926
(3) (4) (5)
(6)
BENJAMIN BARNETT AND WILLIAM E. VAUGHAN
ponents; the Chloroplant Substance of Spinach Leaves,” Doctoral thesis, Purdue University, Lafayette, Indiana, 1941. COMAR,C. L., A N D ZSCHEILE, F. P.: Plant Physiol. 17, 198 (1942). DRABKIN, D . L.: J. Optical Sac. Am. 36, 163 (1945). GIBSON,K . S., A K D KEEGAZI, H . J . : J.Optical Sac. Am. 28, 372 (1938). HOGNESS, T . R., ZSCHEILE,F . P . , JR.,A N D SIDWELL, .4.E., J R . : J . Phys. Chem. 41,
379 (1937). (7) MELLON, bl. G.: Colorinietry f o r Chemists. The G . Frederick Smith Chemical Co., Columbus, Ohio (1945). (8) hhCIiAELSON, J. L.: J. Optical SOC. Am. 28, 365 (1938). (9) LIILLER.E. S.:Plant Physiol. 12, 667 (1937) (10) MITCHELL, J. H . , J R . ,AND KRAYBILL, H . R . : Ind. Eng. Chem., Anal. Ed. 13, 765 (1941). (11) MUNCH,R . 11.: J . Am. Chem. SOC.67, 1863 (1935). (12) PENICK, D. B . : Rev. Sci. Instruments 6, 115 (1935). (13) RENTSCHLER, H. C . , A N D H E N R YD, . E . : Rev. Sci. Instruments3, 91 (1932). C., , A N D STATES,I\I. N . : J . Optical SOC..4m. 31, 64 (1941). (14) S H E A R D €1. C.: J. Am. Chem. Sac. 68, 247 (1936). (15) SJIITH,JAMES (16) ZECEI~IEISTER,L., A N D POLGAR, A , : J . Am. Chem. SOC.6 6 , 1522 (1943). (17) ZSCIIEILE, F. I’., A N D COXAR,C. L.: Botan. Gaz. 102, 463 (1941). (18) ZSCHEILE, F. P., COMAR, C. L., A N D ~ I A C K I N N GE .:Y Plant , Physiol. 17, 666 (19.12). (19) ZSCHEILE,F. P., A N D ~ I A R RDI S .G , . : J.Phys. Chem. 47, 623 (1943). (20) ZWHEILE, F. P., ASD H E N R YR , . I,,: Ind. E n g . Chem., Anal. Ed. 14,422 (1912). T . F . : J. Phys. Chem. 38, 1 (1934). (21) ZSCHEILE,F . P . , JR.,HOGSESS,T. R . , . ~ S DYoIJN.~~, (22) ZSCHEILE,F. P., WHITE,J. W., J R . , B E A D L EB. , W . , A N D RO.ACH, J . R . : Plant Physiol. 17, 331 (1942).
THE DECOMPOSITIOS OF UESZOYL PEROXIDE.
I
THEKINETICS AKD STOICHIOMETRY IS BENZEXE BEKJAMIK BARNETT
ASD
WlLLIAlLI E. VAUGHAN
Shell Development Cornpanil, Emeryville, California Received .Warel& 19, I947
I.
IKTRODUCTIOS .\KD L I T E R i l T E R E
REVIEW
Although benzoyl peroxide has been extensively used for many years as a polymerization catalyst, the manner of its decomposition under different conditions is still not fully understood. At the concentrat>ionsof the peroxide normally used in catalyzed polymerizations, it seems generally accepted (1, 2, 3, 18, 21, 22) that the initial step
DECOMPOSITION OF BENZOYL PEROXIDE.
