PREPARATION OF CHLOROFLUOROCARBONS FUuorination of €+ins=&ype condensation produets Rl. Couper, F. B. Downing, R. N. Luleli, RI. A. Perkins, F. B. Stilmar, and W. S. Struve E. I. DU PONT D E NEMOURS & COMPANY, INC., WILMINGTON, DEL.
STABLE perhalogenated products with 6 to 20 carbon atoms were prepared by stepwise build-up and fluorination from simple starting materials such as carbon tetrachloride, trichloroethylene, and trifluorotrichloroethane. To minimize elemental fluorine requirements, preliminary fluorinations with hydrogen fluoride and especially antimony pentafluoride were utilized. Final treatment with cobalt trifluoride fluorinated the residual double bonds and completed replacement of the small number of hydrogen atoms present.
The chlorohexene shown in Equation 2 was first reported by Prins and has now been found t o condense readily with trichloroethylene a t 25-30" C. presumably as illustrated. KO proof of structure of the octene has been attempted. Removal of more than 2 moles of HC1 from the octene resulted in decomposition. The second type of condensation, dechlorination with copper bronze, was used on olefins which contain fluorine. Such olefins are not susceptible to the Prins reaction since they convert aluminum chloride to the inactive aluminum fluoride ( 5 ) . Owing to the impossihlity of close control of the extent of the condensation, coniplex mixtures n-ere obtained, but the products were suitable for subsequent reaction with antimony pentafluoride. Equation 3 illustrates such a condensation of the product mixture from the reaction of hydrogen fluoride and dodecachloroheptene prepared in Equation 1:
S
IKCE the preparation of elementary fluoiine for the production of metallic fluorides is relatively expensive, it was desirable to effect economies in fluorocarbon manufacture by niinimizing its use. This study vias therefore undeitaken in an effort t o prepare stable polyhalogenated hydrocarbons with the smallest possible amounts of cobalt trifluoride. For this purpose chlorinecontaining aliphatic compounds of more than 6 carbon atoms and of low hydrogen content were subjected to treatment mith antimony pentafluoride. The mixture of products, n hich retained varying amounts of hydrogen and chlorine, and usually contained unsaturated compounds m well, was then converted t o stable saturated chloroperfluorocarbons and perfluorocarbons with a minimum requirement of cobalt trifluoride. The necessary quantity of antimony pentafluoride could also be reduced by preliminary hydrofluorination with anhydrous hydrogen fluoride in the presence of antimony pentachloride.
C7H2C112E-+ C7HFOC14(av.) Cu 3SbCls 250" C. C Mand C21 compounds (13% C1, 45.5% F)
(3)
Condensation of 2 or 3 molecules of partially fluorinated dodecachloroheptene is indicated here from molecular weight determinations. The products contained about 1 chlorine atom for each C7 unit, which constitutes considerable dechlorination. The weight yield of the copper condensation was 447". The copper-bronze condensation of trifluorotrichloroethane (Equation 4) gave, in about 50% yield, a high boiling mixture of products in which the ratio of carbon, fluorine, and chlorine atoms was about 8 to 7 to 1:
STARTING MATERIALS
CFC12-CF2Cl
TWOtypes of condensation were used to prepare compounds of halogen content sufficiently high to resist degradation during reaction with antimony pentafluoride. The first, the Prins reaction (6, 7, 8), is exemplified by the aluminum chloride-catalyzed addition of carbon tetrachloride to trichloroethylene to give symmetrical heptachloropropane. Equation 1 shows the preparation of a dodecachloroheptene as reported by Prins. Of the two isomeric products formed, the higher melting one gave superior yield8 in reaction with hydrogen fluoride and was chosen for the present studies:
cu
160
- 220" C.+ condensate boiling above 200' C.
