Six papers are presented from the Symposium on Development, Stabilization, and Uses of Organic Peroxides, 748th Meeting of the Amerzcan
Development, Stabilization, and Uses of Organic Peroxides
Chemical Society, Chicago, Ill., Ahgust 7964. Chairman, Orville L. Mageli, Lucidol Division, Wallace @ Tiernan, Inc.
Three other
papers from this symposium are being published simultaneously in the
I
December issue of Industrial and Engineering Chemistry.
DETERMINATION OF DECOMPOSITION RATES OF DIACYL PEROXIDES J . E. G U I L L E T , ' T . E . B . T O W N E
Tennessee Eastman Co., Dimion
of
R . W A L K E R , M . F. M E Y E R , J . P. H A W K , AND
Eastman Kodak Co., Kzngsport, Tenn.
A method is described for determining the rate of thermal decomposition of peroxides a t elevated temperatures where the half life is of the order of only a few seconds. The apparatus consists of a heated metal capillary fed by a motor-driven syringe pump. The method is used to determine accurate rate data a t various temperatures for a number of commercial and experimental diacyl peroxides. The preparation of a new class of a-unsaturated diacyl peroxides i s described. Some of these decompose at rates faster than the saturated, straight-chain diacyl peroxides and are of interest as initiators for vinyl polymerization.
are known to decompose thermally in hydrocarbon solution by a unimolecular mechanism, following first-order kinetics. T h e rate of this unimolecular decomposition is highly dependent on the structure of the alkyl group. For example, Cooper (2) showed that whereas the chain length of the alkyl group had little effect, substitution in the a-position could change the rate by a factor of 105: which was attributed to steric factors. Similarly, unsaturation in the P-position gave a large increase in the rate, which was attributed by Cooper to the stability of the allyl radical formed after splitting out CO:! from the initially formed acyloxy radical. T h e studies by Bartlett and coworkers ( 7 ) on the rates of decomposition of various peresters indicate that both steric factors and the relative stability of the radicals produced influence the rate of decomposition of the peroxy group. Smid and Szwarc (5) studied the kinetics of decomposition of isobutyryl peroxide and attributed the higher frequency factor and lower activation energy to a lower energy path involving the more or less simultaneous rupture of three bonds thus : I A C Y L PEROXIDES
D dilute
2
0
Lll
1
I
R-C-0-0-C-R
0 3
111
T h e energy gained by splitting out the relatively stable
CO, molecule reduced the total activation energy required for the decomposition.
This lower energy path seems to be
1 Present address, Department of Chemistry, University of Toronto, Toronto. Canada.
made possible by steric strain introduced by substitution on the carbon a to the carbonyl group. Bartlett and coworkers ( 7 ) , explain the effect in peresters in t e r m of reduction in the entropy of activation owing to increased rigidity of the transition state. T h e purpose of the present work was to establish more accurate values of the rate constants and temperature coefficients for the unimolecular decomposition of a variety of diacyl peroxides and to obtain quantitative relations between structure and reactivity. Experimental
P r e p a r a t i o n of Peroxide. T h e peroxides were synthesized from the appropriate acid chlorides by reaction with sodium peroxide by the following general method : T h e preparations were carried out in a jacketed borosilicate glass flask equipped with a Teflon-coated magnetic stirrer. T h e flask was cooled to the desired temperature using either chilled water or 2-propanol which was carried through the jacket by a small circulating pump. Peroxides, whose shock sensitivities were unknown: were never made in quantities in excess of 1 gram. After information on shock sensitivity was obtained, the peroxides were sometimes made in batches of 5 to 10grams. A solution of sodium peroxide was made by adding 570 sodium hydroxide solution to the desired quantity of Becco L.S.P. 30% hydrogen peroxide to make a 2 to 1 mole ratio of sodium hydroxide to hydrogen peroxide. This solution was cooled to 0--5' C.: and the required quantity of acid chloVOL. 3
NO. 4
DECEMBER 1 9 6 4
257
Table I.
Acid Chloride 2-Ethylhexanoyl c,hloride Nonanoyl chloride 2-Nonenoyl chloride 3-Nonenoyl chloridr
Identification of Acid Chlorides
1 4331
64/10 61/0 2 4.5/0 1 46-48/0 8
2-Ethyl-4-methyl-2-pentenoyl chloride
62/31 81-83/20
4-Ethyl -2-ortenoyl chloride
70 2/1 8
2-Ethyl-2-hrwnoyl chloride
Table II.