I
927
is followed by such reactions as: C6HsCO
II
+
CaH6
f
Con
(2)
0
C&
+ +
hydrogen donor -+
C&
f radical
(5)
Rz Rz 1 1
CeHs RzC=CRz --+ C8HbC-C(6) The activation energies reported (5,20, 21) are of the order of 30-31 kg.-cal. per mole. This may (24) be the energy required to rupture the oxygen-oxygen bond, and consequently equation 1 may represent the rate-determining step. Recent studies (23, 25) indicate that a similar mechanism probably holds for the decomposition of di-te>rt-alkylperoxides. On closer study, however, the kinetics are found not to be as straightforward as this scheme would indicate. It has been known for a number of years that the rate of decomposition of benzoyl peroxide is not the same in all solvents (14). The faster rate in mixtures of benzene and vinyl acetate than in benzene alone has been ascribed (18) to the formation of a complex betmeen peroxide and vinyl acetate, more reactive than the peroxide itself, more recently this effect has been interpreted (22) as the disturbance, to a greater degree by vinyl acetate than by benzene, of an equilibrium between benzoyl peroxide and benzoate radicals. The assumption of a primary association step leads to the possibility of subsequent decomposition by a non-radical path of the complex so formed, and some evidence has been presented (10, 29) in support of such a course, particularly in more conrentrated solutions. The only free-radical mechanism proposed (17) for these conditions appears to be excluded on energy grounds (26), and in fact no experimental evidence has been obtained (20) for the occurrence of the products to be expected from the radicals assumed. The stoichiometry of the decomposition of benzoyl peroxide in concentrated solutions has been studied extensively (14), and the participation of the solvent in the reaction is well established (6, 9, 19, 27, 28), although in the higher-boiling solvents, such as nitrobenzene and paraffin wax, rapid decomposition of the peroxide to biphenyl and carbon dioxide (the so-called Kolbe reaction (10)) may result in the entry of only small amounts of the solvent into the final products (9). The Kolbe reaction is the main decomposition path of pure benzoyl
928
BENJAMIN BARNETT AND WILLIAM E. VAUQHAN
peroxide, both thermally (IO, 11, 12, 30) and photochemically (13). For concentrated solutions the earlier “RH” scheme (14) summarizes the stoichiometry in convenient form (“(so1)H” represents solvent) : CeHbCOH
+ COz + (sol)CoH5
(7)
In benzene equation 7 represents the main reaction ( I I ) , in cyclohexane reactions 7 and 8 proceed to about equsl extents (16), and in isobutyl alcohol reaction 8 is the main course (15). Not only the rate, therefore, but also the stoichiometry of the decomposition is known to vary from solvent to solvent. Evidence for a deviation from the simple first-order kinetics represented by equation 1 can be found by examination of the recent literature. There is a distinct trend of the first-order decomposition rate with the concentration of the peroxide in benzene (20),in benzene containing 20 per cent by volume of styrene ( 8 ) , and in allyl acetate (2). The possibility arises that the nature of the decomposition products may also change with the initial peroxide concentration. No thorough study of either effect has heretofore been made. The problem, however, would appear to be of technical importance, for in those solvents in which the deviation from first-order kinetics is greatest, the peroxide might be decomposing to a large extent by a process which, if non-radical, would be useless for sensitizing polymerization. The present paper includes the results of a study of the variation of the observed decomposition rate with the initial concentration of benzoyl peroxide in benzene. The change in stoichiometry was followed to the extent of determining the initial rates of formation of carbon dioxide and free acid. The effect of the peroxide concentration on the velocity constant was investigated, and it was also found that the stoichiometry of the decomposition changes steadily as the initial concentration of peroxide is increased. The limiting reaction a t low peroxide concentrations is kinetically of the first order, the ratedetermining step probably being that represented by equation l, and the main over-all change is apparently the formation of carbon dioxide and biphenyl. At high concentrations of peroxide, the reaction may be represented by a secondorder equation; the mechanism is apparently of a non-radical nature, and the stoichiometry approaches that of reaction 7, the main course of the “RH” scheme in benzene. To amplify the meager data in the literature on the effect of different solvents on the decomposition rate, a study was also made with a large variety of solvents, in which differences in rate of over twentyfold were obtained a t the same peroxide concentration. The results of this study are presented in the following paper (see page 942).
DECOMPOSITIOiX OF BENZOYL PEROXIDE.
