GEORGE F. COHOEAND W. D. WALTERS
2326
The Kinetics of the Thermal Decomposition of 3,3-Dimethyloxetane'
by George F. Cohoe2 and W. D. Walters Department of Chemistry, University of Rochester, Rochester, New York (Received February 16, 1967)
The kinetics of the pyrolysis of 3,3-dimethyloxetane have been investigated over the temperature range 400450' for pressures near 10 mm. The kinetic and analytical data indicate that the main reaction in a seasoned unpacked vessel is a homogeneous decomposition yielding isobutene and formaldehyde. Experiments with initial pressures from 7 to 74 mm at 420' have provided the major evidence for the first-order character of the reaction. The rate constant for the decomposition into isobutene and formaldehyde can be expressed as (3.8 f 0.1) X lox6exp(-60,700/RT) sec-'. From experiments with added nitric oxide or propylene, it appears that free-radical chain processes do not contribute to the decomposition to an extent greater than a few per cent, if at all.
Studies concerning the homogeneous, vapor-phase pyrolyses of cyclic compounds, particularly hydrocarbons containing small rings, have been carried out in a number of laboratories and have provided much information of interest to kineticists. Among the compounds with small heterocyclic rings, ethylene oxidea has received considerable attention, but less work has been done on the kinetics of the thermal decomposition of molecules having a four-membered ring containing an oxygen atom. Oxetane (trimethylene oxide) has been found to undergo a first-order, homogeneous decomposition at 420-460" in the presence of a free-rsdical i n h i b i t ~ r . ~Earlier Barbot5 had observed that oxetane pyrolyzes at 450" and that its 2,2-diethyl derivative decomposes to give two sets of products : (a) formaldehyde and 2-ethyl-l-butene and (b) diethyl ketone and ethylene. Later studies6 Of the pyrolysis Of substituted Oxetanes Over quartz packing and in the vapor phase have also been concerned with the kinds of products and their relative importance. Searles and his co-workersBc~d have r e ported that 2,2-diethyloxetane decomposes more slowly and a 3,3-dialkyloxetane more rapidly than the parent compound. It was of interest to investigate the kinetics and mechanism of the decomposition Of 3,3dimethyloxetane and to compare the results with those obtained for unsubstituted oxetane and other cyclic molecules.
Experimental Section Materials and Apparatus. The Journal of Physical Chemistry
3,3-Dimethyloxetane was
synthesizedl from neopentyl glycol according to the method of Schmoyer and Case.s From this synthesis a fraction IV with a boiling point of 79" (742 mm) 1.3948 [lit.* bp 79.2-80.3" (756 mm) and and nZ4as~ na5a1~ 1.39561 was purified in the present study on a gas chromatographic column (with diisodecyl phthalate as the liquid phase). This sample designated as IVa was found to be about 99.9% pure on two different columns, namely, a Golay Type R (UCON LB-550-X) in a Perkin-Elmer Model 154D chromatograph with flame ionization detector and a column with Carbowax 20M on Chromosorb P. A second sample (IVb) was (1) This work was supported by a grant from the National Science Foundation. Abstracted from the Ph.D. thesis of G. F. Cohoe, University of Rochester~ 1964. (2) Sherman 1962* (3) M. L. Neufeld and A. T. Blades, Can. J . Chem., 41, 2956 (1963); 9. W.Benson, J . Chem. Phys., 40, 105 (1964). These articles refer to the earlier work. (4) D. A. Bittker and W. D. Walters, J . Am. Chem. SOC.,7 7 , 1429 (1955). (5) A. Barbot, BUZZ. SOC.Chim. France, [5]2 , 1438 (1935); Ann. China. (Park). 1 11.519 (1939). . . . .111 . , . . (6) (a) E. Kovacs, N. I. Shuikin, M. Bartok, and I. F. Bel'skii, Bull. Acad. Sei. USSR, D ~ vC. h a . S C ~ .111 , (1962); (b) E. A. S. Cavell, R. E. Parker, and A. W. Scaplehorn, J. Chem. SOC.,Sect. C, 389 (1966); (c) 5. Searles, Jr., in "Heterocyclic Compounds with Three- and Four-membered Rings," Part 2, A. Weissberger, Ed., John Wiley and Sons, Ltd., London, 1964,p 991; (d) P. E.Throckmorton and S. Searles, Jr., 152nd National Meeting of the American Chemical Society, New York, N . Y . , Sept 11-16, 1966. (7) The synthesis and several exploratory pyrolyses were performed in this laboratory by Roger W. Nelson as part of his senior research for the B.S. degree, 1960. (8) L. F. Schmoyer and L. c. Case, Nature, 183,389 (1959).
