KINETICS OF DISSOCIATION OF CHLORINE PENTAFLUORIDE
1183
The Kinetics of Dissociation of Chlorine Pentafluoride by J. A. Blauer, H. G. McMath, F. C. Jaye, and V. S. Engleman Air Force Rocket Propulsion Laboratory, Air Force Systems Command, United States Air Force, Edwards, California 98698 (Receiued September 96,1969)
The thermal dissociation of C1Fs was investigated over the temperature range of 520-1120°K. At temperatures above 800°K the course of the reaction occurring behind incident shock waves was followed by monitoring the uv absorption of the reacting system at 2200 A. At lower temperatures the study was conducted with static reactors which were coupled to a mass spectrometer. The unimolecular decomposition rate constants are compared to the current statisticaltheory. The theory is found to give a good description of the data taken at elevated temperatures; however, it is at large variance with data taken from the static reactor. The difference is attributed to the presence of the wall in the latter case.
Introduction The photochemical reaction of ClFa with FZto form ClF6 has been subjected to a thorough investigation by Schumacher, et uLJ1who proposed the following mechanism
+ = 2F F + ClFa + M = ClF4 + M F + C1F4 = ClFs* C1F4 + F = ClFs + ClF6* + ClF3 = 2ClF4 ClF6* + M’ = ClF6 + M’ C1F6* = ClF4 + F F + F + M = F2 + M F2
~ Z J
F2
(1) (2) (3) (4) (5) (6)
(7) (8)
Here M’ is a third body which does not include ClFa, and CIFs* is an activated form of the product molecule. The maximum temperature achieved in their study was 343°K. Under the conditions considered in the present study, step 1 will be replaced by the reverse of step 8. Similarly, due to the higher temperatures involved here, the reverse of steps 2 and 6 will have increased importance. A further complication arises in the shock tube study due to the thermal degradation of the product molecule, C1Faa2 Sullivan and Axworthy3 have made a study of the dissociation of ClFb in a flow reactor over the temperature range of 525-581°K. The course of the reaction was monitored by means of infrared spectroscopy. Empirically they found that the reaction is unimolecular, although they did not establish its pressure dependence. They reported the following expression for the rate constant
kl
= 1018.0 exp(-36,800/RT)
sec-’
(9) Our choice of an ultraviolet absorption technique for following the course of the reaction behind incident
shock waves was made possible by the very large extinction coefficient of C1F6 relative to those for F2 and ClF at 2200 8. Although CIFa also has a large extinction coefficient at this wavelength, its presence can be accounted for by a consideration of the results of our previous paper.2
Experimental Section Both experimental techniques and associated apparatus have been adequately described e l s e ~ h e r e . ~ ~ ~ * ~ The experimental design used at high temperatures was identical in every detail with that used to study the thermal dissociation of CIFs. The experimental design used at low temperatures was identical with that used to study the thermal dissociation of OF2.6 In the present instance the course of the dissociation at low temperatures was followed by monitoring evolved F2at m/e 38 andClFd+atm/e 111. The chlorine pentafluoride used in this study was provided by the Rocketdyne Division of North American-Rockwell CorporationaB Analyses by gas chromatography and mass spectrometry indicated a minimum purity of 99.5%. The argon carrier gas was Matheson preparative grade; it was used without further purification. Gaseous fluorine having a purity of 98.2% was purchased from Allied Chemical Corporation. A mass analysis revealed the presence of 0.7% Oz and 0.2% H F as the only significant impurities. After passage through a column of NaF pellets, the gas was used without further purification. Gaseous ClF having (1) R.L.Krieger, R. Gatti, and H. J. Schumacher, 2.Phys. Chem., 51, 240 (1966). (2) J. A. Blauer, H. G . McMath, and F. C. Jaye, J. Phys. Chem., 7 3 , 2683 (1969). (3) J. M. Sullivan and A. E. Axworthy, private communications, 1968. (4) J. A. Blauer and W. C. Solomon, J . Phys. Chem., 72, 2307 (1968). (6) W. C. Solomon, J. A. Blauer, and F. C. Jaye, ibid., 7 2 , 2311 (1968). (6) Rocketdyne Division of North American-Rockwell Corp., Canoga Park, Calif. Synthesized under Contract AF04(611)-10544. Volume Y4,Number 6 March 19,19YO
J. A. BLAUER, H. G. MCMATR,F. C. JAYE, AND V. S.ENGLEMAN
1184
Table I : Incident Shock Parameters and Initial Reaction Rate Constants ClFs, %
Test no.