I
929
11. EXPERIMENTAL METHODS
The decompositions were carried out in a n all-glass apparatus. The reactors, varying in capacity from 100 cc. to 1000 cc., were provided with three openings and an oil-filled thermometer well. X mercury-sealed glass stirrer passed through the middle neck, and a t one of the side openings a standard-type, seven-bulb condenser was attached. A capillary sampling tube passed don n through the third neck and bore a t the top a turned-down enlargement of about 5-cc. capacity closed at each end by a stopcock. This tube permitted removal of samples at any time by application of suction. The system could be flushed with nitrogen or oxygen, as required, by passing the gas in through the sampling tube. From the top of the condenser the evolved gases could be passed to a water-jacketed gas buret for volumetric measurement, to a standard carbon dioxide-absorption train in which the gas was absorbed in Ascarite, or to a reservoir from which samples of the gas could be removed for analysis. When the thermostated oil bath in which the empty reactor was immersed had attained the proper temperature, the prepared reaction mixture was poured in through the middle opening, and the stirrer then set back in place and started. Tap water, usually a t 14-18"C., was used to cool the condenser. When the reaction mixture had reached the desired temperature, as indicated by the calibrated thermometer (accurate to &0.02"C.),a sample of the reaction mixture was removed by means of the sampling tube and at once, a t "zero time," a stopcock in the line was opened to permit gas to collect in the buret or to be absorbed in the Ascarite. When the gravimetric method \vas used, a sloiv stream of nitrogen was passed through the system throughout the experiment. I n the volumetric determinations of carbon dioxide, a correction was applied for the partial vapor pressure of benzene a t the temperature of the tap nater used to cool the condenser. The experiments in which evolved gas was determined gravimetrically were run for a maximum of 5 hr., corresponding to decompositions of ca. 55 per cent. Thereafter, to follow the decomposition, samples of the reaction mixture were removed from time to time for analysis, and the amount of gas evolved either read off the buret or determined by weight. Since, when the initial concentration was lorn, the presence of air or oxygen in the system resulted in slightly too high values for the observed ratio of the number of moles of carbon dioxide formed per mole of peroxide decomposed, the system was always thoroughly flushed with nitrogen a t the start. Total acidity was determined by titration with dilute standard sodium hydroxide. Analyses for oxidizing power were carried out in the following manner: The weighed mmple (3-4 g , ) was mashed into a 500-cc. Erlenmeyer flask with 10 cc. of glacial acetic acid, and after the air in the flask had been removed by sublimation of a few small pieces of dry ice, about a gram of sodium or potassium iodide was added and the mixture refluxed on the steam bath for 15 min. After dilution with about 100 cc. of water, the liberated iodine was titrated with standard thiosulfate solution. The benzoyl peroxide used was an Eastman White Label product, different
930
BENJAMIN BARNETT AND WILLIAM E. VAUGHAN
samples of which analyzed from 98.8 to 99.4 per cent peroxide. For a few experiments the White Label material was further purified by dissolving in chloroform and reprecipitating with methyl alcohol; however, this material, which assayed 99.G per cent, gave no sensibly different results. 111. THE DECOXPOSITION OF BENZOYL PEROXIDE IN BENZENEAT 80°C. A. KIKETICS
The deviation from first-order kinetics mas readily observed in any one experiment when the logarithm of the peroxide concentration, P , was plotted against I
I
I
I
I
I
2
3
4
Time, hr. FIG.1. Dccomposition of benzoyl peroxide in benzene. Increasing deviation from firstorder kinetics with time. T = 80°C.
the time, as in figure 1. Data for a typical run (curve (a) in figure 1) are given in table 1; from this a good idea of the precision of the measurements may be obtained. It is evident that a first-order expression holds only at the start. Since concentrations arc expressed as moles per kilogram of solution and since carbon dioxide is evolved, kl would normally decrease. Approximately onetenth of the decrease noted is due to this factor. However, it is difficult to corrcct for this effect because the stoichiometry changes as the reaction proceeds (see Section B). The pure first-order reaction was isolated by plotting the initial slopes of graphs such as that in figure 1 against the initial concentrations. The initial slopes in figure 1 are:
DECOMPOSITION O F BESZOYL PEROXIDE,
93 1
I
and if lil (obs.) is a true first-order constant, it obviously should be independent of the concentration of peroxide. The data at 80°C. in table 2, which are plotted in figure 2, show hoxever that this is not the case: even from the start the limiting first-order reaction is accompanied by a formally second-order process.’ For all values of P studied, therefore:
+
= kl (obs.) = k1 kzP dt In figure 2, since kl is the intercept at P=O, and k? the slope,
-__ -d(loge )‘
(10)
kl = 0.158 hr:’
kz = 0.145 kg./moles-hr. = 0.164 liters/moles-hr.
at low peroxide concentrations. The fraction (see table 2) of the over-all decomposition that.is due to the second-order reaction is:
and consequently this reaction becomes the major course a t concentrations for xhichf > 0.5: namely, for conditions where Since this paper was written, the authors have conversed with Dr. Kenzie Nozaki,. who worked on this same problem with Professor Bartlett a t Harvard University prior to joining the staff of the Shell Ilevelopment Company. Nozaki and Bartlett, a n account of whose work is presently appearing (4, 21), have also found a departure from first-order kinetics t h a t is the more observable a t higher peroxide concentrations. They feel that under certain conditions the rate of the reaction accompanying the basic first-order change is better expressed by a 3/2-order equation, and ascribe (21) the departure from first-order kinetics to a radical-sensitized decornpoaition of benzoyl peroxide, thus:
CaHj
0
0
I/
II
+
CoHSCOOCCeHj --+
+
;I; C6HjCOOCCeHj
or
0
11
CGH~CO
non-radical products
‘y ,/
L
-----+
non-radical products
+
CGII~ 0 It
+
CaHiCO
Such a scheme conforms to the 3/2-order rate observed. I t is interesting that this interpretation, also, leads to the possibility of “peroxide-wastage”: namely, the dissipation of the “catalyst” without forming free radicals which might propagate chains. Under certain conditions, however, these authors concede the possibility of a n accompanying second-order reaction. I n support of their conclusion that the side reaction is of the 3/2 order, Nozaki a n d B a r t lett found t h a t certain radical-producing substances, such as hexaphenylethane, increase the decomposition rate of the peroxide. On the other hand, i t had previously becn found in the present work, in a single experiment, that the oxidizing power of a benzcne solution containing 0.2 mole/lig. of benzoyl peroxide and 0.2 molc/kg. of hexaphetiylethanc decreased a t the same rate ns t h a t of n solution containing 0.2 molc/kg. of the peroxide alone in l ~ n zene.