THERMAL DECOMPOSITION OF 3,3-DIMETHYLOXETANE
obtained by the purification of IV on the Carbowax 20M column. Sample IVa was used in most of the kinetic experiments. To test whether undetected impurities in IVa might be influencing the rate, 10-mm experiments were also carried out with each of the following: (1) sample IVa after thorough contact with a doubly distilled sodium surface, (2) sample IVb, and (3) a prepyrolysed sample of IVa. The ratio of the rate of the special sample to the mean rate of sample IVa under similar conditions was (1) 0.99 f 0.03, (2) 1.025 f 0.003, and (3) 0.99. Thus the rates for the various samples showed no significant differences. The propylene and isobutene used in this study were found to be 99.9 and 99.6% pure, respectively. Nitric oxide (99%, Matheson) was distilled under vacuum two times between traps at -159 and -196". Prior to use it was degassed and passed as a gas through a spiral at -78". Experiments below 40 mm involved the use of two cylindrical reaction vessels of Pyrex glass. The unpacked vessel had a volume of 313 ml, and the vessel packed with thin-walled glass tubes had a volume of 290 ml with a 27-fold greater surface-to-volume ( S / V ) ratio. The electrically heated furnace and vacuum system were similar to those used in earlier work.g The temperature was determined with a platinumplatinum-13% rhodium thermocouple standardized at the melting point of zinc (419.5"). Pressures in the reaction vessel were measured with a wide-bore mercury manometer and a Gaertner M930-303 cathetometer. Experiments above 70 mm were carried out in a 350-ml glass vessel contained in a second furnace, and pressures were read on a capillary mercury manometer. In two experiments a vessel packed with copper wire was used. In all experiments the tubing adj, cent to the reaction vessel was heated sufficiently to prevent condensation. Before each experiment isobutene was kept in the reaction vessel for at least 1 hr (in some cases, overnight) to season the surface, and then the vessel was evacuated to mm or lower. More intensive seasoning was used for the packed vessel. In experiments with added nitric oxide the reaction vessel was flushed with a preliminary sample of nitric oxide. In most cases any added substance was introduced into the vessel before 3,3-dimethyloxetane, but when the order was reversed for propylene, no change in rate was observed. Products. I n a number of experiments with the unpacked vessel in the region 410-450", the reaction mixture was removed for analysis when the decomposition had reached about 30% (indicated by the pressure-time curve). The procedures which are described below gave evidence that the main products of
2327
the decomposition are isobutene and formaldehyde. Gas chromatography involving the use of four different columns was employed for identification (by retention time) and separation of isobutene from the entire reaction mixture or from the fraction volatile at -78", which had been treated with water or an aqueous solution of hydroxylamine hydrochloride and sodium acetate to remove formaldehyde. This material which was collected from several experiments gave an infrared spectrum on a Perkin-Elmer 421 instrument in good agreement with that of pure isobutene. Likewise in other experiments similar fractions separated from the reaction mixtures in two ways gave mass spectra which agreed satisfactorily with the fragmentation pattern for isobutene measured on the same Consolidated Model 21-620 mass spectrometer. The water-soluble portion of the reaction mixture was mixed with lithium hydroxide-lithium chloride solution. The half-wave potential (us. pool) obtained with a Fisher Elecdropode and a Sargent Model XV polarograph was the same (within 0.01 v) as that for a solution of Mallinckrodt formaldehyde and agreed with the value of Whitnack and Moshier.lO The formaldehyde in the reaction products was identified and determined quantitatively by the use of a colorimetric meth0d.l' Tests showed that under the conditions used isobutene and 3,3-dimethyloxetane did not interfere. Precautions were taken to minimize the polymerization of formaldehyde after removal of the mixture from the reaction vessel. The size of the gas fraction noncondensable at - 196" (PNc)from the ll-mm experiments averaged only 0.005 times the pressure increase (AP). In two 75-80-mm experiments at 420" with and without 1.5 mm of added NO, the values of PNCwere 0.007AP and O.O16AP, respectively. Since noncondensable gases were not significant products in the unpacked vessel, they were not analyzed quantitatively, but a mass spectrum for m/e >11 indicated the presence of CO. The third major component of the reaction mixture was found in fraction 3, not volatile at -78", and was established as undecomposed 3,3-dimethyloxetane by comparing its retention time with that of the starting material on two packed chromatographic columns (dimethylsulfolane and Carbowax 20-M) and a capillary column (Perkin-Elmer R). A combined sample of fraction 3 from several 20-mm experiments also showed a trace of isobutene. Fraction 3 had an infrared spec(9) B . C. Roquitte and W. D. Walters, J . Am. Chem. Soc., 84, 4049 (1962).