(4)
PRELIMINARY HYDROFLUORINATION
This preliminary introduction of fluorine by halogen exchange nas ctudied only with dddecachloroheptene. The knon n procedure 15 as used, in which a mixture of organic material, hydrogen fluoride, and antimony pentachloride was held at 150" C. and the pressure vas regulated to 400-450 pounds gage by controlled venting of the evolved hydrogen chloride ( 8 , AlCl ,CHCl-CCls 9). After one recharge of hydrogen fluoride, ( l ) the product was largely a mixture of di-, tri-, and CCI~=CCl-CCl~ +SCHCl=CCl, 3+CCle=CCl-CCI 'CHCl-CClj tetrachloro compounds boiling in the range llC(CrH&Id 220" c.: Thie product could be easily dehydrochlorinated with alcoholic C7H2Cll2 SbCl, H F --+ CiHF,oC13 (av.) ( 5 ) sodium hydroxide to give an oil of composition C7c110. The average formula shown lyas assigned from halogen analysis Similar methods used to prepare an 8-carbon compound are on the assumption of one dehydrohalogenation during the reacbelieved to proceed as follows: tion. This dehydrohalogenation would be expected from thermal inhtability 2CHCI4Cl-CHCl~ CHCl=CCl-CHCl-CHCl-CC12-CHCl2 CHC1=CC12+ AIC1, (compare heptachloropropane) as well CHCl=CCl-CH-CHC1-CCI~-CHcl*-'_t CsHaCle (2) as from the action of antimony salts; I for example, antimony pentafluoride is CHC1-CCls
+
*a
346
+
INDUSTRIAL AND ENGINEERING CHEMISTRY
March 1947
known to promote such a process (4). Dehydrohalogenation may have been single or multiple and, as will be sliox-n later, can lead to further side reactions so that a complex mixture of products may result. This mixture was treated with antimony pentafluoride without attempts to characterize completely the individual components. R E 4 C T I O S S U'ITH AXTIVOUY FEVTAFLL'ORIDE
Antimony pentafluoride n-as prepared by the controlled addition of hydrogen fluoride to antimony pentachloride in aluminum equipment. It n as distilled before use and generally employed in sealed aluminum-lined bombs under autogenous presrure.
347
pentafluoride have resulted in increased unsat uration as evidenced by tests with potassium permanganate. Apparently, antimony pentafluoride a t its boiling point may generate double bonds, the saturation of which would require the use of superatmospheric pressure. Also, addition of fluorine t o double bonds by antimony pentafluoride generates antimony trifluoride Tyhich is known to promote dehydrohalogenation (4). The over-all reactivity of antimony pentafluoride was ordinarily insufficient to give saturated perfluorocarbons under the conditions used here. Even after recycling, the products Tere generally characterized as containing small amounts of chlorine and hydrogen; the latter was shown by the liberation of hydrogen fluoride in subsequent treatment with cobalt trifluoride. In the single run tested (reaction 8), the products also Fihon-ed some unsaturation n.hen treated with potassium permanganate. CYCLIZATION
SbFs.
Starting I1:iterial
Parts by W t .
Clr a n d C?Ichlorofluoro condexate Conden-ate from c2FsC13 a
5.5" 5
Roiling Rppe,
e.
Products
m1xt. 1Iiur.
( 4 . 3 q C C1) 11ixt. ( 5 % Cl!
Reaction so.
< 145 (10 mm ) 145 (10 mm.) to 179 ( 0 8 xnm.)
IIost
(9)
(10)
As the formulas of t,he products of reaction 7 ?,how, cyclization also took place during the reaction n-ith either hydrogen fluoride or antimony pentafluoride. An analogy and possible explanation exist in the rearrangement of perchloropentadiene to perchlorocyclopentene tyhich occurs with extreme ease in the presence of aluminum chloride (7'). Equations 11 and 12 illustrate this and shoxy a parallel course possible for dodecachloroheptene, in the presence of antimony salts:
*a
cc12=cc1-cc1=ccI-cc13
C1C=CCl I I c12c CCln
Total required for t w o step-. original a n d one recycling
(11)
\ /
The formula calculated for the product of reaction 6 is based on chlorine analysis and molecular weight from vapor density (yo C1 found 12.7, calculated 12.7; molecular n-eight found 278.5, calculated 278.5). The folloh-ing constants n-ere also determined: :'n = 1.3204; d:" = 1.66F. Either an unsaturated or a bicyclic moleciile is indicated. Further treatment of this compound with antimony pentafluoride-resulted in much decomposition, as evidenced by formation of 1017 boilers. The formulas suggested for reaction 7 products are based largely on the molecular weight computed from the vapor density of the chlorine-free lowest boiling compound (found, 326; calculated for CIHF,,, 332), and the boiling point increments in the chlorine-containing part of the series. Subsequent treatment of C7HFIz n-ith cobalt trifluoride gave a saturated compound, C7FL4(boiling range, 72.5-74' C.; molecular weight found 351, calculated 350), which must be cyclic. The increase in boiling point after elimination of the final hydrogen atom, howel-er, is not clear. Other physical const'ants were determined as follon-s: Refractive index n k o Specific gra\-ity, d:' Specific refraction r