932
BENJAMIN BARNETT AND WILLIAM E. VAUOHAN
k '2CI > 0.5 or P> A ( P > 1.09) (12) k r P kz This corresponds to concentrations greater than 10.4 mole per cent. It should be noted that the concentrations of benzoyl peroxide used in technical applications are much lower than this (usually of the order of 1-2 per cent), and the participation of the higher-order reaction a t these concentrations will be correspondingly less. The abrupt drop in observed rate near the ordinate axis suggested in figure 2 is not to be interpreted as a deviation from the proposed kinetics, but appears to TABLE 1 Decomposition of benzoyl perozide i n benzene P O (starting concentration): 0.5861 mole/kg. T = 80°C. PEALTION T l Y E
oNCENTPAT1oN PEROXIDE, P
hmrrr
mJIcrIkK
0 0.33 0.50 0.67 0.83 1.00 1.17 1.33 1.50 1.67 1.83 2.00 2.17 2.33 2.50 2.67 2.83 3.00 3.17 3.33
0.5861 0.5425 0.5135 0.4954 0.4791 0.4615 0.444' 0.4184 0.4135 0.3941 0.3719 0.3551 0.3541 0.3378 0,3264 0,3063 0.3000 0.2891 0.2806 0.2699
1'
ki
INCBNTPATION C PEPOXIDX, P
(OW.)
hr.-l
0.233 0.263 0.251 0.243 0.239 0.237 0.253 0.232 0.262 0.249 0.250 0.232 0.236 0.234 0.243 0.237 0.236 0.232 0.233
kI
(OBS.)
hours
molcs/kg.
hr.?
3.50 3.69 3.83 4 .OO 4.17 4.33 4.50 4.67 4.83 5.00 5.17 5.33 5.50 5.67 5.83 6.00 6.17 6.33 6.67 6.83
0.2616 0.2512
0.230 0.230 0.227 0.232 0.232 0.228 0.228 0.222 0.227 0.229 0.228 0.223 0.227 0.227 0.223 0.221 0.223 0.222 0.218 0.220
0.2460 0.2311 0.2229 0.2176 0,2098 0.2076 0.1960 0.1867 0.1807 0,1768 0.1677 0.1600 0.1593 0,1552 0.1483 0.1434 0.1365 0.1303
be due to the formation of traces of peroxide of the solvent, the effect of which on the rate would be observed only when the total oxidizing power is low from the start. This effect is not very great in benzene even in the presence of oxygen, but in other s o l v e n t s s u c h as alcohols, acids, cycloparaffins, and notably, substituted benzenes-the total oxidizing power may increase several fold in the presence of oxygen over that of the initial benzoyl peroxide content. Data illustrating this effect are presented in the following paper. From the data in table 2, the activation energy of the first-order reaction has been calculated to be 32 kg.-cal./mole (see figure 6), in essential agreement with the literature (18, 20). This point is further discussed in the following paper. The activation energy of the higher-order reaction can be roughly estimated from the data in table 2. They lead to a value of approximately 28 kg.-cal./
TABLE 2 Decomposition of benzoyl peroxide i n benzene Initial rates as function of initial peroxide concentration, Pa TEUPEPATUPE
krPo ‘ = k l O
ki
(*O.lO)
75
80
hr.7
kt./molcs-hr.
0.0420
0.0450
0.0944
0.0724
hr,-1
%.
