(20) Perry, P. H., “Chemical Engineering Handbook,” p. 238, McGraw-Hill, 1950. (21) Reid, R. C., Sherwood, T. K., “Properties of Gases and Liquids,” McGraw-Hill, New York, 1958. (22) Rihani, D. N., Doraiswamy, L. K., IND.ENG.CHEM.FUNDAMENTALS, 4, 17 (1965). (23) Rossini, F. D., et al., J . Res. Natl. Bur. Std. 42, 225 (1949). (24) Rossini, F. D., et al., “Selected Values of Physical and Thermodynamic Properties of Hydrocarbons and Related Compounds,” American Petroleum Institute Research Project 44, 1953. (25) Rossini, F. D., et al., “Selected Values of Properties of Hydrocarbons,” Natl. Bur. Standards Circ. C464 (1947).
(26 Skinner, H. A,, Snelson, A,, Trans. Faraday Soc. 56, 1776 019601. (27) Sohders, M., Mathews, C. S., Hurd, C. O., 2nd. Eng. Chem. 41, 1037, 1048 (1949). (28) van Krevelen, D. W., Chermin, H. A. G., Chem. Eng. Sci. 1, 66 (1951). (29) Ve;ma, ‘K. K., “Catalytic Vapour Phase Condensation of Methanol and Aniline to Dimethylaniline,” M. Sc. (Tech.) thesis, Bombay University, 1964. (30) Weltner, W., Jr., J . Am. Chem. SOC.77, 3941 (1955). RECEIVED for review August 7, 1964 ACCEPTED February 15, 1965 National Chemical Laboratory Communication 714.
KINETICS OF THE PYROLYSIS OF C H LO ROD I FLU0 RO METHAN E JOHN W. EDWARDS AND PERCY A. SMALL Plastics Division, Imperial Chemical Industries, Ltd., Welwyn Garden City, Herts, England
The thermal decomposition of chlorodifluoromethane has been studied at 1 -atm. pressure over the temperature range 533’ to 750’ C. The reaction i s essentially homogeneous and gives hydrogen chloride and tetrafluoroethylene as main products. Reaction mechanisms are proposed and the Arrhenius parameters for the main reaction steps obtained using a computer to integrate the kinetic equations and to make a leastsquares fit to the experimental results are given; an empirical correction term was used to take account of side reactions. Values for the heats of formation in the gas phase at 25’ C. were derived for chlorodifluoromethane as - 1 12.3 kcal. per mole and for difluorocarbene as - 39.1 kcal. per mole.
and coworkers (9) first described the pyrolysis of chlorodifluoromethane and the yields of the various products in the temperature range 650’ to 800’ C. and the pressure range 0.5 to 4.0 atm. The principal reaction, which is endothermic, is represented by the over-all equation 2 CFzHCl + CZF4 2 HCl; tetrafluoroethylene is made commercially by this route. Norton (8) found that chlorodifluoromethane decomposed on silica in the range 425’ to 525’ C. in a reaction initially of the first order with a n activation energy of 49 kcal. per mole. More recently, Gozzo and Patrick (4)have reported a n activation energy of 51.4 =t2.5 kcal. per mole for the homogeneous reaction CFzHCl + CFz f HCI, and one of 15 =t5 kcal. per mole for its inverse. All products other than C2F4 and HC1 will be called collectively “high boilers,” following Park (9). The main byproducts are cyclo-C4F8, CF2: CF.CF1, H(CF2)&1, and H(CF2)&1, though there are many others, often in trace amounts; lists of these are given by Park (9) and Serpinet PARK
+
(72).