1 4381
1 4660 1 4521
+ B' ( i / r
-
46 60 87 73 53 88
i/r&\) B'
A'
-2 341 f 0 -2 519 zt 0 - 2 137 zt 0 - 2 590 i 0 - 2 523 f 0 - 2 421 i 0 -2 521 f 0 - 2 212 f 0 -2 580 zt 0 - 2 504 f 0
Caprylyl Nonanovl Decanoyl Lauroyl Bis( 2-nonenoy 1 ) Bis( 3-nonenoyl ) Bis(2-rthyl heuanoyl) Bis[2-rthvl hexrnoyl) Bis(4-rthv I-2-octenoyl) Bis[ 2-ethvl-4-methvl-2-pmtenoyl)
ride. dissolved in about 20 ml. of heptane. vias dropped in slo\vly with rapid stirring. maintaining the solution below
5' C. T h e solution was stirred 2 to 4 hours a t this temperature, after which more heptane was added, and the solution was washed with cold 5% sodium hydroxide and then ice rvater until neutral. The yield was determined by titration or infrared assay. The purity of the concmtrated peroxide was determined by infrared analysis. I t was not possible to purify the peroxides subsequent to their preparation because they could neither be distillrd nor crystallized in most cases. Consequently the acid chlorides involved lvere subjected to rigid purification and analysis before they \vere used. T h e structures ivere confirmed by infrared and SMR analysrs. and purity was confirmed by preparing their methyl esters which were then subjected to quantitative assay by gas chromatography.
049
06060 052 109 181 081 082 146 149
L
Thermal decomposition unii
DEVELOPMENT
--7
256 030 314 247 514 '44 289
450 802 231 i 0 663
l/T,, 2 2 2 2 2
3 2 2 2 2
X IO3 582 628 550 628 634 013 987 572 629 677
CHLOSynthesis of Acid Chlorides. ~-ETHYLHEXAKOYL \vas made by reaction of redistilled 2-ethylhexanoic acid (b.p. 110" C. a t 6.5 mm.) and thionyl chloride, yield 717'. NOXANOYL CHLORIDEwas obtained from Distillation Products Industries. RIDE
to
Figure 1.
101 i 0 -6 639 zt 0 -6 88' zt 0 -'201 zt 0 -6 728 f 0 -5 640 i 0 -5 55" f 0 119 f 0 -6 656 f 0 -7
Data on the phi-sical constants. purity. and structure of the acid chlorides are given in T-able I along lvith the yields of peroxide obtained in various preparations. TVith the exception of bis(nonanoy1) peroxide: none of the peroxides listed has been described previously. Other peroxides used in this study !\;ere lauroyl (Alperox C). caprylyl (Lupersol MMO). and decanoyl peroxides (all obtained from LucidoJ Corp.). These compounds were used \vithout purification after it \vas established by gas chromatographic analysis of their decomposition products that the peroxides contained a t least 957i of the correct isomer. Dit&-butyl peroxide (97.07c minimum. obtained from Shell Chemical Corp.) was used \\ithour further purification.
Thermocouple Wells
PRODUCT RESEARCH A N D
c /c
20 rts >95 trans >95 >95
1 4682
log k = A'
I&EC
Peroaide I i e l d ,
Constants for Regression Equations
PfroxidP
258
i f i n i m u n i Purity 6) Gas Chromntoqrtiphy. ''I >95 >95 >95 trans 80 trans
nl?
2 .o
4 Figure 2. Rate of decomposition of bis(2-nonenoy1)peroxide 1 .c
SEC
7
40 000 -
lo'oOob
.
