolysis, hydrolysis, ammonolysis, and alcoholysis occur frequently in chemical parlance and indicate a chemical reaction in which hydrogen, water, ammonia, or alcohol, respectively, effects a rupture of the molecule undergoing these reactions. Following this analogy, the term "chlorinolysis" is suggested as affording an apt description of a chlorination reaction performed under conditions which cause a rupture of the carbon-carbon bonds in the reactant molecules to form chloro compounds with fewer carbon atoms. The chemical literature contains only a limited number of examples of chlorinolysis. Hartmann (3) chlorinated 2,5-
dimethylhexane, 2,7-dimethyloctane, l-iodohexadecane, hexadecanoic acid, and Pennsylvania petroleum with antimony pentachloride in sealed tubes a t temperatures of 60-320" C. (140-608" F.) and obtained carbon tetrachloride, hexachloroethane, hexachlorobenzene, and a compound having the empirical formula C4Cls (presumably hexachlorobutadiene). Krafft (4) chlorinated l-iodobutane and l-iodohexane in sealed tubes with iodine trichloride at 100-240" C. (212464" F.) and obtained carbon tetrachloride, hexachloroethane, and a compound which was considered to be hexachlorobutadiene. The only reference to a chlorinolysis reaction in which elemental chlorine was used as the chlorination agent is that of Grebe, Reilly, and Wiley (2) who patented a process for the preparation of chlorocarbons, principally carbon tetrachloride and tetrachloroethylene, by passing preheated chlorine and paraffin mixtures through a molten salt bath. From the limited number of references to chlorinolysis it is evident that this reaction has received but little attention, presumably because of the difficulties involved, the lack of industrial potentialities in chlorination by means of compounds of chlorine, and the frequent occurrence of carbonization and explosions when large amounts of chlorine are mixed with paraffins.
Explosibility of Dichloropentanes-Chlorine Mixtures
It was desired to apply the chlorinolysis reaction to the isomeric dichloropentanes derived from the thermal chlorination of mixtures of pentane and isopentane; however, there was considerable apprehension as to the possible consequences of
INDUSTRIAL AND ENGINEERING CHEMISTRY
182
heating under pressure mixtures of chlorine and dichloropentanes containing large amounts of the former reactant. It is known that chlorine forms explosive mixtures with hydrocarbons (1) and that, as a hydrocarbon is progressively substituted with chlorine, the violence of the explosire reaction diminishes and undoubtedly the explosive range becomes narrower. Since the most economical operation of the chlorinolysis reaction is a one-step process in which all of the required amount of chlorine is introduced a t one time, i t was essential that the explosibility of mixtures of chlorine and dicliloropentanes be inrestigatetl.
FIOURE 1. APPARaTUS
FOR THE DETERMIIi.4TlON EXPI,OSIVE LIVITS
03
This information wab obtained by pumping various mixtures of chlorine and dichloropentanes through a n electrically heated nickel tube against a pressure of 1000 poundsper square inch. The apparatus (Figure 1) employed in this dynamic method of determining the explosive limits was similar to the one eventually used in the chlorinolysis reaction, with the exception that more elaborate means were taken to accommodate the anticipated explosions. The mixtures with the desired reactant ratios were obtained by admitting dichloropentanes into the reactants reservoir, and by the proper manipulation of valves an appropriate quantity of chloriiie v a s distilled from the chlorine weighing capsule into the reactants reservoir. Chlorine and dichloropentanes are miscible in all proportions at room temperature, and the reactants reservoir was either removed and shaken or allowed to stand for a t least 10 hours to ensure homogeneity of the reactants. The various mixtures were pumped through the reaction tube having a weakened section while the temperature in the tube m-as being increased from room temperature to 400" C. (752" F.) and/or until a n explosion occurred. By this means the lower explosive limit mas established as being approximately 6 moles of chlorine per mole of dichloropentanes, and the upper limit was located in the proximity of 9 moles of chlorine per mole of dichloropentanes. The establishment of the explosive range below the theoretical amount of chlorine required to substitute the hydrogen atoms remaining in the dichloropentanes molecules marked a definite progress toward the solution of the problem for i t made possible all of the economies attending mixing the total quantities of reactants and obviated the need for a inultiple stage process involving recycling. It is considered that the use of excess chlorine as a diluent is a n innovation in high-pressure chlorination technique. The explosive limit determinations were made in a tube heated by a resistance mire wrapped around the tube. It is possible that the explosive range would have been obviated
Vol. 33, No. 2
or narrowed had provision been made for more rapid dissipation of the heat liberated by the exothermic reaction.