55 60 70
0.07112 0,06699 0.0983 0.1745 0.3304 0.4804 0.6180 0.6955 0.8222 0.8650 0.9454 1.064 0.1379 0,1432 0.2526 0.2642 0.4749 0.5047 0.6336 0.8222 0.8933 1.035 0.01095 0.02138 0.03837 0.07980 0.1633 0.2759 0.3162 0.5861
0.00411 f O.ooOo7 0.00996 f 0.00031 0.0467 f 0.0018 0.0504 f 0.0026 0,0573 f 0.0031 0.0622 f 0.0027 0.0696 f 0.0018 0.0722 f 0.0042 0.0831 f 0.0020 0.0808 f 0.0021 0.0875 f 0.0047 0.0883 f 0.0013 0.102 f 0.005 0.105 f 0.007 0.115 f 0.007 0.114 f 0.008 0.128 f 0.006 0.132 f 0.007 0.135 f 0.004 0.155 f 0.004 0.162 f 0.005 0.168 f 0.004 0.150 f 0.006 0.158 & 0.008 0.163 f 0.006 0.171 f 0.005 0.181 =t0.004 0.198 f 0.005 0.205 rt 0.006 0.243 & 0.008
I
P,.
0.158
0.145
0.095 0.156 0.260 0.348 0.400 0.433 0.445 0.482 0.486 0.512 0.098
0.099 0.159 0.168 0.269 0.277 0.340 0.385 0.399 0.446 0,011 0.020 0.034 0.067 0.131 0.202 0.224 0.350
I
0.2 0.4 molrs/kg. solution
FIQ.2. Decomposition of benzoyl peroxide in benzene. Dependence of observed firstorder rate on initial concentration. T = 80°C. 933
934
BEh'JAMIK BARNETT AND WILLIAM E. VAUGHAN
mole. This figure is not accurate, partly because of the small temperature range involved and partly because of the unknown influence of temperature on the solvent effectof the peroxide. Nozaki and Bartlett (21) found an activation energy of 25 kg.-cal./mole for their higher-order process. The only data in the literature which show the presence of higher-ordcr reactions have been obtained a t lower temperatures (see table 3). An analysis of the data of McClure, Robertson, and Cuthbertson (20) for 78°C. gives k? = 0.18 and for 66'C. k, = O.OG for very low peroxide concentrations. Thc specific rate to be expected at (34°C. from the higher-temperature data of these authors is TABLE 3 Decomposition of benzoyl peroxide Evidence from the literature for a simultaneous higher-order reaction EKPEBATUP.
SOLVENT
(f0.1") ~~
PO
~
hr.?
"C.
Benzene*. . . . . . . . . . . . . . . . . . . . . . . . . .
66
78
0.0772 0.126 0.188 0,0770 0.124 0.185
0.0237 0,0278 0.0316 0.130 0.136 0.149
Benzene containing 3.46 moles styrene per liter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
64
0.0508 0.0811 0.216
0.0205 0.0211 0,0257
Allyl acetate . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
80
0.0413 0.0887 0.248 0.413
0.191 0.196 0.227 0,235 I
*Brown (7) gives results a t 80°C. in benzene, which show a similar trend, but his firstorder constants are uniformly lower than those given by other investigators a t the same concentrations. It would seem t h a t in these experiments either the peroxide had not completely dissolved at the instant taken as zero time, or t h a t the temperature was actually somewhat less than 80°C. Brown concluded t h a t the first-order decomposition is accompanied by a second-order reaction.
kl = 0.15 and k2 = 0.06 a t low concentrations of peroxide. Cohen's data (8) extrapolate to kl = 0.18 and k2 is found to be 0.03 at this temperature. The styrene used in Cohen's experiments may exert a definite solvent effect on k2 different from that of benzene, but from these data the effect on kl would appear to be small. This is in accord with results obtained in this laboratory and t o be reported later. If the styrene were displacing an equilibrium between peroxide and benzoate radicals, an appreciably higher limiting rate might be expected under these conditions than in pure benzene (22). Table 3 also includes data on the decomposition of benzoyl peroxide in allyl
DECOMPOSITION O F BENZOYL PEROXIDE.
935
I
acetate. In this solvent also a higher (apparently second) order process occurs as a side reaction, the rate of which is not greatly different from that of the second-order decomposition in benzene at the same temperature. Other data obtained in this Laboratory indicate that higher-order decomposition reactions accompany the normal first-order decomposition also with other peroxides (23). TABLE 4 Decomposition of benzoyl peroxide i n benzene DeDendence o n the initial Deroxide concentration of the amounts of carbon dioxide and free acid formed per mole of peroxide decomposed PO
/II
TEMPERA-
(%&
PO
n
B
-"C.
55 60 70
75
80
0.07112
0.06699 0.09827 0.1745 0.3319 0.4798 0.6194 0.6950 0.8241 0.8590 0.9485 1.059 0.1371 0.1442 0.2650 0.4749 0.5059 0.6353 0.8308 0.8955 1.035 0.00928 0.01016 (a)* 0.01144 0,01475 0.01856 0.02138 0.03736
0.72 0.86
1.14 0.81 0.34 0.35 0.43 0.43 0.53 0.52 0.57 0.53 0.59 0.65 0.27 0.29 0.35 0.46 0.47 0.50 0.58 0.58 0.63
1 ~
1 ~
0.24 0.47 1.87 1.90 1.89 1.68
0.36
o.21
I.