(10) G. C. Whitnack and R. W. Moshier, Ind. Eng. Chem., Anal. Ed., 16, 496 (1944). (11) C. E. Bricker and H . K . Johnson, ibid., 17, 400 (1945).
Volume 71, Number 7 June 1967
GEORGEF. COHOEAND W. D. WALTERS
2328
trum corresponding to 3,3-dimethyloxetane and gave rates of pressure increase at 410 and 450" which agreed with those of pure 3,3-dimethyloxetane. When the entire reaction mixture after about 30% reaction was analyzed gas chromatographically, the only peaks of significance were those attributable to the compounds mentioned above. In this study definite evidence was not obtained for any isomerization of 3,3-dimethyloxetane.'* If an isomerization were about 0.01 as fast as the decomposition, the isomer in the early stages of the pyrolysis would have a very small concentration compared to that of the undecomposed oxetane and might have escaped detection. However, such an amount of isomerization, which produces no pressure increase, should not have an appreciable effect on the rate constants calculated for the decomposition under the present conditions. Stoichiometry. The data indicated that the decomposition in the unpacked vessel proceeds mainly by CH3
I I
CHI
CH3--G--CH2
I
I
+CH3--CcCH2
+ CH2O
(1)
CH24 reaction 1. Table I shows the results of the analyses of the products (millimeters) in the heated vessel at reaction conditions. For the experiments in which the pressure of isobutene P I Bwaa compared with the pressure increase (corrected for dead space), the ratio P I B / A P had a Table I: Major Products from the Decomposition of 3,3-Dimethyloxetane in an Unpacked Reaction Vessel
440 440 430 430 420" 420 420 420 420 420 410 410 400'
10.0 10.6 11.0 10.3 9.5 11.2 28.5 73" 10.8 28.5 10.6 10.9 10.2
3.64 3.89 3.88 3.60 3.20 3.78 9.81 26.1 3.66 9.45 3.50 3.64
3.06
3.67 3.78 3.93 3.45 3. 14b 3.78 9.94 26.0 3.58 9.07 3.47 3.48 2.92
a Intensively seasoned reaction vessel. Determined on the same chromatogram as 3,3dimethyloxetane. The latter In the presence of 40 mm of amounted to 1.02(P0 - e). added propylene.
The Journal of Phyaical Chemktry
mean value and average deviation of 1.00 f 0.01. Moreover, several experiments with isobutenela indicated that the correction to AP for any subsequent reaction of isobutene during the first quarter of the decomposition of ll mm of 3,3-dimethyloxetane at 440" probably would not exceed the error in the pressure measurement (-0.01 mm). From the formaldehyde analyses a value of 0.96 f 0.01 was obtained for PCE,O/ AP. This slightly lower value compared to that for isobutene may result from some loss of formaldehyde (e.g., poIymerization) after removal from the reaction vessel. I n view of the analytical results, particularly those for isobutene, it was concluded that the decomposition proceeds mainly according to eq 1 and that the pressure increase can be used to determine the amount of decomposition (up to at least 30% reaction). Some analyses were also made on reaction mixtures from experiments performed in the packed bulb after only a few preliminary decompositions. The pressure of the fraction volatile at -78" after removal of formaldehyde was l.ll(AP). Gas chromatograms of the reaction mixtures had two additional peaks of moderate size, one lying between and one after the peaks of isobutene and 3,3-dimethyloxetane. Since processes other than (1) were occurring, the packed vessel was given a strong seasoning by allowing 1 atm of isobutene to remain in the vessel for 2 days at 460". After this treatment the two extraneous chromatographic peaks disappeared. After -30% of the oxetane had reacted in the packed vessel, the formaldehyde was 0.80-0.89(AP), and the pressure of noncondensable gases was 0.002(AP). It appears that even with a 27-fold larger S / V , the reaction shown in eq 1 predominates if the packed vessel is well seasoned.