E
1,000 tog
c
g/1
100
LAUROYL
s DECANOYL
OCTANOYL
1 i
0 l0k
b Figure 3. Half life
, / , ,
&'--&:I
of lauroy', decanoyl, -1
(
0
IS
30
bo
120
T I M E , min
~ - N O S E N O YASD L ~-SOSENOYL CHLORIDE were prepared by reaction of heptaldehyde \i-ith malonic acid, Ivhich on treatment \vith pyridine. gave 2-nonenoic acid. This was converted to the acid chloride by reaction Lvith thionyl chloride. I t was carefull). fractionated to give both 2-nonenoyl and 3nonenoyl chloride. ~ - F , T H Y L H E X E S O Y CHLORIDE L \vas prepared from 2-ethylhexanoic acid by reaction \vith thionyl chloride. T h e acid \vas prepared by air oxidation of 2-ethylhexenal obtained from the aldol condensation of butyraldehyde. 2-ETHYL-4-METHYI.-2-PESTENOYL CHLORIDE. 2-Ethyl-4methyl-2-pentenoic acid \vas prepared by air oxidation of 2-ethyl-4-methyl-2-pentenal and convrrted to the acid chloride by treatment with thionyl chloride. 4 - E T H Y L - 2 - O C T E S O Y L CHLORIDE \vas prepared from the corresponding acid by treatment with thionyl chloride. T h e acid \vas prepared by treatment of 2-ethyl-hexanal with malonic acid in pyridine. Peroxide Titration Procedure. T h e peroxide sample was introduced into a n iodine flask using a calibrated pipet. T h e flask also contained 40 ml. of 2-propanol, 10 ml. of saturated sodium iodide in 2-propanol, and 3 ml. of glacial acetic acid as catalyst. T h e solution was refluxed under nitrogen for 30 minutes and titrated to a colorless end point with 0.01.V sodium thiosulfate solution. Rate Measurements. To obtain more accurate temperature coefficients, equipment had to be designed to study the rates of decomposition over a very wide range of temperatures. Conventional techniques were satisfactory at temperatures where the half life of the peroxide was of the order of 30 minutes or more. However. to study shorter reaction times. a tubular reactor, as shoxvn in Figure 1. was employed. This consisted of a coil of stainless steel tubing. 0.406 m m . in I.D. and 50.75 cm. long. Lvhich \vas wound around a I-inch diameter cylindrical copper block containing a cartridge heater. T h e block was insulated with asbestos, and the temperature was controlled with a LVheelco proportioning controller to Lsithin 1 0 . 2 ' C. After leaving the heated section, the tubing \vas brazed inside a piece of '8-inch I.D. stainless steel tubing which served as a Mater-cooled condenser. T h e inlet end of the tubing was fitted with a syringe fitting which was attached by means of a flexible metal tube to a 50-ml. syringe. T h e syringe was driven by a synchronous motor \\.hose speed could be varied over a \vide range by changing the frequency of the power supply. T h e temperature of the reactor was measured by a copper-constantan thermocouple fastened close to the
I80
T E 40~ - - ~
TEMPERATURE, ' C .
and octanoyl peroxides at various temperatures
1.9
2.2
25
2 T L"K)
28
3.1
x 103
middle of the tube. Electromotive force readings \viere made Lvith a potentiometer and converted to trmperature readings based on calibration against a standard thermometer. T h e volume of the heated section of the tube was 0.0640 ml.; the flow rate could be varied from 0.007to 1 . 5 ml. per minute giving contact times from 10 minutes to 2 seconds-the average density of solution \vas 0.'1 gram per ml. T h e rates Lvere determined by the following procedure :
A solution containing O.Sycof the peroxide \vas made u p in a light-\veight. completely saturated mineral oil and purged Lvith nitrogen. This \vas placed in the syringe and run through the capillary a t constant temperature. Peroxide assay \\'as carried out before and after the runs by the sodium iodide method (-I)or in certain cases by infrared analysis. The concentration of prroxide \vas plotted as a fnnction of time on semilog paper. giving in most cases a good straight line from Lvhich the half life of the peroxide a t the temprrature in question could be determined by intrrpolation. Figure 2 sho1l-s typical rate data. determined for bis(2-nonenoy1)peroxide at a slightly higher initial concentration (1.47,). \vhich shon. that the rate is first-order in peroxide over a t least a 10-fold range of concentration. T h e half life was converted to the first order rate constant by the relation: 0.693 k = -~ At lo\ver temperatures. where the half life exceeded one-half hour. the rates ivere determined by heating 3 to S ml. of 0.57, solution in borosilicate glass tubes in aluminum blocks thermostatically controlled to r0.3' C. T h e tubes \\'ere fitted \vith self-sealing gaskets and flurhed \vith nitrogen by alternate application of vacuum and dry nitrogen through a syringe needle. T h e tubes \vere maintained a t the desired trmpera-
Table 111.
Arrhenius
Peroxide
Capry-lyl Sonanoyl Decanoyl Lauroyl Bis(2-nonenoyl) Bis( 3-nonenoyl) Bis( 2-ethylhexanoyl) Bis(2-ethyl-2-hexenoyl) Bis( 4-ethyl-2-octenoyl) Bis(2-ethyl-4-methyl-2pentenoyl )
VOL.