0.5 0.5 0.5 2.0 0.2 0.9 0.5 1.8 0.5 4.5 0.17 0.17
05 07 01 18 08 10 06 14 03 17 11
12
lOe(ClFs), mol/cma
lOS(Fz), mol/cma
0.673 0.664 0.598 1.004 0.287 0.208 0.683 0.891 0.426 1.060 0.218 0.180
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0 * 000
0.000 0.000 0.257 0.212
€1 x 10-6,
T,
PI
kd
x
iob4,
atrn
OK
cma/mol
000 -1
10.0 10.0 9.1 3.9 10.2 1.8 11.1 4.1 7.8 2.2 9.4 9.5
904 915 924 940 962 991 994 1004 1116 1120 892 1093
0.60 0.67 0.71 0.55 0.75 0.75 0.71 0.59 0.71 0.59 0.67 0.66
0.09 0.12 0.23 0.22 0.38 0.18 0.75 0.59 8.2 2.3 0.16 13.3
a purity of 98% was purchased from Ozark-Mahoning Co. and was further purified by trap-to-trap distillation at - 140". Gaseous mixtures were prepared and stored in stainless steel cylinders. All mixing operations were monitored with Wallace and Tiernan precision manometers or with Heise gauges for higher pressures. A11 components of the apparatus were initially passivated with Fzand C1F5. Prior to each test conducted in the shock tube, the optical density of the static gas sample was measured a t 220 mp. The results were well described by Beer's law. No change in C1F5 concentration was noted for gaseous samples which had been stored in the cylinders for as long as 1week.
Data Analysis A . Shock Tube. A typical absorption trace is shown in Figure 1. The trace was linearly extrapolated to the shock front to obtain the optical density of the mix under conditions of negligible dissociation. The incident shock parameters were furnished by a solution of the Rankine-Hugoniot equations. The resulting absorption coefficients are tabulated in Table I along with other pertinent information. These coefficients were found to be statistically independent of the temperature in contrast to the results obtained for CIFs. Since the absorption coefficients for ClF3 were found to be of the same magnitude as those for ClFs, its presence had to be accounted for in the data analysis. Furthermore, the simultaneous degradation of these two species was an additional complicating factor which had t o be taken into consideration. The absorption coefficients for C1F and Fz were obtained in the manner described in the previous paper.2 The dissociation of CIFS probably proceeds initially by means of steps -6 and 7 of the mechanism given by Schumacher,' i.e.
+ M' = C1F6* + M' ClFs* = C1F4 + F
C1F5
The Journal of Physical Chemistry
Figure 1. Absorption trace for test no. 6; 0.5% CIFs in argon, 994"K, 11.1 atm; 0.1 V/cm, ordinate; 10 and 50 psec/cm, abscissa.
The C1Fd will then be further degraded by the other postulated reactions, i.e. ClF4
+ RI = C1F3 + F + 11 C1F4
CIF4
+F
=
(-2)
ClF5*
+ F = ClFB + Fz
(3)
(4)
The further degradation of the product species, C1F3, was assumed to follow the mechanism described in the previous paper.2 The rate equations were numerically integrated by means of a computer program' which uses assumed kinetic parameters in conjunction with the computed shock performance parameters to simulate conditions in the chemically reacting, isentropic system. The optical density, A , of the reacting mix was computed according to the form A = P(C1Fs)
+ K(ClF3) + N(C1F) + L(F2)
(-6)
(7)
(7) Furnished by Dr. T. A . Jacobs of Aerospace Corporation, El Segundo, Calif.