We have studied the effects of temperature and pressure on this reaction and obtained plausible mechanisms and rate constant parameters for the more important steps, which reproduce our experimental results within analytical errors, a t I-atm. pressure. Experimental
Flow System. To study the initial stage of the reaction a t high temperature, two flow reaction vessels were constructed in platinum; one of 60-cc. volume was used in the range 500’ to 396
l&EC FUNDAMENTALS
620’ C. and the other of 4-cc. volume used in the range 620’ to 750’ C. They were shaped like pipets with the central zone (length 20 cm.) arranged in a tubular electric furnace in a position corresponding to the central constant temperature zone of the furnace (constant to f 2 ’ C. a t 700’ C . over a length of 25 cm.). Chlorodifluoromethane was passed a t a known flow rate and pressure through a filter to the reaction vessel in the furnace. The products then went through a control valve and were collected in a trap a t liquid nitrogen temperatures. The inlet a r m of the reaction vessel was fitted with a small electric preheater and a Pt-Pt, 13% R h thermocouple was fitted inside the vessel with its hot junction a t the vessel’s center. Similar thermocouples were fitted inside the furnace tube and on the outside surface of the reaction vessel a t its middle. While gas was being passed, the preheater voltage was adjusted so that the internal thermocouple indicated the same temperature (to within =t2’ C. during a run) as the external one in the furnace tube; the thermocouple on the surface of the vessel indicated the same close agreement in temperature. While steady flow and preheater conditions were being set up the pyrolysis products were discarded to a dummy trapping system. When all was steady, a tap was quickly turned and the products were collected in the main trap for a known time. Mercury manometers fitted a t the inlet and exit ends of the reactor showed a pressure drop of about 1 cm. along the reaction vessel. The contact time in seconds was calculated from the expression t = 2V1T1/(L‘2 V3)T2, where V1 is the volume of the central section of the reactor in cubic centimeters, T I is the CF2HCl inlet temperature (room temperature) in ’ K., Vz is the flow rate (cubic centimeters per second) of CFzHCl into the reactor a t T I , Va is the flow rate (cubic centimeters per second) of products out of the reactor a t T I ,and Tz is the temperature a t the center of the reaction vessel in OK. This expression thus includes a n approximate correction for the expansion arising during the reaction a t constant pressure.
+
Static System. Below about 550' C. the reaction is slow enough for a static system to be used. For this purpose a cylindrical platinum vessel with hemispherical ends and a narrow inlet tube (length 20 cm., volume 226 cc.) was used in the same furnace as thse flow reaction vessels and in the same constant temperature zone. A glass spiral pressure gage was used for pressure-time studies. The pressure-time d,ata correlated well with the analytical results of the various products formed. The static system proved useful for the study of the pyrolysis of individual high boilers. Experiments using propylene as a radical scavenger yielded inconclusive r e d t s because of the difficulty of identifying the various products. However, HCl was formed faster when propylene was present than without it. We interpret this as due to removal of CFz by reaction with the propylene and a consequent decrease in the rate of re-formation of CF2HC1 by Reaction l a . Analysis of Producits. Mass balance calculations showed that in any one flow system run, the loss of material as carbon or low molecular weight polymer was negligible and products were entirely gaseous a t room temperature.
use platinum reaction vessels in this work, it would have been inconvenient to investigate the occurrence of wall reactions by using packed vessels. There is, nevertheless, fairly good indirect evidence that the main reactions (Reactions 1, l a , 2, and 2a) are homogeneous, in that in the temperature range where it was possible to use both the flow reaction vessels, the same relation between contact time and extent of reaction was found, despite the difference in their surface-to-volume ratios, which were 2.0 and 7.0 cm.+ Further, the kinetic constants derived in this work have been used to predict the course of the reaction carried out on a much larger scale in reaction vessels made of other materials (Inconel and alumina), satisfactory agreement with experiment being obtained. We suspect that certain minor by-products such as CHF:CFz, CH2F2, and CF2C12, which occur only in amounts of the order of O.Ol%, but the concentrations of which are rather irreproducible, may be formed in surface reactions, but the main reaction and the reactions leading to the major high boilers appear to be substantially homogeneous under our conditions. Reaction Mechanisms
Hydrogen chloride :in the products was determined by a titration method in which a sealed tube containing a known amount of frozen products was broken under the surface of standard borax solution, and the excess borax back-titrated. For fluorocarbon analysis 2 cc. (STP) of products were measured into a dosimeter and passed using helium carrier gas through a tube containing moist Deacidite ion exchange resin (to remove HC1) and onto a jacketed chromatography column operated a t room temperature. A drying tube packed with anhydrous calcium chloride on Chromosorb was placed between the Deacidite tube and the column to keep traces of moisture out of the latter. The column (18 feet long) contained Chromosorb coated with perfluorotributylamine. The products eluted in the sequence C2F4, CFzHCl, CF2:CF.CF3, H(CF,),Cl, cyclo-C4F'g, H(CFp)oCl. Product elution was detected using a 200-ma. katharometer and a Sunvic potentiometric recorder (0.5- to 10-mv. full scale deflection). Frequent calibrations using the peak area method were made with each of the main coinponents. Trace products were also detected, identified, and measured. Satisfactory carbon and chlorine mass balances were obtained throughout the analyses. Materials. Chlorodifluoromethane was obtained in cylinders from Imperial Chemical Industries (General Chemicals Division) and chromatographic analysis showed a purity of 39.870. Materials such as CIF,, C F ? :C F . CF3, H.. (CF?)&I, and H(CF2),C1 were o'btained in >39% purity during a Podbielniak distillation of products from the IC1 chlorodifluoromethane pyrolysis plant, and used to calibrate the G L C apparatus.