'C.
80
0.03837 O.lM333 0.05668 0,05788 (a) 0.07508 0,07700 0.07980 0.09137 (b) 0.09457 (a) 0.09708 0.1051 0.1079 0.1196 0.1344 0.1473 0.1486 (a) 0.1633 0.1738 0.2022 0.2759 0.3162 0.3445 0.5861 0.6397 0.6530 0,7452 1.116 1.237
0.25 1.61t 1.68t 1.79 1.49 1.56
0.67
0.27 1.48t 1.85 1.491 1.64 1.52 1.46t 1.52 1.49 1.50
0.44
0.68 0.37
1.44t 1.42 1.33 0.46 1.30 0.51
1.22t 1.28 1.17 1.17 1.19
0.49 0.50 0.59 0.68
* Atmosphere nitrogen except for case8 marked (a) (oxygen) or (b) (air)
t Carbon dioxide
determined volumetrically. B. STOICHIOMETRY O F T H E DECOMPOSITION
The stoichiometry of the decomposition has been studied only to the extent of determining the number of moles of carbon dioxide and free acid formed per mole of benzoyl peroxide decomposed. These ratios are represented by cy and p , respectively. The co-presence of the second-order reaction should be reflected
936
BENJAMIN BARNETT AND WILLIAM E. VAUGHAN
increasingly in the observed stoichiometry as the concentration of the peroxide is raised, for it is clear that if the concentration becomes great enough the secondorder reaction will become the main path (see equation 12). The data appear in table 4 and are plotted in figure 3. The a-and @-valuesare clearly dependent on the initial concentrations of peroxide. This is in accord with earlier data by Brown (7), who found that when two solutions of benzoyl peroxide in benzene, 0.1 M and 0.8 M a t the start, were decomposed to the Same extent, the carbon 2
0.4
1.2
0.8
,.P mokr proxide/kq. solution FIG.3. Decomposition of benzoyl peroxide i n benzene. concentration. T = 80°C.
Change in stoichiometry with
dioxide and benzoic acid appeared in the following proportions (on a solvent-free basis) : 0.8 M
Carbon dioxide.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Benzoic acid.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 0 . 2 per cent 23.9 per cent
12.3 per cent
35.5 per cent
The total free acid comprises benzoic and a very small proportion (6, 15) of p-phenylbenzoic acid. The right portions of the curves in figure 3 appear to extrapolate to moles carbon dioxide evolved = moles peroxide decomposed
a =
and moles free acid formed @
= moles peroxide decamposed =
’
DECOhfPOSITION O F BEKZOTL PEROXIDE.
I
93 7
but since P may increase indefinitely, no exact value can be estimated for these ratios from the figure. A more useful ~ a r i a b l eis the fraction f (see table a), which varies from zero to unity nhile P increases indefinitely from zero. In the plot in figure 4 of a and fl against f, the observed values, cy1 and pl, a t f = 0 are those of the 1;1 reaction (since then P = 0). As f -+ 1, kl (obs.) -+ k?P, so that the values, a2 and 02, vhen f = 1 refer essentially to the stoichiometry of the k2 reaction. The experimentally realizable region in figure 4 is, of course, limited to the interval 0 5 f < 0.805, since when f = 0.805, P corresponds to pure benzoyl peroxide. Alctually,the experimental limit is even lower because of the limited solubility of the peroxide in benzene a t 80°C. The realizable
0.2
0.4 1.
0.6
0.8
!k!L
k , (obs.)
FIG. 4 . Decomposition of henzoyl peroside in benzene. Stoichiometry as function of fraction decomposed in the higher-order reaction. T = 80°C.