Apparatus One of the most troublesome phases of this investigation involved pumping liquid chlorine against superatmospheric pressure. Attempts to pump chlorine with a proportioning pump in the conventional manner met little success. Although several gland packings were testad, none was found which would resist the action of liquid chlorine for more than 30 minutes of operation. The chlorine rapidly dissolved the packing lubricant, causing either prohibitive leakage or piston galling, depending upon the gland pressure. The method rrhich was successfully employed is shown iii detail in Figure 2. The chlorine reservoir was filled by opening the valve of the inverted chlorine cylinder and allowing t,he liquid to flow from the cylinder into the reservoir. Cold tap water was run through the mater jacket surrounding the chlorine reservoir to lower the ressure of the chlorine, thereby facilitating the transfer vapor of the I)iquid. ' The gas ballast, was added to the chlorine reservoir to preclude the possibility of a liquid-filled closed system. Since the vapor pressure of chlorine in the chlorine level gage was greater than that in the reservoir, nitrogen was admitted through valve 2 until the total pressure in the reservoir was approximately 175 pounds per square inch. By proper manipulation of valves the chlorine level in the reservoir was indicated by the glass capillary level gage. In pumping liquid chlorine it is esscntial that the cones and Beats of the check valve make an intimate contact because of the low viscosity of the liquid. The glass wool filter was inserted in the chlorine line to remove small solid particles which, prior to this modification, frequently interfered with the successful operation of the check valve. A ball-type check valve was found to be inferior to a cone type for use in the chlorine line. The pressure of 175 pounds per square inch in the chlorine reservoir precluded the possibility of vapor lock in the chlorine check valve. Concentrated sulfuric acid was placed in the line connecting the check valve with the pump body. This acid and liquid chlorine have low mutual solubility, and there was no appreciable loss of acid not attributable t o the slight leakage by the pump packing. After protecting the pump packing by sulfuric acid and the check valve by the glass wool filter, no difficulty was encountered in pumping liquid chlorine against pressures as high as 1000 pounds per square inch.
GAJ MLLAJT
4-
JS OL
7WNlNG
WMP
FIGURE 2.
APPAR.4TVS FOR P U M P I N G
LIQUIDCHLORITE
Thc chloropentanes were pumped in the conventional manner with the other unit of the dual proportioning pump from the chloroparaffin reservoir (Figure 3). The establishment of flow in the chloropentanes line was visually demonstrated by means of the drop meter located between the chloroparaffin reservoir and the check valve, and the chlorine flow was evidenced by the fall in liquid in the chlorine level gage at each intake stroke of the pump.
February, 1941
INDUSTRIAL AND ENGINEERING CHEMISTRY
The streams of reactants converged at a tee and ascended the vertical reaction tube. The product then passed through the condenser and into the receiver, The reaction tube, which was fabricated from 43 inches of 1/2-inch iron-pipe-size cold-drawn nickel tubing, was surroundedby a bath containingequimolecular quantities of sodium nitrite and potassium nitrate heated by two 750-watt immersion heaters. The temperature of this bath was measured by three thermocouples located at various depths in the bath. The desired pressures in the system were obtained by filling the reactor and receiver with nitrogen and allowing the nitrogen to escape through the exhaust valve as product collected in the receiver. The apparatus was fabricated from ferrous metals with the exception of that portion between the mixing tee and the receiver which was of nickel t o resist the corrosive action of hot chlorine and hydrogen chloride. The safety blowout mounted on the top of the reaction tube was made from a standard */pinch iron-pipeshe nickel nipple machined to an outside diametei of 0.637 inch. This outside diameter corresponds to a wall thickness of 0.007 inch, and the resulting weakened portion ruptures at approximately 2000 pounds per square inch
Methods of Analysis and Identification of Products At the conclusion of a run the semisolid yellow product was drained from the receiver, and the hydrogen chloride and excess chlorine were allowed to boil off. The refrigeration provided by the vaporization of these liquids was found to be sufficient to prevent the loss of substantial amounts of carbon tetrachloride. The residual organic material was steamdistilled to remove carbon tetrachloride and hexachloroethane, and these components were subsequently separated by distillation. The residue from the steam distillation was dried and then distilled under reduced pressure. By the latter manipulation a fraction boiling a t 140" C. (10 mm. pressure) was obtained. Upon cooling, this material crystallized; and after recrystallization from 95 per cent ethanol, it had a melting point of 38.0" C. The material exhibited a pronounced tendency to supercool, and the liquid had an index of refraction (D line) of 1.5660 and a density of 1.817 grams per ml. a t 50" C. A Carius determination gave values of 81.6 and 81.4 per cent chlorine. A literature survey indicated that the material was a chlorocarbon having the empirical formula C4C16, presumably hexachlorobutadiene (8, 4 ) . This compound is reported as having a melting point of 39" C., a boiling point which calculates to 139" C. a t 10 mm., and contains 81.58 per cent chlorine. Further confirmation of this structure was obtained when it was found that the same material resulted from the chlorinolysis of chlorides derived
183
from isopentane and several higher paraffins but was not obtained from the chlorinolysis of isobutane and lower hydrocarbons. This work will be discussed in detail in a subsequent paper. The identity was considered to be established when it was found that the application of the LorenzLorentz equation to the refractive index and density data calculated to a molecular refractivity of 46.82 as compared to the theoretical value of 48.94 for hexachlorobutadiene. When the small amount of the solid residue from the distillation of the hexachlorobutadiene was recrystallized from benzene, a trace of hexachlorobenzene was always obtained. This latter compound never comprised over one per cent of the product, and since it was an undesired material, no attempt was made to augment its formation.