KINETICSOF DISSOCIATION OF CHLORINE PENTAFLUORIDE The absorption coefficients K , L, and N were taken from the previous study.2 Accordingly, the time histories of the various molecular species as given by the computer calculation allowed a calculation of the time dependence of the measured quantity A . The data analysis proceeded by matching the computed and observed absorption traces. During this process, the rate constants pertinent to the ClFs decomposition reactions were not altered and the rate constant for the assumed initial step of the reaction C1F5 = C1F4
+F
Table I1 : Rate Data Obtained from the Static Reactor
(10)
was altered until the computed profile matched the initial portion of the observed profile. During the course of these calculations, it was found necessary to assign a large value to either k-2 or k4 (ca. L2greater than lo8 sec-l) in order to obtain an analytical description of the data. This is tantamount to assuming that ClF4 is highly unstable or reactive under the conditions encountered in the shock tube. The rate constants obtained by this process for reaction 10 are tabulated in Table I. B. Static Reactor. The time history of C1Fs determined directly by monitoring m/e 111(ClFd+) and indirectly by monitoring m/e 38 (F2+)is illustrated in Figure 2 for a typical test. I n the early tests where only the production of fluorine was followed, it was found that the final fluorine concentration was less than the initial CIFs concentration. When mass peak m/e 111 was monitored it exhibited two regions of decay, a region of very rapid change (ca. 60 sec), followed by a region of much slower change which continued through the remainder of the reaction. The rate of fluorine production in the second region was found to correspond with the rate of disappearance of ClFs (Figure 2). These results were taken to indicate an absorption of C1F6on the walls of the reactor during the first region of the reaction, possibly with compound formationls which reduced the amount of C1Fs available for gas phase decomposition. Absorption of chlorine trifluoride on nickel fluoride has been studied by Farrar and Smith.g The rate constants reported here are based on observations made in the second region of the reaction. The resulting data are given in Table 11.
Results A . Shock Tube. The analytical description of the data which resulted from the computer analysis is illustrated in Figure 3 for three tests. It was found that, in most instances, once the value of klo was adjusted to allow a good description of the initial portion of the reaction, a good analytical description of the entire reaction had been obtained. This result, coupled with the results of the previous paper12lends credence to the overall mechanism assumed for the decomposition; i.e., the C1F4 formed in step 10 is rapidly decom-
1185
Test
PclFI,
P total,
no.
mm
mm
12 13 14 15 16 17 18 19 20
2 8 80 240 80 130 100 100 100
T,
2.5 10 100 300 800
960 120 120 120
kd X loss
O K
800 -1
543 543 543 543 543 543 534 525 517
3.1 3.3 3.1 2.5 2.4 2.7 1.5 0.67 0.35
Diluent
Ar AI? Ar Ar C1F
NO Ar Ar Ar
80
50
40i4 4oc: 30
-
-
,&
P 20in
it: Q Y
m a VI
3
E 10-
g E
n. 4
987 654 -
0- C I F j BASEO ON 111 PEAK HEIGHT
3 - 0 -1CIFg #::-TOTAL
2 0
I 100
BASED ON F2 CALIBRATION
IF2I CIFg OECOMPOSEO BASED ON FINAL IF2) [ p l o t t e d f o r r e f e r e n c e only] I I 1 200
300
400
I
I
I
500
600
700
T I M E [SECONDS I
Figure 2. Pressure of C1Fs us. time for a typical test in the static reactor.
posed into ClF, which is itself further degraded in the manner described previously. A close examination of the results given in Table I reveals that the rate of step 10 depends strongly upon the total pressure. An analysis of variance of these data which was based upon an assumed function of the form Iclo/(Ar)" = A e-E/RT = kll
(11)
gave the best analytical description of the data when the pressure exponent n had a value of 0.7 0.2. The resulting 1.7 order rate constant had the following form
*
(8) The nature of the absorption was not determined, but possibly ClFa W WFz*ClFs. (9) R.L. Farrar, Jr., and H. A. Smith, J.Amer. Chem. Soc., 77,4602
+ +
(1955).