C;HFir
C;Fir
1.282 1.66 0.1064
1.2i6 1.756 0.0986
The products of reactions 8, 0, and 10 lvere not identified as to structural formula. The latter two viere designed t o give fluorochlorocarbons of chain length greater than CZu;this as apparently accomplished in reaction 10, judging from the boiling range of the products. Several types of reaction can be recognized as caused by antimony pentafluoride. Addition of fluorine to double bonds and replacement of chlorine by halogen interchange are the most obvious. Cnder certain conditions new double bonds are created, probably by dehydrohalogenation. For example, n-hen fluorochloroheptanes containing 4 or fen-er hydrogen atoms were treated with antimony pentafluoride a t atmospheric pressure and the products were distilled from the reaction mass, 1 to 2 moles of hydrogen fluoride were giren off. Also, atmospheric pressure treatments of some chlorofluoro hydrocarbons with antimony
-HC1
--+
CCI~=CCI-CCl-CHCl-CCI~ I
CHCl-CCIj rcci,=cci-c=cci-cci,i
1
I
CHCl-CC13
1
+c i c = c - c H c i c c i r '
CliC
(12)
I
CCI,
\ /
eCL
The cyclization observed with C,H?Cl,, resiet,ed all efforts t o preclude it. These included (a) elimination of the HF.SbC16 step, ( b ) attempted saturation of both the usual starting material and C,ClI, with CIn during HF. SbClj treatments, and (c) cobalt trifluoride treatment to remove unsaturation and hydrogen in C7H,Cli2 prior t o reactions involving antimony. The relatively slight amount of cyclization caused by cobalt trifluoride in its action on ri-heptane is reported in another paper (1). TREATBIENT WITH COBALT TRIFLUORIDE
Cobalt trifluoride is well suited to replace the residual hydrogen atoms and to cause saturation of the ethylenic linkages resistant to antimony pentafluoride. The liberated hydrogen fluoride and the finally negative permanganate test corroborate this. Cobalt trifluoride is not effective, hon-ever, for replacing the last few
TABLE 11. REACTIOSSWITH COBllLT TRIFLUORIDE Starting llaterial C~HFIS C7HF nC1 C7HF1"212 Rlonochloro f r o m CsHaCls CISa n d C21 compounds after SbFs reaction
Temp., O
c.
400-475 400 400 300-320
200
Product ClFir CiFiiCl C7FnC12
{chF;aC11
....
Boiling Range,
c.
72 5-74 100-102 128-1 30 100-110 115-132 S o change
To
c1
Reaation No.
348
INDUSTRIAL AND ENGINEERING CHEMISTRY
chlorine atoms, as illustrated by reactions 13 to 17 (Table 11). Reactions 13 through 16 n-err carried out in thc vitpov pl~aic.; 17 was run in the liquid phase under rvflux. Ilrl:iti\~rly lwgc increases in density after cobalt fliioridr treatmc~iit n-ere found with some antimony fluoride producti. Thrl?e are bclicvxl to be due mostly t o addition to double bonds by tlie reagmt. Evidence as to the structure of the products of rc:iction+ 13, 11, and 15 is not a t hand, but the niechanism of cyclization postulated above suggests that these are derivatives of cth?-lcyclopcnt:ine. LITERATURE CITED
(1) Benner, R. G., e t . al., IXD. EXG.CHmr., 39, 329 (1947) (2) Benning, A. F., U. S. Patent 2,230,925 (Feb. 4, 1941).
Vol. 39, No. 3
(3) Dsudt, H. I”., nnd Touker, 51. .I., I M . , 2,005,710
(.June
18,
1935). (4) Henne, -4.L., and Midgley, T., Jr., J . Am. Chem. Soc., 58, 884 (1936). (5) Hcnne, A. L., and Nemman, M. S., Ibid., 60, 1697 (1938). (6) Prins, H. J., dissertation, Delft, 1912; J . prakt. Chem., 81, 414 (1914). ( i )Prins, H. J . , Rec. traa. chim., 51, lOG5 (1932). ( 8 ) Ihid., 57, 659 (1933). before the Syuiponiurn o n Fluorine Chenli.try a i pnper 7 5 . Division of Industrial a n d Engineerinp C‘hemistry, 110th LIeeting of the . ~ \ I E R I C . ~ PC H E \ l I C . 4 L S o C I E . r Y , Chicago, Ill. T h e work described in this paper is covered a l ~ oi n n c o n i p r r h e n ~ i r ereport of work with fliiorine a n d fluorinated compounds undertaken in connection with the l l n n h n r t n n Project. This report is s m n t o be published as 1-olume I of Division \-I1 of the M a n h a t t a n Projert Terhnical Series. PREsEUrEo
(PREPARATION OF CHLOROFLUOROCARBONS)
RUuorination of poUychUoroterphen yUs F.B. Stilrnar, W. S. Struve, and W.V. Wirth E. I. DU PONT D E VEMOURS & COMPANY, INC., WIL1lINGTON, DEL.