Discussion of Reaction After several preliminary chlorinolysis experiments i t became evident that the relative amounts of hexachlorobutadiene and hexachloroethane could be varied a t will over a rather wide range, but the production of carbon tetrachloride was independent of this variation. When it was found that the chlorinolysis of hexachlorobutadiene gave a 75 per cent conversionto hexachloroethane without a detectable formation of carbon tetrachloride, it was concluded that hexachlorobutadiene is an important intermediate in the chlorinolysis of chloropentanes to carbon tetrachloride and hexachloroethane, and the reaction proceeds, to a substantial extent, according to the following equations:
+
C6HsCL f 7C12 --c ChCle SHCl CdCle 3C12 --L 2CzCla
+
+ CCl,
This conclusion is in harmony with the fact that the more drastic the conditions as to temperature, pressure, and exposure time, the greater the formation of hexachloroethane a t the expense of hexachlorobutadiene. Operating data and analyses of products of representative runs made in the chlorinolysis of tetrachloropentanes are given in Table I. The graphs which will be considered subsequently are based upon these data. It was found impossible to preclude the formation of small amounts of carbonaceous material when dichloropentanes were used as the starting material. This situation might be circumvented by maintaining a temperature gradient in the reaction tube; however, it was found that further substitution of dichloropentanes to the tetrachloride stage gave a starting material which resulted in substantially no carbonization under optimum conditions, and all data included in this paper were obtained by using material having the average composition of tetrachloropentanes.
Effect of Exposure Time
I FIQIJRE 3. CHLORINOLYSIS APPARATUS
As has been noted, the treatment of chloropentanes with chlorine yields as organic products polychloropentanes, hexachlorobutadiene, carbon tetrachloride, and hexachloroethane in relative amounts which can be widely varied. Of these products only the latter two were of interest in this investiga-
INDUSTRIAL A N D E N G I N E E R I N G C H E M I S T R Y
184
Vol. 33, No. 2
TABLE I. CHLORINOLYSIS OF POLYCHLOROPEKTANES Run No. Temperature
c.
F. Pressure, Ib./sq. in. Chlorine, moles CsHaClr, moles Mole ratio Moles/hr. P r o d u c t , % by weight CC14 clcl6 C4C16 Residue
;s5}
30
31
33
37
38
39
40
42
425 797
500 932
450 842
425 797
426 797
428 797
425 797
425 797
1000
1000
1000
32 30 32 1.7 1.0 1.4 19:l 2Q:l 22:l 16.9 20.7 15.9 14 72 14 0.5
14 49 33 4
17 60 21 2
43
425 797 1000 1000 1000 1000 1000 1000 34 23 30 23 23 26 1.7 1.4 1.3 2.0 1.5 1.4 20:l 1 6 : l 2 3 : l 12:l 15:l 19:l 17.8 12.2 31.3 18.8 15.5 18.0 18
64 16 2
17
20 62
43 3; -
... .