Volume 74, Number 6 March 19, 1970
J. A. BLAUER, H. G. MCMATH,F. C. JAYE,AND V. S. ENGLEMAN
1186 IC11
= 1016*s*0.6 exp(-41,500 f 2500/RT)
X
cc0J/m010-~sec (12) This expression applies only to binary mixes of Ar and ClF6. The exponent of (Ar) indicates that step 10 is a unimolecular reaction which is in its intermediate pressure region under the conditions cited. This result is illustrated in Figure 4. The relatively large scatter inherent in these data may be ascribed in part to the uncertainty in the absorption coefficient for CIFacoupled to the errors incurred in the evaluation of the rate constants pertinent to its dissociation. An examination of the results of Figure 4 reveals that if the initial reaction mix contains 2% Fz, there is a statistically discernible increase in the reaction rate. Since the presence of fluorine was found to inhibit the dissociation of C1F3, the effect may be due to an exchange reaction such as ClFs
+ F = ClF4 + Fz
+ M'
= ClFs*
+ M'
+F CIF4 + F = C1F3 + Fz F + F + ?1/I = Fz + 14 CIF~*= C1F4
(- 6)
(7) (-2) (8)
The effect of reaction 5 cannot be estimated from the present data. The further degradation of ClF3 proceeds as previously discussed.2 B. Static Reactor. These data were collected over wide ranges of pressure (ca. 2.5 to 960 mm) and concentration of CIFs (ca. 0.7 X to 1 X 10-6 mol/cc). An examination of the resulting rate data, shown in Table 11, reveals that changing the total pressure by a factor of 400 does not significantly alter the magnitude of the assumed unimolecular rate constant. Accordingly, at pressures above 2 mm and temperatures below 550"K, the reaction has the characteristics of a unimolecular reaction which is above its high pressure limit. The temperature dependence of the data is graphically illustrated in Figure 5. The results for binary mixes of argon and C1FS can be expressed in the following form kd =
10le*l exp(-46,300
The Journal of Physical Chemistry
f
2300/RT) sec-'
Ill0
70U l d B O R d l O R l l l M I PSIC
ebo
100
Figure 3. Observed us. computed reaction profiles for data taken with the shock tube. The assumed rate constant is given by the expression klo = 0.16 X 1014 exp(-42,403/RT) sec-1.
(13)
A second possibility is that Fz is an efficient collision partner in step -6. Since these data are relatively uncertain due t o the various cumulative errors involved in their derivation, no attempt was made t o establish the nature of this phenomenon. The resulting error in our derived results will be small since the effect is only apparent in mixes containing large excesses of molecular fluorine. Accordingly, it appears that the initial phase of the dissociation of CIFs in the shock tube can be approximately described by the following reaction model CIFs
i
(14)
09
IO
11 TI
11
,103 ( K . I
Figure 4. Temperature dependence of 1.7-order decomposition rate constants as measured with a shock tube: 0, binary mixtures of argon and ClFj; A, mixes containing initial added fluorine.
The addition of large amounts of NO and CIF to the reacting mixture appeared to have no significant effect on the rate constant. Although these data are somewhat uncertain due to a probable wall effect, the gross kinetic behavior of the system is apparent. Comparison with Statistical Theory The statistical theory of unimolecular reactions as developed by Rice and Ramsperger,'O and Kassel" has recently been extended by Keck and KalelkarlZto the case of a large number of coupled equations which represent the kinetics of a molecule in all the possible vibrational and rotational modes of the ground electronic state. The spontaneous decay rates and vibrational transition probabilities required were derived from the statistical theoryola If the dissociation energy is known, the model will provide ah estimate of the dis(10) 0.K. Rice and H. c. Ramsperger, J.Amer. Chem. Soc., 49,1617 (1927). (11) L.S.Kassel, J. Phys. Chem., 3 2 , 225, 1065 (1928). (12) J. C.Keck and A. J. Kalelkar, J. Chem. Phys., 49, 3211 (1968). (13) J. C.Keck, Advan. Chem. Phys., 13, 85 (1966).