Results Order and Homogeneity. The first-order character of the decomposition was indicated by the observation that the quarter-times from the pressure-time curves at 420" were not altered significantly by an ll-fold change in the initial pressure. The values for the first-order rate constants at 420" are shown in Table 11. Also consistent with first-order behavior is the fact that a plot of log [ P o / ( ~ P-o P,)] us. time was linear as far as the decomposition was carried in the kinetic study (-33% reaction). The slope of the line yielded a rate constant in agreement with that calculated from the quarter-time. For 42 experiments without added gas, the ratio t t / J t I l a averaged 2.12 f (12) Likewise in the work cited in ref 6d, the decompositions of 3,bdialkyloxetanes were observed, but not the isomerization into 2,2-dialkylpropanals. (13) Work of P. C. Rotoli in this laboratory.
THERMAL DECOMPOSITION OF 3,%DIMETHYLOXETANE
Table II: First-Order Rate Constant for the Decomposition of 3,3-Dimethyloxetane for Various Pressures at 420"" PO,
104k,
mm
sec -1
6.63 9.48 10.09 10.49 10.64 10.62
2.74 2.79 2.77 2.80 2.76 2.72
P*# mm
10.76 11.23 28.46 28.68 73. 5b
2329
Table III: Comparison of Apparent Quarter-Times for Various Reaction Vessels
10'k,
sec --I
2.76 2.76 2.72 2.71 2 . 94b
Temp, O C
430 420
" Temperatures are within f0.2" of this, and small corrections were applied when appropriate. apparatus.
Experiment in the second
0.02 in comparison with the value 2.15 for a firstorder rate law. For the purpose of obtaining kinetic data about the homogeneous decomposition, the experiments were performed in a seasoned unpacked vessel, but, to investigate the importance of surface processes in the rate studies, experiments were carried out also in packed vessels (Table 111). With only moderate seasoning of the packed vessel (S/V increased 27-fold), the rate of pressure rise was severalfold greater than that observed for the unpacked vessel. After strong seasoning, the quarter-times in the packed bulb were close to those for the unpacked vessel at a corresponding temperature. In the unpacked vessel, an intensification of the seasoning treatment above that normally used did not alter the rate significantly. From the results in Table I11 one can conclude that the decomposition in the seasoned unpacked vessel was essentially free from any surface catalysis on the walls. An experiment at a higher pressure in a vessel packed with copper wire had a quarter-time in accord with that for an unpacked vessel under similar conditions. The presence of the copper packing provides a large surface area and tends to eliminate any temperature gradient in the vessel.14 E$ect of Added Gases. Since free-radical chain processes have often been detected by the addition of inhibitors, rate experiments in the presence of propylene or nitric oxide were performed between experiments without added gas. Comparisons are given in Table IV in terms of the ratio of the rate constants in the presence and absence of the inhibitor. The ratio is quite close t o unity in most cases. No evidence for an appreciable amount of chain decomposition has been found, but in view of the experimental error an inhibition of about 1-3% cannot be established or ruled out by the present rate data. Activation Energy. The temperature dependence
400
Unpacked vessels Seasoning Strong Normal
PO, mm
9.8 7.5 7.9 10. lb 10.8 74 78 10. 3b 10.4
9.4
Packed vessels Seasoning Mod-
Strong
erate
9.3+0.1" 9.2O 2.6'
17.4b
17.4 f 0.1" 16.7" 16. Id 16.2d"
63.6b
64.1" 60.7=
Average of quarter-times (in min) for experiments under usual conditions for the unpacked vessel. Average of two experiments. Packed with glass tubes. S/V was 27-fold greater than unpacked bulb. Experiment performed in the second apparatus. a Packed with copper wire. S/V was 4.5fold greater than unpacked bulb.