20 61
17 56 25 1
,
,
I
I
44
45
48
49
50
51
52
53
54
425 797 1000 24
425 797 1000 16
400 752 1000 32 1.6
22:l 23.2
15:l 17.1
2O:l
376 707 1000 32 1.6 2O:l
330 626 1000 16 0.8 20:l 16.8
400 752 800 24 1.2 20:l 16.8
400 752 600 30 1.6 l9:l 15.8
400 752 400 26 1.4 18:l 15.7
400 752 800 31 1.6 2O:l 16.3
19 66 14 1
18 68 12 0.3
21 68 10
10 35
12 76 8 4
19 58 22 1
10 40 49 0.5
10 69 19 2
1.1
24 62 13 1
18
2
' I 1 l '
75
18 24 M O L E S 1 H R.
I
65
I
12
16.8
1
16.8 18
57 24 0.5
.. ,.
851--i
E
U
02651*Jj\
1.1
30
I
,
l
of runs in which greater chlorine excesses were employed. It is evident that the excess of chlorine prevented untoward carbonization, presumably by diluent action, and reduced the hexachlorobutadiene content of the product by further chlorination of this material. The location of a maximum plateau on the mole ratio-com-
FIGVRE4. EFFECT OF EXPOSURE TIMEAT 425' C. AND 1000 POUNDS PER SQVARE INCH
tion. After a series of runs had been made under widely varyconcentration of chlorinein theliquid phase are desirable for the ing conditions, the preferred temperature for performing the optimum performance of the chlorinolysis reaction. Using a temperature of 400" C. and a mole ratio greater than 16:1, the chlorinolysis reaction mas located a t approximately 425" C. hydrogen atoms of tetrachloropentanes were completely re(797' F.); with this temperature and a mole ratio greater moved at all of the pressures investigated; however, the comthan 16 moles of chlorine per mole of tetrachloropentanes, a series of runs was made in which the exposure time was varied. position of the products varied widely with pressure changes The percentage of carbon tetrachloride and hexachloroethane (Figure 6). At a pressure of 400 pounds per square inch the (Figure 4) remained constant as the total reactants flow was product contained 49 per cent hexachlorobutadiene and 50 per cent carbon tetrachloride and hexachloroethane. As the presincreased from 12 to 23 moles per hour (volume of reactor, 210 cc.) but decreased sharply when the latter rate was exceeded. Of the runs made with too brief an exposure time the part of the product, which was not carbon tetrachloride and hexachloroethane, was principally hexachlorobutadiene. It seems probable that the further chlorination of hexachlorobutadiene SO-necessitates more drastic conditions than are required to substitute the hydrogen atoms of ( I , I I , I the chloropentanes. $00 700 or? 3 GO 400 500 .-The reproducibility of conditions and acPRESSURE TEMPERATURE curacy of the analyses probably were not suffiOF PRESSVRE AT FIGURE 7. EFFECT OF TEMPERATURE FIGURE 6. EFFECT ciently high to give data for a curve which would 400" C. AT 1000 POTJNDS PER SQU.4RE INCH extrapolate to 100 per cent carbon tetrachloride and hexachloroethane as the exposure time was sure mas increased, the formation of carbon tetrachloride and increased. The establishment of a plateau on the exposure hexachloroethane increased a t the expense of hexachlorobutatime-composition curve indicates that rigorous control of diene until, a t a pressure of 1000 pounds per square inch, the rate of flow is not essential to high conversions. carbon tetrachloride and hexachloroethane constituted 89 per cent of the product, the remainder being hexachlorobutadiene. Effect of Mole Ratio From these data it is evident that, when carbon tetrad study of the effect of mole ratio upon the conversion t o chloride and hexachloroethane are the desired products from carbon tetrachloride and hexachloroethane was made in order the chlorinolysis reaction, pressures less than 1000 pounds t o determine the lowest chlorine excess compatible with high per square inch are incapable of eliciting the optimum results. conversions to the desired products. There appears to be Attempts to perform the reaction a t atmospheric pressure were no advantage (Figure 5) in using mole ratios in excess of 16 discouraging since considerable carbonization and only a commoles of chlorine per mole of tetrachloropentanes since there paratively small amount of chlorination was obtained. was only a negligible change in product composition as the mole ratio was increased from 16: 1t o 22: 1. Runs made using Effect of Temperature a mole ratio less than the limiting value of 16: 1 gave products An investigation t o determine the effect of temperature on which were dark in color and contained higher contents of the chlorinolysis of tetrachloropentanes a t a pressure of 1000 carbonaceous material and hexachlorobutadiene than those
February, 1941
INDUSTRIAL AND ENGINEERING CHEMISTRY
pounds per square inch and a mole ratio of 19-22:l (Figure 7) revealed that a t 330" C. (626" F,) polychloropentanes and hexachlorobutadiene constituted a substantial portion of the product. As the temperature was elevated to 375" C. (707' F.) the polychloropentane content of the product became essentially nil; however, hexachlorobutadiene still constituted 24 per cent of the product. At 400" C. (752O F.) a maximum carbon tetrachloride and hexachloroethane content of 89 per cent was obtained, and a t this temperature the hexachlorobutadiene content was 10 per cent. It might be anticipated that further increases in temperature would be attended by a continued decrease in the hexachlorobutadiene content of the product. Actually it was found that the products from runs made a t 450" and 500" C. (842" and 932" F.) contained larger amounts of hexachlorobutadiene than did those runs made a t 400" C. This apparent anomaly is probably due to the decreased chlorine concentration in the liquid phase a t the higher temperatures.