General Features of Pyrolysis. Preliminary work in the static reaction system showed that C2F4 and HC1 are the first (stable) products, their concentrations being initially proportional to the reaction time; the concentrations of the high boiler components [including H(CF2) 2C1] were initially proportional to higher powers of the reaction time. These relations were consistent with the formation of the high boilers by Reactions 3 to 5, if it is assumed that [CFZ]increases very rapidly to a value which then remains approximately constant (the detailed kinetic treatment discussed below gave results checking this assumption). At 500' to 550' C. the initial rates of pressure rise and of formation of C2F4 and HC1 were approximately proportional to the initial pressure, indicating the initial reaction to be of the first order. The pressure rose to a maximum and then decreased as reactions resulting in high-boiler formation predominated. Much more C2F4 than HC1 was consumed in this process. T h e maximum yield of CzF4 under such conditions was low. Homogeneity of Reaction. Because it was necessary to
Main Reaction. believed to be
The first breakdown step is now generally
CFzHC1-
CFz
+ HC1
(1)
-
as has been stated by Hudlicky ( 5 ) and by Errede and Peterson (2.). According to Simons and Yarwood (73) flash photolysis of CF2HBr yields C F Z and HBr. A step such as CFzHCl CFzH' C1' seems unlikely as the main reaction, as CF2H.CFzH, which would be formed by dimerization, is not found in the products; further, addition of Cl2 does not appreciably affect the rate, though it does produce CF2ClZ. Reaction 1 is reversible :
+
CF2
+ HCl
+ CFzHCl
(la)
Mahler ( 6 ) and Gozzo and Patrick (4) have mentioned this reaction. We found that when a n equimolar mixture of CFzHCl and HCl was heated a t 510' C . for 500 seconds, only 6y0of the CF2HC1 had reacted, whereas in pure CFIHCl a t the same partial pressure 53% decomposed in this time. When CFlHCl and HBr were heated together, CFzHBr was the main product. The analogous decomposition of chloroform is inhibited by HCl (77). These facts strongly support Reaction 1a. We assume the next reaction step to be the dimerization of CF2 to C Z F ~and , we believe this to be reversible: 2 CF2 + CzF4
(2)
C2F4 + 2 CFz
(24
Mahler ( 6 ) observed dimerization of CF2 produced from (CF3)3.PFz a t 120' C.; and C F Zis formed from Cfi4 by flash photolysis (73). A possible alternative reaction leading to CzF4 is:
presumably with excited H(CFz)zCl* as intermediate. Unless AH', for CF2 is less than -61 kcal. per mole, which seems unlikely, Reaction 2 ' would be exothermic. I t might be expected that some H(CF2)zCl would be formed by stabilization of H(CFz)ZCl*; but there is good kinetic evidence that the H(CFJnC1 found is not formed in this way but by addition of HCl to CzF4. If the CFz:CF.CF3 found is formed by attack of CF2 o n the ?r-electrons of C a 4 (forming excited cyclo-C1FG, most of VOL. 4
NO. 4
NOVEMBER 1 9 6 5
397
which isomerizes, but small amounts of which are stabilized and can be identified as a trace by-product), we obtained no evidence for any insertion reaction of CFZunder our conditions except the reaction with HC1 (Reaction l a ) . Mahler (6) observed no reaction of CFz with several reagents which might have been expected to undergo insertion reactions, and the relatively low activity of dihalocarbenes is well recognized ( 3 ) . We therefore consider Reaction 2' unlikely, even though we have no conclusive evidence against it, and we postulate the formation of C2F4 by Reaction 2, which has actually been observed (6). By-product (High-Boiler) Reactions. We have carried out a sufficient number of pyrolyses of CFzHCl and individual high boilers and made enough studies of the concentrations of products as functions of contact time to be fairly confident that the main side reactions during CF2HCl pyrolysis are: (3) CzF4
+ HCl 2 CZF4
CFzHCl
H . (CF2)zCI
s C4FS
+ CzF4 $ H(CF2)&1
e C2F4 + 2H'Cl
(5) (6)
(7)