Parameters for Various Peroxides Actiration Enpry). Frequency Kcal. /.Mole Fmtor. Sec -I 32 49 i 1 1' 9 . 8 x 1015 30 38 f 1 39 8 . 4 X 10l4 31 52 i 1 44 2 . 7 x 1015 32 95 i 1 1'3 2 2 x 10'6 1 . 6 X 10'j 30 '9 i 2 35 3 . 7 x 1014 2.5 81 i 3 40 25 43 i 1 32 1 2 >( 1014 32 58 f 2 06 1 6 X 10lfi 8 . 2 x 10'4 30 46 3
+
33 09 f 3 03
3
NO. 4
DECEMBER
7 1
x
1O'fl
1964
259
T.
1
rec
c
180
Id0
103
4G
T , "C
2w
260
80
I40
40
Figure 5. Half life of 2-ethyl-2-hexenoyl (curve 1 ), 2-ethyl-4-methyl-pentenoyl (curve 21, and 2-ethylhexanoyl peroxide (curve 3) at various temperatures
T ."C
Figure 4. tures Curve Curve Curve Curve
1. 2. 3. 4.
Half life of peroxides at various temperaDi-ferf-butyl peroxide (reference curve) Lauroyl peroxide (from Figure 31 Nonanoyl 0, 2-nonenoyl 0 , 4-ethyl-2-octenoyl 3-Nonenoyl peroxide
A
ture before the peroxide solution was added. Peroxide concentrations were determined before a n d after heating as described above. Results
R a t e constants were determined for each peroxide a t six temperatures over the range from 30' to 250' C. T h e d a t a were fitted to the best straight line using a regression equation of the form: log k = A'
+ B'
(117- -l/Tax.)
T h e regression coefficients obtained by this procedure for a variety of peroxides are shown in Table I1 with the appropriate 95% confidence limits. From these data, one can obtain the usual Arrhenius parameters, the activation energy EA. and the frequency factor, A . These are shown in Table 111. T h e straight-chain diacyl peroxides from Cs to C I Uhave activation energies of 31 to 33 kcal. per mole and are nearly identical within the precision of the experimental method. Nonanoyl peroxide (C,) containing a n odd number of carbon atoms may possibly have a value slightly lo\ver than the three peroxides containing a n even number of carbon atoms (Cg. CIO.and '212). although the difference is marginal since the 95% confidence limits overlap. Cnsaturation in the 2-position apparently does not affect either the activation energy o r the frequency factor; rvhereas, unsaturation in the 3-position does cause a significant reduction in the activation energy. This is shown when one compares the data for nonanoyl peroxide with those for 2-nonenoyl, 4-ethyl-2-octenoyl. and 3-nonenoyl peroxide. Substitution in the 2-position. as demonstrated previously 260
I&EC
PRODUCT RESEARCH A N D DEVELOPMEN1
by Cooper (2) and Smid and Snvarc (5). does cause a large reduction in the activation energy-shown here by the data on 2-ethylhexanoyl peroxide. T h e activation energy is significantly lo\ver than the value of 27.3 kcal. per mole determined by Smid and Szwarc for isobutyryl peroxide. This is probably due to the larger steric effect of an ethyl group compared with a methyl group. However, when unsaturation is also present in the 2-position, as in 2-ethyl-2-hexenoyl or 2-ethyl-4-methyl2-pentenoyl peroxide, the activation energy is about the same as for the straight-chain peroxides. Data of this type can probably be more readily interpreted from a practical point of view in terms of plots of the logarithm of the half life of the peroxide as a function of the reciprocal of the absolute temperature. Such a plot is shown in Figure 3 for lauroyl, decanoyl, and octanoyl peroxides. Each point represents the average of five or more titrations, and an excellent straight line is obtained over a range of nearly 100' C . As mentioned previously, there appears to be no significant difference in the rates of decomposition of these three peroxides. I n other plots the reciprocal scale is omitted for convenience. Figure 4 shows the large reduction in half life by unsaturation in the 3-position (curve 4, 3-nonenoyl peroxide). Figure i show-s the effect of 2-unsaturation in increasing the half life. Conclusions
These studies show that the rate of decomposition of aliphatic diacyl peroxides is controlled by t\vo main factors: T h e relative stability of the radical formed by splitting out
CO,. Steric hindrance about the carbonyl group caused by substitution on the adjacent carbon atom. kt'hile one can increase the rate of decomposition of a peroxide by increasing the stability of the radical formed, this