185
hexachlorobutadiene. I n addition to these conversions to the desired products, it was shown that hexachlorobutadiene can be converted to hexachloroethane by further chlorinolysis. The over-all yields of carbon tetrachloride and hexachloroethane would approach the theoretical value in a process involving further conversion of hexachlorobutadiene.
Acltnowledgmen t The authors are indebted to the Sharples Solvents Corporation for defraying the expense of this investigation. J. F. Olin (Sharples Solvents Corporation) and G. A, Hawkins (Purdue University) made suggestions concerning this investigation which were of considerable value.
Evaluation of the Chlorinolysis Reaction
Literature Cited
The observation was repeatedly made that the proportion of carbonaceous material in the product and the carbonaceous deposit on the reaction tube walls were always very low when the optimum operating conditions were employed. These observations are in accord with the fact that it has been possible to isolate 95 per cent of the carbon contained in the starting material as carbon tetrachloride, hexachloroethane, and
(1) Brooks, B. T., IND. EN*. CHEM.,17, 752 (1925). (2) Grebe, J., Reilly, J., and Wiley, R., U. S. Patent 2,034,292 (March 17, 1936). (3) Hartmann, E., Be?., 24, 1011 (1891). (4) E ~ K.,~ Ibid., ~ , (1877). BASEDupon a thesis submitted b y E a r l Pierson t o t h e faaulty of P u r d u e University in partial fulfillment of t h e requirements for t h e degree of doctor of philosophy, 1941.
HIGH-PRESSURE CHLORINATION OF PARAFFINS E. T. MCBEE, H. B. HASS, AND J. A. PIANFETTI Purdue University and Purdue Research Foundation, Lafayette, Ind.
T
H E discovery (8) that liquid-phase chlorination of paraffin hydrocarbons yields a higher percentage of primary substitution products than are obtainable in the vapor phase at the same temperature led to an investigation of highpressure chlorination a t elevated temperatures (1). During this investigation i t was found that in the chlorination of propane at 300" C. there is a n increase in the yield of the primary substitution product with a n increase in pressure. This discovery was advanced tentatively as chlorination rule 11: "In vapor-phase chlorination of saturated hydrocarbons, increased pressure causes increased relative rates of primary substitution.'' Since this work is not only of considerable theoretical interest but also of potential practical value in obtaining higher yields of the generally more desirable primary substitution products, the investigation-has been extended with several hydrocarbons to include pressures as high as 4000 pounds per square inch. Figure 1 shows a drawing of the apparatus used. A charge was prepared by first introducin the material to be chlorinated into the mixing and supply tan%, 9. To obtain a known and desired amount of chlorine, a tank of liquid chlorine was attached
Liquid-phase chlorination of paraffin hydrocarbons yields a higher percentage of primary substitution products than are obtainable in vapor phase at the same temperature. Thus, in the thermal chlorination of paraffin hydrocarbons, high pressures may be desirable in order to maintain liquid phase. In vapor-phase chlorination of saturated hydrocarbons, increased pressure causes increased relative rates of primary substitution. Since there seems to be a direct correlation between the change in the relative chlorination rate and in the molar volume of the hydrocarbon with increasing pressures, it appears that pressure alters the relative chlorination rate by effecting a greater absolute concentration of the hydrocarbon in the vapor phase.
at 2 and the vapor was allowed to flow into a cooled receiver 1, where it condensed. The receiver was weighed before and aker filling by detaching it from the apparatus at union 3. When the desired amount had been collected, the valves in the line were opened t o allow the chlorine to vaporize and pass into tank 9 where solution with the material t o be chlorinated occurred. A bath of warm water was applied t o receiver 1 to facilitate com-