T h e quantity p(C2F,)pz(HCl)/p2 (CFZHCI) increases with contact time and attains a maximum, after which it decreases slowly as high boilers form. I t was possible to estimate by extrapolation the final value that this expression would have attained a t 600' C. if high boilers had not formed; this value, 8.7 X 10-2 atm., is assumed to be the equilibrium constant a t 600' C. Tabulated thermodynamic functions (7, 70, 75) lead to a value of 30.1 cal. mole-' deg.-' for ASo7for Reaction 7 a t 600' C., whence we estimate AHo7 as 30.5 kcal. per mole a t 600' C. and 29.06 kcal. per mole a t 25' C. If the heat of formation of CZF4 is -151.5 kcal. per mole a t 25' C., that of CFzHCl is -112.3 kcal. per mole. AHo7 enters into kinetic calculations discussed below. Since ACop for Reaction 7 is small in the temperature range 500' to 800' C., ASo7 and No7 are essentially constant in this range.
Kinetics Estimation of Arrhenius Parameters. We desired to obtain the best possible estimates of the rate constants for Reactions 1, la, 2, and 2a, expressed in Arrhenius form. T h e differential equations corresponding to our reaction scheme have no analytical solution, even if the volume change in a constant pressure system and the effects of high-boiler formation are neglected. A digital computer was therefore resorted to, and programs were used that integrated the kinetic equations and varied the Arrhenius parameters so as to minimize a n error function consisting of a weighted sum of the squares of the differences between calculated and found gas compositions a t the experimental times, the weighting being determined from the estimated uncertainties of the analytical results. The kinetic differential equations were 398
l&EC FUNDAMENTALS
-d[CzFi]/dt
=
hd[HB]/dt
(8)
where [HB] stands for the total high-boiler concentration, and X is a constant determined from the analysis of the high boilers to be 1.5; and we found that high-boiler formation could be represented by the equation :
(4)
T h e structure of the product written as H(CF2),C1 was confirmed as such by N M R studies and it is a different material from that formed by addition of HC1 to C F 2 : C F . C F 3 . This latter addition product is detectable in trace amounts after prolonged pyrolysis of chlorodifluoromethane. We have obtained some rate constant parameters for these reactions which may be the subject of a later publication. H e a t of Reaction. Under certain conditions, a n equilibrium can almost be set u p in the over-all reaction: 2 CFzHCl
written in a form that took the effect of the volume expansion (due to reaction at constant pressure in the flow system) on the reactant concentrations into account. If Reactions 3 to 6 leading to high-boiler formation had been included in the kinetic scheme in full detail, it would have been too complex to handle. Under our conditions, the total amount of high boilers formed was always small compared with the amount of tetrafluoroethylene. We therefore decided to neglect the consumption of HCl and CFzHCl by Reactions 4 and 6, respectively, and to represent the considerably greater consumption of C2F4 by introducing kinetic terms into the differential equations as below:
d[HB]/dt = kg[CzF4I2
(9)
Since [HCl] = ~ [ C Z F the ~ ] , rates of both Reactions 4 and 5 are in fact proportional to [CzF4l2. High-boiler formation was adequately represented by this equation with kg = l o B exp [-E,/RT] liter mole-' set.-', with B = 8.6 and E9 = 25.3 kcal. per mole, though in the least-squares adjustment B (but not E,) was left as a parameter. In the series of experiments finally used for determination of the Arrhenius parameters, a maximum of 25% of the C2F4 was lost as high boilers, though in most cases this loss was only a few per cent; the effect of this simplified representation on the rate constants calculated for the main reaction should be negligible. In order to reduce the number of independent parameters to be found to the minimum, available information on the thermodynamics of the reaction was used. A molecular model was assumed for CF2 (C-F, 1.28 A.;