range is indicated in figure 4. The data obtained corer a large portion of this interval. I . Stoichiometry of ihe first-order reaction The values of a and /3 a t P = 0 are a1 = 1.67 and 41 = 0.22, so
that
-
$- o'22) or only some 5.5 per cent of the reacted peroxide is not 2 accounted for by the carbon dioxide and free acid formed. This is in contrast to the results of Gelissen and Hermans 114), who a t a concentration of P = 0.92 (see figure 3 for comparison) found that about 30 per cent of the reacted peroxide
('"'
938
BEKJAMIN BARNETT AND WILLIAM E. VAUQHAN
appeared in the products as a dark, resinous material, and that a large portion of the acid was bound in the form of some unknown ester. Since aI does not approximate a whole number, a t least two over-all reactions must occur. The major over-all course is almost certainly the formation of biphenyl and carbon dioxide : 0
0
II
II
CeH6COOCCoH6 -+
CoH6CsH6
f 2c02
(13)
Benzoyl peroxide is known to decompose in the pure state by this path (9, 12, 13, 30) and recently it has been shown (20) to contribute to the carbon dioxide formation at moderate concentrations of the peroxide in benzene. The appearance of acid could be explained, perhaps, on the assumption of another reaction accompanying reaction 13, such as, for example: CeH5COOCCsH6 4- CsHs
I/
0
8
-+
CeHjCOH f
I/
coz
+
CeH6CeH6
(14)
0
which would appear to be pseudo first order under the conditions. Hot\ever, it seems more probable that only one first-order reaction is involved, the rate step being :
Only (2 - 1.67)/2, or 16.5, per cent of the radicals form acid bya reaction such as:
The greater part decomposes:
The fraction of the benzoate radicals formed which react by equation 16 is actually less than 16.5 per cent, because radical combination appears t o account for about 5.5 per cent. (18)
Using these equations and the limiting values of al and fil a t P = 0, it can readily be shown that of the peroxide decomposed, 83.5 per cent forms carbon dioxide and biphenyl, 11.0 per cent forms benzoic acid and, presumably, biphenyl, and 5.5 per cent forms phenyl benzoate. The appearance of phen7.l benzoate in the decomposition is well established (14).
DECOMPOSITION O F BENZOYL PEROXIDE.
I
939
2 . Stoichiometry of the second-order reaction
While the initial peroxide concentration is varied, or even in a single experiment as the peroxide decomposes, the a- and p-values of the two reactions may also vary, and the values of a2 and 8 2 found ivill therefore depend on the possihility of variations in a1and 81. However, the linear curve in figure 4 for p (the value of p actually determined) indicates that both p, and p2 are constants, independent of P.? On the other hand, the curve of a (the over-all value of a for the concurrent reactions) falls rather rapidly, and apparently linearly, a t the start; then as f approaches unity, CY appears asymptotically to approach unity also. .-Is P increases from zero, the concentration of benzoate radicals would also increase, and since as the concentration of peroxide is raised less solvent becomes available to react with benzoate by reaction 16. the carbon dioxide evolution may be expected to increase as the fraction f increases. Clearly, however, if this effect exists, it is not the factor that determines the shape of the plot of a against f , for othernise the initial slope of this plot could not be the observed maximum possible slope, that is, the slope for a1 = 1.G7 and a2 = 0. Thus it appears that the factor that controls the shape of the LT plot is the variation of a2 vith f . Some 80 per cent of the experimental points in figure 4 fall on the straight line connecting the values a1 = 1.G7 and a2 = .O (at f = 1). Combining this requirement of zero carbon dioxide evolution with that of the linear nature of the 8 plot (namely, Q 2 = 1) gives as the probable initial stoichiometry of the second-order reaction :
At higher values of j , however, the approach of the experimental curve atf = 1 to If Pr is the observed value of 8 , (A acid)l and A P I are, respectively, the number of moles of acid formed and the number of moles of peroxide decomposed in the first-order reaction and (A acid), and A P z are the corresponding quantities for the higher-order reaction, then:
But: j=-
:.
... !A5P 1 = 1f -f
APn API+AP,
81
82 =
+ 82
(f) ~
1f - f = ( 1 - f ) o l + f 8 z
1 + T f
Consequently, when 81 and 8%are constants,
d8r = (Pz - 81) df so t h a t the plot of PZ against f should he linear, with slope pz - B1
940
BENJAMIN BARNETT AND WILLIAM E. VAUQHAN
the limiting value a2 = 1 makes it probable that the final stoichiometry of the higher-order reaction is represented by equation 14. Presumably, the stoichiometry of the second-order reaction shifts gradually, as the peroxide concentration is raised, from reaction 19 a t the start, to the limiting reaction (14) a t high concentrations of peroxide. Equation 14 represents the main reaction of the “RH scheme” (14) in benzene a t high peroxide concentrations. The side reaction of this scheme:
can be accounted for as a summation of reactions 17 and 18.
8’. Discussion of the stoichiometries These results give an insight into the meaning of the accepted “RH scheme” not heretofore possible because of the absence of kinetic data. They show for TABLE 5 Decomposition of benzoyl peroxide i n benzene Extent of the higher-order reaction a t low peroxide concentrations CONCENTRATION OF PEBOXIDE
f Weight per cent
Mole per cent
Po in rnoleslkg.