CHLOROFLUORO oils can be made from heavily chlorinated aromatics with either AgF2 or ShFj. Economy in the elemental fluorine requirement is demonstrated orer the hydrocarbon-AgFg process. Although the resulting chlorofluoro oils are stable compounds, they do not haye the thermal and chemical stability of the perfluoro oils. The fluorination of highly chlorinated aromatics with AgFz or SbFs appears to he general. Preliminary indications are that certain nitrogen and ovygen linkages survive fluorination.
C
HLOROFLCORO compounds have been preparcd that are
comparable in stability with the perfluorinated oils discussed in the third and fourth papers of this series (pagcs 350 and 352). -4 number of heavily chlorinatrd polynnclcsar compounds yield chlorofluoro oils, either directly with .\gFn or ivith 8hFj followed by a sniaI1 amount of AgFz. Such products arc. most readily prepared from either chlorinatcd 0- or m-terphclnyl. The advantage of preparing a chlorofluorocarbon rathw than a fluorocarbon oil lies entirely in the minimum use of elcmcntal fluorine. Equations 1 and 2 illustrate tlic comparative amounts of .4gF2 required in each of the conversions:
c1
+27.igF,+
c1
>CH*]
-
+ 72.W‘~
I
I
+
+
ClsCI1, 18SbF6--+ (product containing 22’% Cl) (mol. vt., 713) (product from Equation 3)
(3)
+ 14AgF2 --+ ClsCi~F~~(15.9% CI) (mol. wt., 800) (4)
The over-all yield is 60% of theory, according to this i,cprcacntation. CHLORINATION
The chlorinated starting materials containing G7-G!)C; C1, were obtained in quantitative yield by passing clilorine into 0- or m-terphenyl a t 200-300” C. in the presence of lC; FeClr (theory, 60.7q chlorine for Cl&l14). The clilorinntetl material TTRS fluorinated directly. Further Cl8C1,F2,+ 27AgF + O(C1) (1) chlormntion n l t h ShC1, arid crys(mol. n t , tallizntion from dichlorobcnwne prior 906.5) to fluorination shoired no difference in yield of chlorofluoro 011.
c1 c1
18[ >,CF*]
fluoro oil production over fluoro oil output amounts to 70% per unit of AgF, instead of 85% as implied by the above equations and yield data. A furtl1r.r saving in active fluorine results when a halogenated tcrphenyl is first rcacted with SbFa (from SbClb HF). hddition as well as substitution occurs, and t,he resulting chlorofluoro oil can tic completely halogenated with only two parts of AgFz. This saves more than 70y0of the AgFg needed for the conversion of the chlorinated terphenyl directly to chlorofluoro oil. The steps may be represented as:
LIQUID-PI14SE FLUORIAATIOY WITH AgFs
- 72.%F + 36HF
(2)
(mol. wt., 900) I n Equation 1 fluorine has not only added to the double bonds, but has replaced two thirds of the chlorine to give a complicated mixture of products containing 4 to 6 chlorine atoms. The yield by the first process is 45‘3 as compared v i t h 65% for the second “hydrocarbon” route based on organic starting material. However, it is advisable t o maintain a larger excess of AgF2 in the first process. Therefore the net weight increase in chloro-
Clilor.inated tcrphenyls were fluorinated with .IgI>z on a semiworks scale in a n oil-jacketed, 1 x 2 foot, horizontitl qtccl rcacbtor cquippcd iyith scraping agitation. This equipment \vas similar to that describcd on page 352 by Struve et al., x i t h the cxccyition that a screw feeder was employed to charge the solid chlorinatcd terphenyl. An inert fluorocarbon solvent was used to modcrate the reaction, as described by Struve. Fifteen pounds of chlorinated terphenyl (0.029 pound mole) wcre added at 150-160° C. over 5 hours to a slurry of 13 pounds of fluorocarbon solvent (boiling range 180-137” C. a t 10 mn.) and 125 pounds of AgFl (0.856 pound mole or 40.8 moles based on chloroterphenyl).