KINETICSOF DISSOCIATION OF CHLORINE PENTAFLUORIDE
1187 ~~
Table I11 : Comparison of Results Obtained a t Low Temperatures with Statistical Theory
OK
PT,mm
517 525 543 525 580
120 120 2.5 760 760
T,
D, kcal/mol
P
47 47 49 37 37
m = ( b In kd/d In [MI )T.
k
0 0 9
0.2 0.2 0.5 0 0
...
...
d
Calcd
mtheo4
mobsd
0.005 0.010 0.001 14 352
,
8W-l
x
-0 1
Method
Obsd
0.35 0.67 3.1 4.8 135
Static reactor Static reactor Static reactor Flow reactorb Flow reactor
Sullivan and Axworthy, ref 3.
kd
.3
--
STATISTICAL THEORY
0 -SHOCK TUBE RESULTS
5’
b
-DIRT DISSOCIATION ENERGY &- EvO/AT E y b ZERO LEVEL VIBRATIONAL ENERGY ko- PARAMETER GIVEN BY THEORY A PARAMETER GIVEN BY THEORY 01~EFFECTIVE COLLISION CROSS4ECllON 023- EFFECTIVE DECAY CROSS.SECTION D
4 -
4
-
-
r2 Y .5 m .J
-
1x103 9 -
8 -
-6
7 6 -
4
-
v-CIF5/CIF:Ar
-7
3 1.82
1.84
1.86
1.88
1.90
1.92
1.94
1.96
[ l / T ] x l O 3 “K’’
Figure 6. Correlation of experimentally observed first-order dissociation rate constants with statistical theory. (See ref 12.) D = 47 d~ 2 kcal/mol, U ~ A= uzs= 1 X 10-16 cm2.
Figure 5. Temperature dependence of decomposition rate constants for static reactor.
sociation rate constant as well as its pressure dependence. If, as in the present instance, the dissociation energy is unknown, the model employs the observed activation energy to compute a dissociation energy which it then uses to estimate the desired quantities. Although the model neglects the conservation of total angular momentum, thus likely leading to an overestimate of the reaction rate, nevertheless, the theory was used with considerable success to correlate experiment with theory for 21 molecules having from 3 to 6 atoms. Chlorine pentafluoride has been shown to have a square-pyramidal structure in its ground electronic state. l4 All required molecular parameters for this molecule were taken from the “JANAF Thermochemical Tables.”l6 In the absence of any data concerning the postulated product, CIR, it was assumed to have a squarE-planar configuration with a C1-F bond distance of 1.7 A. The required vibrational frequencies for this molecule were assumed to be identical in magnitude with those of the corresponding vibrational modes of ClFS. Although
these assumptions lend uncertainty to the resulting calculation, the errors introduced are probably less than the errors introduced by the assumptions made in the derivation of the mathematical model itself. The model was first applied to the results obtained from the shock tube. The required activation energy was calculated from eq 12. The model estimated the dissociation energy at 47 f 3 kcal/mol for an average temperature of 1000°K. The pressure dependence of the reaction was estimated to have an exponent of 0.5 f 0.2 as compared to the observed value of 0.7 0.2. The analytical description of the data is illustrated in Figure 6. Overall the model gives a good description of all aspects of these experimental results. The model was also applied to the results obtained from the static reactor. Here again the dissociation energy was estimated by the theory at 47 3 kcal/mol; however, neither the absolute magnitude of the rate
*
*
(14) F. P. Gortsens and R. H. Toeniskoetter, Inorg. Chem., 5 , 1925 (1986). (15) “JANAF Thermochemical Tables,” Dow Chemical Co., Midland, Mich., 1968.