Table IV: Effect of Added Gases upon the Rate of Decomposition Temp,
PO,5
OC
mm
400 410 420
430 440 450
10.0 10.6 10.6 10.2 10.3 74 74 79d 13.1 9.4 9.7 10.7 10.7 9.2
Added gas, mm
CsH, 7 . 3 NO, 0.13 NO, 0.01 CsHe, 6 . 8 C3He, 7 . 0 NO, 1 . 4 C3He, 39 CaHe, 33 CaHe, 7 . 6 CaH, 5 . 3 CsHe, 7 . 9 CaHe, 6 . 0 NO, 0.07 NO, 0.03
ka/knb
0.97 0.98 0.99 1.00 0.99 0.98 1.ooo 0.9gd 0.96 0.98 0.95 0.96 0.98 0.99
" Po denotes the initial pressure of 3,3-dimethyloxetane. k./kn represents the ratio of the rate constant with added gas to the rate constant without added gas. ' Average of two experiments. Experiments with and without added propylene (CIH6)in the vessel packed with copper wire. In other cases, unpacked vessels. of the rate was found from a series of 33 experiments at 40O45O0 in an unpacked vessel with initial pressures kept in the pressure range from 9.1 to 11.8 mm. ~
(14) The question of thermal gradients in various gaseous reactions haa been discussed by S. W. Benson, "The Foundations of Chemical Kinetics," McGraw-Hill Book Co., Inc., New York, N . Y., 1960, pp 426-431, and by D. J. Wilson, J . Phys. C h m . , 62, 653 (1958).
Volume 71,Number 7 June 1967
GEORGE F. COHOEAND W. D. WALTERS
2330
The rate constant was determined by the use of the integrated first-order rate equation and the measured quarter-time (and eighth-time) corrected for dead space. The two values for each experiment were in good agreement. The values of log k (from quartertimes) were plotted against (l/!P), and the slope of the line (visual fit) gave an activation energy of 60.8 kcal/mole. 13y a least-squares analysis of the same rate constant data (see Figure l),the activation energy With E = was found to be 60.7 =t 0.2 kcal/mole. 60.7, a value of the Arrhenius factor was calculated from each experiment. The rate expression for the decomposition in the unpacked vessel without added 0.1) X 1015 exp(-6O1700/RT) gas is k = (3.8 sec-'. From a series of 11 experiments with 9.2-13.1 mm of 3,3-dimethyloxetane at 400-450" in the presence of propylene or nitric oxide, the activation energy was evaluated as 60.7 f 0.4 kcal/mole by least-squares analysis and as 60.6 kcal/mole from the slope of a log k os. (1/T) plot. The rate constants for the experiments with 8tdded propylene or nitric oxide could be represented by the expression k = (3.7 f 0.1) X 1015 exp( -60,70O/RT) sec-l, which is almost the same as that for the experiments with 3,3-dimethyloxetane alone.
2.3
_.1.8 l0
8
4
2
+ 10
1.3
*
Discussion From the results of the analyses and rate measurements, the decomposition of 3,3-dimethyloxetane into formaldehyde and isobutene appears to occur in an unpacked vessel as a homogeneous reaction which obeys a first-order rate law. With respect to the mechanism, it is significant that the amount of formaldehyde is close to the quantity expected from eq 1, and the amount of noncondensable gas is very minor. According to previous work,15 formaldehyde in the presence of free methyl radicals or H atoms would undergo a chain decomposition forming chiefly H2 and CO, but formaldehyde alone reacts only slowly near 400". I n an earlier study4 the decomposition of 100 mm of oxetane (trimethylene oxide) without added inhibitor a t 440" occurred t 3 the extent of about 10% by freeradical chain piozesses which produced some H2, CO, and C& in addition to H2CO and C2H4. I n the absence of tnhibitor the time for 25% pressure inthe pressure and for crease depended 'Ornewhat 100 mm of oxetane was about 2094 . _ shorter than that observed when nitric oxide (1.9 mm) was added.4 For an experiment in the Present system with a lower pressure of unsubstituted oxetane (11 mm) purified by gas ChrOInatofPPhY, the time for 25% Pressure increase was about 10% Smaller than that for a similar The Journal of PhyeicaZ Chemistry
0.8 1.37
4. 1.40
1.43 10B/T.