1.0 2.0 4.0 6.0
0.324 0.653 1.32 2.01
0.0413 0.0826 0.165 0.247
0.037 0.070 0.131 0.185
the first time that the ratio of the main and side reactions (equations 14 and 20) can vary even in the same solvent, taking on different values a t different peroxide concentrations. The changing stoichiometry of the formally second-order reaction is consistent with the assumption of the formation of some intermediate complex, possibly (C,H,COO), C6&, in a preliminary, rapidly reversible step, and its subsequent decomposition in a bimolecular process by different paths. A non-radical process for decomposition has been advanced before, and also the possible formation of a mole-to-mole complex with the solvent (10, 11, 29). If the concurrent apparent second-order reaction does not involve the formation of free radicals, the fraction of the decomposing peroxide available to form radicals will be correspondingly reduced, and from the practical side it is important to attempt to estimate the effect a t peroxide concentrations normally used in technical applications, such as polymerization. For polymerization in benzene at SOT., if the monomer does not affect the kinetics, table 5 illustrates the extent to which the higher-order reaction enters. The f-values given are
DECOhlPOSITION O F BEKZOYL PEROXIDE.
1
941
the fractions of peroxide which would be “wasted” in not initiating chains, if a non-free-radical decomposition obtains.
Iv.
SUMMARY
1. The kinetics of the decomposition of benzoyl peroxide in benzene is strictly first order only a t infinite dilution. At all finite concentrations the first-order course is accompanied by a formally second-order reaction. At 80°C. the specific rate of the first-order reaction is kl = 0.158 hr.-l, and of the second-order process, kz = 0.145 kg./moles-hr. At this temperature the higher-order reaction becomes the major decomposition path at initial peroxide concentrations greater than 1.09 moles per kilogram of solution. The activation energy of the first-order reaction is 32 kg.-cal. per mole (55-8OoC.), while that of the side reaction is approximately 28 kg.-cal. per mole (70-80°C.). 2. At 80°C. the over-all stoichiometry of the first-order reaction consists chiefly of the formation of biphenyl and carbon dioxide. Benzoic acid is a minor product. The over-all stoichiometry of the higher-order reaction appears to shift gradually from the initial reaction:
to the final limiting course
as the peroxide concentration is increased.
REFERENCES BARTLETT AND ALTSCHCL: J. Am. Chem. SOC.67,812 (1945). BARTLETT AND ALTSCHCL:J. Am. Chem. SOC.67,816 (1945). BARTLETTAND COHEN:J. Am. Chem. SOC.66,543 (1943). BARTLETT AND NOEAKI: J. Am. Chem. SOC.68,1495 (1946). ( 5 ) BEREZOVSKAYA AND VARPHOLOMEEVA: J. Phys. Chem. (U.S.S.R.) 14 (2), 936 (1940). (6) BOESEKENS AND HERMANS: Ann. 619, 133 (1935). (7) BROWK:J. Am. Chem. SOC.62, 2657 (1940). (8) COHEW:J. Am. Chem. SOC.67, 17 (1945). (9) DIETRICH:Helv. Chim. Acta 8, 149 (1925). (10) ERLENMEYER: Helv. Chim. Acta 10, 620 (1927). (11) FICHTER AND ERLENMEYER: Helv. Chim. Acta 9, 144 (1926, (12) FICHTERAND FRITSCH: Helv. Chim. Acta 6, 329 (1923). (13) FICHTER AXD SCHNIDER: Helv. Chim. Acta 13, 1428 (1930). (14) GBLISSENA h D HERMANS: Ber. 68,285 (1925) and subsequent papers. (15) GELLISSEN AXD HERMANS: Ber. (8,765 (1925). (16) GELISSENASD HERMANS:Ber. 69, 662 (1926). (17) HEY ASD WATERS:Chem. Rev. 21, 186 (1937); see also reference 20. (18) KAMENSKAYA AND MEDWEDEV: Acta Physicochim. U.R.S.S. 13, 565 (1940). (19) LIPPMANN:Monatsh. 7, 525 (1886); see also references 14 and 27. (20) MCCLURE,ROBERTSON, AND CUTHBERTSON: Can. J. Research 20B, 103 (1942). (1) (2) (3) (4)
('eHs('0
f
C's13t;
0 ('nH,C'O
+
CGH~
+
('eHs('OH
II
II
I/
+
C'eH5
(3)
0 --*
('sHjC'OCoHs
(4)
0
.it finit? coiicentrations. ho\vever, n i'ornially second-order side reaction is observed, which predominates above i i certain critical peroxide conrentration. The stoichiometry of this higher-order reaction is initially ('sHsC'OOCC:sH,
/I
0
/I
0
+
CsHs
-+
C',HsCOH
/I
0
+
('eHsCOCsH5
II
0
(5)