Volume 74,Number 6 March 19, 1970
F. A. THOMASSY AND F. W. LAMPE
1188 constant nor its pressure dependence was reproduced satisfactorily. Typical results are shown in Table 111. I n every instance, the observed rate constant was at least a factor of 50 larger than the calculated rate. It would appear that this difference is due to the influence of the reactor walls although the actual mechanism is not immediately apparent. Finally, the resulta of Sullivan and A x ~ o r t h y ob,~ tained in a flow reactor, were compared with the theory. Although the theory does reproduce the magnitude of their rate constants very well, it gives a dissociation energy of 37 kcal/mol which is at considerable variance with the results obtained from the shock tube. Although the disagreement between the static reactor data and the shock tube data is probably due to the influence of the wall in the former case, the source of the disagreement between the flow reactor data and the shock tube data is not clear. I n the absence of independent thermal data for the C1F4 molecule, the statistical theory offers no clear choice between these latter two sets of experimental results. Although the above treatment assumes that C1F4 and F are initial decompositioll products (ca. eq lo>,it must be admitted that Re "le Out the following process ClFrj = ClFa
+ Fz
(15)
Indeed, a statistical treatment based upon this reaction path is in equally good agreement with the data.
Conclusions I n conclusion, the thermal dissociation of C1F6 a t elevated temperatures is well described by the statistical theory of unimolecular reaction rates. This theory points out a large disagreement between these data and the data taken a t lower temperatures (ca. 550°K) by Sullivan and Axworthy from a flow reactorag I n the absence of independent thermodynamic data for the CIFl molecule, we are unable to clearly define the nature of the decomposition of the CIFs molecule ( i e . , unimolecular or chain) although by analogy to ClFa,2both mechanisms may be operative. The static reactor data appear to favor the unimolecular decomposition path, although this may be due to the influence of the reactor walls.
Acknowledgments. The authors wish to thank Dr. T, A. Jacobs of Aerospace Corporation and Drs. J, Sullivan and A. Axworthy, both of Rocketdyne Division of North American-Rockwell Corporation, for very useful discussions concerning this and related work. The authors also wish to acknomrledge the assistance of A h . M. A. Abrego in the experimental phases of the work.
A Mass Spectrometric Study of the Dimerization of Nitrosomethanel by F. A. Thomassy and F. W. Lampe Vhitmore Laboratory, Department of Chemistry, The Pennsylvania State University, University Park, Pennsylvania (Received August 26,1969)
16808
It is shown directly that the disappearance of monomeric nitrosomethane in a second-order reaction is accompanied by the simultaneous formation of dimeric nitrosomethane. The dimerization rate constant has cm8 been measured by following the monomer disappearance and has been found to be 0.36 ==I 0.08 X molecule+ sec-1. Studies of the initial formation rate of CHsNO+ as a function of electron energy lead to an ionization potential of 10.8 f 0.3 eV for monomeric nitrosomethane.
in nitric oxide, to produce methoxy radicals and nitrous For some time it has been known that the monomeric oxide; such reactions are of importance only when the form of nitrosomethane, CH,NO, is the primary product of the reaction of methyl radicals with nitric o ~ i d e ~ - ~ and also that this initial product reacts further in a (1)Based on a thesis submitted by F. A. Thomassy to The Pennsylvania State University in partial fulfillment of the requirements for variety of ways.lO Thus Christielo and Johnston and the M.S. degree. Heicklen" have shown that monomeric nitrosomethane (2) L.A. K.Staveley and C. N. Hinshelwood, Proc. Roy. Sac., A159, reacts with nitric oxide, in a reaction second order in 192 (1937). nitric oxide, to form nitrogen and nitrate radicals, NOa, (3) J. Collin, Bull. Sac. Roy. Sci. (Liege), 23, 201 (1954). while simultaneously undergoing a reaction, first order (4) W.A. Bryce and K. U.Ingold, J. Chem. Phys., 23, 1988 (1955). The Journal of Physical Chemistry