1.46
1.49
Figure 1. Temperature dependence of the first-order rate constant for the decomposition of 3,3-dimethyloxetane: 0,single experiment; 6, two experiments; 0, three or experiments with values too close to be shown separately.
sample in the presence of 0.14 mm of NO. Thus, in the decomposition of oxetane free-radical processes apparently are not as close to being insignificant as in the case of 3,3-dimethyloxetane. To ascertain the influence of methyl substitution upon the nonchain decomposition, comparison has been made between the rate constant for 3,3-dimethyloxetane (11 mm) and the earlier result for oxetane (100 mm)la in the presence of inhibitor. The rate constant for 3,3-dimethyloxetane a t 420" is about 32/3times that of the parent molecule. It is interesting that the increase in rate constant due to the dimethylation of oxetane in the 3 position is near to that observed for the dimethylation of cyclobutane.la The Arrhenius factor for the decomposition of 3,3dimethyloxetane into isobutene and formaldehyde has a magnitude comparable to the preexponential factor measured in this laboratory for various alkyl derivatives of cyc10butane.l~ The value of the Arrhenius factor for (CH&C%H40is larger than (ekBT/h), and the main factor in producing this result is probably the positive entropy of activation associated with a ring(15) J. E. Longfield and W. D. Walters, J. Am. Chem. Soc., 77,6098 (1955).
(16) The rate constant for each compound at the pressure used is essentially independent of the initial pressure. The rate constant for 11 mm of oxetane with added NO in the uresent atmaratus - was about 14% smaller than the rate constant obtained for-100 mm in ref 4. This difference may be the result of a slight falloff with pressure or a discrepancy in the temperature. (17) M. Zupan and w . D. Walters, J. P ~ W . 67,1845 (1963).
ch.,
THERMAL DECOMPOSITION OF 3,3-DIMETWLOXETANE
cleavage process. If the over-all decomposition were a one-step reaction, an entropy of activation (-9 eu) could be estimated from the present results, but, if the decomposition involves more than one step, more information is needed. The activation energy for the oxetane decomposition with or without inhibitor was found earlier to be 60 f 1 kcal/mole, but, on account of the size of the combined experimental error for the decompositions of the two oxetanes, the amount of the difference in the activation energy resulting from 3,3-dimethyl substitution cannot be determined from the existing data. In a consideration of the details of the ring cleavage one can envision the breaking of two of the original bonds in the ring either (a) in succession or (b) simultaneously. These same alternatives have been considered for the cleavage of the cyclobutane ring, with preference being given to mechanism a in the majority of the recent publications.’* If mechanism a applies to 3,3-dimethyloxetane1 the results indicate that the intermediate biradical decomposes before it can react with formaldehyde or isobutene under the experimental conditions used. With mechanism a there arises the question of whether a C-0 or a C-C bond breaks first, and the results of an experiment in which a biradical was detected or identified experimentally
2331
would be of considerable interest. Various data have indicated that the G O distance in the oxetane ring is slightly longer than the normal value for an unstrained molecule, whereas the C-C distance is nearly normal,l9 but an exact relationship between these data and the dissociation energies of the bonds in the oxetane ring has not been established. Moreover, bond distances for 3,3-dimethyloxetane do not seem to have been reported. If a biradical intermediate exists, the over-all rate of pyrolysis will be dependent upon the rates of formation, decomposition, recyclization, and isomerization of the biradical. In such a case a detailed interpretation of the differences ,observed for related compounds will involve a consideration of the effect of substitution upon the rate of formation of the biradical and upon the rates of the various reactions by which the biradicaI may disappear.
Acknowledgment. The authors are grateful to Mr. Carl Whiteman and Mr. Roger Nelson for their assistance. They also wish to thank the General Railway Signal Co. for the use of an IBM computer. (18) 9. W. Benson and W. B. DeMore, Ann. Rev. Phys. Chem., 16, 421 (1966). (19) 9. I. Chan, J. Zinn, and W. D . Gwinn, J . Chem. Phys., 34, 1319 (1961); E. J. Goldish, J. Chem. Educ., 36, 408 (1969).
Volume 71. Number 7 Juns 1067