is iisually not a deqirable approach, since the radical may then br too stable to initiate polymerization or a t the very least \vi11 b r a much less efficient initiator. By placing a methyl or ethyl group in the 0-position. one can incrrase the rate of decomposition markedly without reducing the activity of the radical appreciably; however. these peroxides generally decompose too rapidly to be useful for many commercial applications. By combining substitution with unsaturation in the 2position, peroxides can be obtained which decompose a t a n intermediate rate: lvhich is of particular interest for commercial polymerization catalysts. Since a vinyl radical is a t least as active (and possibly more active) than the corresponding alkyl radical. these peroxides are effective free radical initiators ( 3 ) . T h e difference in rate is attributed to the fact that the alteration of the bond angles from the usual tetrahedral angle by the double bond reduces the steric effect of the a-substituent. ‘This effect can be clearly demonstrated by construction of the appropriate Fisher-Hershfelder models. T h e capillary tube reactor provides a useful way of obtaining rate data on the thermal decomposition of peroxides a t elevated temperatures \There the half life is of the order of only a
fe\v seconds. When combined with data obtained from conventional procedures a t lower temperatures, these d a t a allow more accurate calculations of the .4rrhenius parameters for thermal decompositions Acknowledgment
T h e authors acknowledge the help of M. V. Otis and V. \Yilson Goodlett in the structural determinations, and the many helpful discussions with H . TV. Coover, Jr. Literature Cited
(1) Bartlett. P. D.? in ”Peroxide Reaction Mech. Conf., Providence, K. I.. 1960,” J. 0. Edwards, ed.. pp. 1-10, Interscience, S e w York. 1962. (2) Cooper, \V., J. Chem. Soc. 1951, p. 3106. 13) Guillet. J. E., Hawk, J. P., Towme. E. B (to Eastman Kodak Co.), U. S. Patent 3,119,802 (Jan.28, 1964). (41 Kolthoff. I. M..Belcher. R.. “Volumetric Analvsis.” Vol. 111. p. 399, Interscience, New York. 1957. (5) . . Smid. ,J.>Szwarc. IM., J . Chenz. Phys. 24, 432 (1958). RECEIVED for review October 5, 1964 ACCEPTED October 14, 1964 Division of Industrial and Engineering Chemistry. 148th Meeting, ACS, Chicago, Ill., September 1964. \
I
,
I
THERMAL DECOMPOSITION AND APPLICATIONS OF n=BUTYL=4,4BIS(ferf-BUTY LPER0XY)VALERATE STANLEY W . B U K A T A , LEONARD L. Z A B R O C K I , M A R Y J A M E S R. K O L C Z Y N S K I , AND O R V I L L E L . M A G E L I Liicidd Dmirton. llhiiucr 3 Twrnan. Inc.. Buffalo. .V,
F. M c L A U G H L I N ,
Y.
The thermal decomposition of n-butyl-4,4-bis(terf-butylperoxy)valerate has been studied in the pure state and in n-dodecane. Kinetic data indicate a first-order decomposition a t least up to two half lives with an energy of activation of 39 kcal. per mole. At 100” C. the half life is 32.8 hours. A mechanism for the bond breakdown of the peroxide is proposed whereby the initial step is the homolysis of the -0-0of one of the fert-butylperoxy groups. The products of decomposition are explained by subsequent reactions. The main gases of the decomposition of the peroxide in the pure state are carbon dioxide, methane, and ethane. When the decomposition is carried out in n-dodecane, the maior gaseous products are carbon dioxide and methane. A formulation consisting of 50% peroxide on an inert filler attains an optimum cure for ethylene-propylene rubber in 18 minutes at 300” F. Resulting cures have low residual odor in both sulfur and sulfur-free systems. ECENTLY,
the Lucidol Division of Wallace S: Tiernan, new organic
R Inc.: has made commercially available a
n-butyl-4.4-bis(tert-butylperoxy)valerate (BPV): peroxide, ivhose structure is CHz l
0 I
0
C)
T h e pure peroxide is a colorless, nonvolatile liquid (vapor pressure less than 0.3 m m . of H g a t 100’ C.) which can be used to initiate free radical reactions a t temperatures above 100’ C. This article discusses the thermal decomposition of the peroxide in the pure state a n d in the presence of n-dodecane, and its application for curing ethylene-propylene rubber. lMost of the published work concerning the decomposition of dialkyl prroxides has been done on di-tert-butyl peroxide ( 7 , 2, 5-7). T h e rate-determining step appears to be the homolysis of the oxygen-oxygen bond :
11
CH:,-LCH,-CH,--C--O--CH,-CH,-CH,-CH, (1 l 0 I
6 I
c H:
