914
ROBERT c. L A M B
A N D JAS.
To t h e filtrate was added 20 ml. of pentane and the mixture was cooled t o 0" t o give 1.85 g. ( M r , ; ) of pyrazoline IV, m.p. 117-119". This material was identical in every way with t h e pyrazoline synthesized by reaction of t h e nitrosourea with lithium ethoxide in diethyl f ~ m a r a t e . ~ Thermal Decomposition of N-Nitroso-N-(2,2-diphenylcyclopropyl)-urea in Varying Concentrations of Cyc1ohexene.-[)ecompositions were effected in a reaction flask t h a t contained a magnetic stirring bar and was held a t 52" by refluxing t-butyl alcohol in an outer jacket. Before use, Eastman cyclohexene was distilled from lithium aluminum hydride and Phillips pure grade heptane was distilled through a column packed with metal turnings. Prior to t h e addition of the solvents, the vessel was brought to temperature and thoroughly flushed with d r y nitrogen. T h e appropriate volumes of cyclohexene and n-heptane were pipetted into t h e reaction vessel to give a total volume (25') of 50 ml. I n cilculating t h e concentration of the cyclohexene, corrections were employed for the change in density of cyclohexene upon heating t o i7.6°14 and the change in density of n-heptane upon heating t o 80".13 T o the hot stirred mixture was then added a weighed pellet (ca. 1.0 mmole) of the nitrosourea.15 T h e reaction was allowed t o proceed for 10 min. during which time gas evolution was measured. I n general, about 1.5 equivalents of (14) F . D Rossini, el n l . , "Selected Values of Physical and Thermodynamic Properties of Hydrocarbons and Related Compounds," Carnegie Press, Pittsburgh, P a . , 1933, pp. 65, 162 ; J Timmerman, "Physico-Chemical Constants of Pure Organic Compounds." Else\ier Press, New York, ?; Y , 1950, pp. 60, 205 (I:) T h e total yield of spiropentane and 1,l-diphenylallene was found For this reason, all of the runs using to vary with the batch of nitrosourea solvent mixtures were made from the same initial batch of starting material.
[CONTRIBUTION FROM THE
DEPARTMENT OF
GRADYPACIFIC1
Vol. 86
gas was evolved. A mark on the side of t h e reaction vessel and the absence of refluxing solvent in t h e condenser in the gas collection line established t h a t the volume of t h e solvent underwent no significant change during t h e course of t h e reaction. A t t h e end of 10 min., the reaction mixture was transferred by a pipet to a 100-ml. volumetric flask. T h e reaction vessel was rinsed with heptane a n d the washings were added to the volumetric flask. T h e solution was then made up to 100 m l . with heptane a n d stored in the refrigerator.'B T h e absolute amount of 1,l-diphenylallene was determined with a Beckman Model U U spectrophotometer employing the absorbance at 250 mp. T h e absolute amount of t h e spiropentane was determined by removing t h e hydrocarbon solvent from the reaction product and adding a known weight of methyl-a-methyl cinnamate (internal standard j t o the residue. Analyses were carried o u t by g.1.p.c. using a 10-ft. column ( I / * in. inside diarneter) packed with 5yc by weight of rZpiezon L on 00-80 mesh Cas Chrom Z using a column temperature of 230' and a flame ionization detector. Integrated areas were used for the calculations and were compared with standard mixtures of spiropentane and ester. I n all cases, t h e total yield of spiropentane plus allene ,anged between 7'5 and 81yo.
Acknowledgment.-The authors are most grateful to the National Science Foundation for support of this work. (16) 1,l-Diphenylallene tends to polymerize upon standing." This polymerization is greatly accelerated when t h e liquid allene is warmed. However, it was found t h a t under our reaction conditions (dilute solution, short heating periods) the allene is stable (17) L . Skattebol, Telrahedvon Lefters, 167 (1961).
CHEMISTRY, UNIVERSITY O F
GEORGIA, &kTHESS, G A . ]
Organic Peroxides. 111. The Behavior of Cyclohexaneformyl Peroxide in the Presence of Excess Stable Radicals. The Simultaneous Determination of Kinetics and Free Radical Efficiencies in the Thermal Decompositions of Free Radical Initiators1
c. L A M B 2 A N D JAS.
B Y ROBERT
GRADY PACIFIC13
RECEIVEDAUGUST13, 1963 T h e first-order rate constants and free radical efficiencies in t h e thermal decompositions of cyclohexaneformyl peroxide i n benzene in the temperature range 30-50' were determined spectrophotometrically by measuring t h e fading of galvinoxyl, which was present in excess. Experiments in which a!y-bis-(diphenylene)-8-phenylallyl (BDP.1) and DI'PH were used as scavengers in benzene a t $5' gave results similar t o those obtained in t h e galvinosyl experiments. A11 of the experiments indicate a low free radical efficiency (247c or less) for t h e decomposition of cyclohexaneformyl peroxide in benzene. T h e volatile products formed in t h e decomposition of cyclohexaneform~-Iperoxide in benzene containing excess galvinoxyl are ester, acid, bicyclohexyl, cyclohexane, and c y c l ~ h e x e n e . ~On t h e basis of t h e slight dependence of t h e free radical efficiencies upon temperature, it is argued t h a t these products are formed in cage reactions of radicals, rather than in rearrangement reactions of the peroxide.
Introduction
A new modification of an old method for determining free radical efficiencies was suggested recent1y.j T h e method consists of using an excess6 of a stable free radical as a scavenger for reactive free radicals formed in the decomposition of an initiator. The use of an excess of scavenger enables one to determine the free ( I ) Supported h y the Petrr,leum Research Fund administered by the American Chemical Society, Grant P I i F - S 6 % A ~ 4 . This paper was read a t the Suutheastern Regional Sleeting of the American Chemical Society in Charlotte, X C . , S o v . 13, 1 9 6 3 Paper I in this series is given as ref. 5 . Paper 11: R C I.amb. P LV Ayers, and SI K . Tuney, J . A m Chem. Soc., 86, ;348;3 :IQ63). ( 2 ) To a h u m correspondence should be addressed. ( t i ) P R F Predoctornl Fellow, 1962--1!~63, 1 4 ) I t has been reported by H a r t and n ' y m a n t h a t the only products of decmnposition oi cycluhexaneformyl geroxide in carbon tetrachloride a t 70' are cyclohexyl chluride and ester. Apparently, the peroxide behaves quite differently in the two solvents, ci H . H a r t a n d L). LX'yman, J . A m . Chem. .So(., 81, 1801 (1839). L.5) K C . I.arnli, €;. F. Rogers, J r . , G . C . Ilean, J r . , and F. LX', L-oigt, J r . , i b t r i . , 8 4 , 2G:j.i (1962). 6 )T h e word Y A C I S S , as used in this paper, means t h a t enough stable radical is used so t h a t some remains after all the initiator has decomposed. I n must of the experiments, the ratio of the molar concentrations of stable radical t u peroxide ( Z o / P d i is less t h a n one, inasmuch as the free radical eficiency of cyclohexaneformyl peroxide is low.
radical efficiency of the initiator and the first-order rate constant of its decomposition in the same experiment. The (oversimplified)' mechanistic scheme used in this work may be described as
K.
+ 2 --+
cage products stable products
where P = peroxide, ( 2 R . ) , = solvent cage containing a pair of reactive radicals, R . = reactive radical, and 2 = stable radical used as scavenger. Assuming the steady state in R.and (2R.),, one obtains - dZ/dt
=
2k,1fP, wheref
=
k,'(k
+ k')
which may be integrated with proper limits to give the two eq. 1 and 2 in
(z - Zm)/(Zo - Zm)
_____
f
=
= -kdt
( Z , - Z=)/2Pa
(1)
(2)
(7) R. M. X o y e s , J . Phys. C h e m . , 66, 763 (19611, and literature cited therein.
March 5 , 1964
THE KINETICS Series no. (no of expt )
AND
915
CYCLOHEXANEFORMYL PEROXIDE WITH STABLE RADICALS
TABLE I FREERADICAL EFFICIENCIES I S THE D E C O h f P O S I T l O S O F C T C L O H E X A S E F O R M Y L EXCESS STABLE RADICALS IN BENZENE Radical
los&a
& / P O range
10s(kd i s . d ) min - l b
t
P E R O X I D E I N THE PRESESCE
1/21 min.c
id
OF
T , OC
0.234 85 0.89 Too fast t o measure BDPX 4.80 BDP.4 4.80 0,44-0.89 Too fast to measure ,216 75 200 60 .44- 89 Too fast t o measure 4.80 BI3P.q BDPX 3 66 .50- .99 30 5 f 0 . 3 22.7 ,173 45 44.8 3 5 6 f 0 9 19.5 191 1 00 DPPH 11.23 Galv 143 2 0.49-0. 78 Too fast t o tneasure 191 60 184 50 4 6 8 f 2 3 14 8 .49- .97 Galv 143 2 168 45 31 2 f 0 . 7 22.2 .49- . 9 8 156.7 Galv. 8 (5) 30.8 f .9 22.4 ,160 45 0 61 190 6 9 (2IE Galv. 37.0 165 40 0.49-0 98 18.7 f . 5 156 7 10 ( 4 ) Galv 45.5 162 38.6 .52- 87 15.2 f .1 144.4 Galv. 11 ( 3 ) 12 ( 3 ) Galv. 156.7 .49- . 9 8 8 79 It . 4 5 78 8 161 35 119.3 135 30 5 81 i . 2 4 .27- . 4 3 138.0 Galv. 13 ( 3 ) a Z0,t h e initial molar concentration of stable radical, was the same for all the kinetics runs within a series. Stable radical concentrations In t h e BDPX experiments, the wave length used was 490 m p ; t h a t used in experiments were determined spectrophotometrically with DPPH was 518 nip; and t h a t used in experiments with galvinoxyl was 772 mp. The extinction coefficients of t h e three radicals a t these wave lengths were 26,289, 12,218, and 607.1° First-order rate constants, k d , were determined by least-squares adjustment of d a t a t o eq. 1. .\I1 runs were followed through 85% reaction. The number of spectrophotometric determinations within each run varied from 6 t o 13, excluding t h e necessary determinations a t t = p . The average value of k , ~is given for each series, along with t h e error between experiments (not intraexperimental precision). The average standard devi:ition in k d within individual experiments was f 3 5 , . Average half-life in rnin. d j = average efficiency. The average error in efficiencies within series was zk0.006. These errors appear t o have been random, i.e., efficlencies did not change in a regular fashion with changing ZolPo. e The kinetics samples used in these runs contained added cyclohexanecarboxylic acid a t 103(H.4) = 5.9 and 19.8 M. 1(1) 2 (2) 3 (2) 4 (3) 5 (2) 6 (2) 7 (4)
taining stable radical a t the same concentration as A system of this type places rather strict requireused in the kinetics vial (but no peroxide), were found ments on the behavior of its components. Ideally, to undergo no fading whatever during the period renone of the components (solvent, initiator, stable radiquired to complete the reaction. cal, products) should react with another component, In one experiment (expt. 17, Table 11), higher renor should any component other than initiator undergo actant coiicentrations were used, which necessitated spontaneous decomposition. Also, if a spectrophotothe use of several samples, each of which, after its metric method of analysis for 2 is used, the reaction heating period, was broken and an aliquot portion of products should not absorb a t the wave length used for its contents was diluted sufficiently with solvent so the analysis. If products do absorb, a high value of Z,,hence a low value of the efficiency (f),results (eq. that the absorbance due to the stable radical could be measured. 2, above). I t is therefore obvious that the systems used in such Results and Discussion experiments must be chosen carefully. G a l v i n ~ x y l ~ ~ ~ The results of 29 kinetics runs on the decomposition is known to be an excellent scavenger for reactive of cyclohexaneformyl peroxide, listed in order of inradicals'O and is also known to be relatively stable a t creasing stable radical concentration, are summarized low temperatures in benzene.l1.l2 We have found in Table I . The rate constants and efficiencies given t h a t galvinoxyl is stable in benzene up to 60°, when the for each temperature are, for the most part, average latter has been freshly distilled and vigorously flushed with carbon dioxide. a,?-Bis-(dipheny1ene)-P-phenyl- values obtained in two or more runs. allylI3 (henceforth called BDPA in this paper) is stable TABLE 11 in the same medium a t temperatures up to 55'. InasVALIDATION EXPERIMENTS: DECOMPOSITIOS OF much as cyclohexaneformyl peroxide decomposes a t a CYCLOHEXASEFORMYL PEROXIDE I N BENZENE substantial rate a t low t e m p e r a t ~ r e s a, ~series of experiments was conducted with this peroxide to de10% zt s.d., Run ZO,M P u ,M min. l 1 l 2 , rnin. T,'C termine the feasibility of using these two radicals and 14(IR)" 0.04 0 04 3 1 . 7 zk 2 . 1 21.9 45.3 DPPH in experiments of this type. Galvinoxyl was 15 (IR)" 0 .04 1 2 . 8 zk 1 7 54.3 38 6 used in the majority of the experiments. 16b 0 08 5 . 9 6 f 0 06 116 3 30 0 Most of the kinetics runs reported here in which the 17 0.04 .08 39 4 f 1 5 17.6 45 0 rates were determined by measuring the fading of the stable radical a t the wave length of one of its absorpInfrared band of the carbonyl group in t h e peroxide a t 1766 c m . - ' used for analysis. Kate determined by iodometric astion bands in the visible were performed as follows: say. Samples diluted by a factor of 25, and t h e 772-nip Each sample containing the stable radical and the galvinoxyl band used for analysis. The eficiency obtained in peroxide was degassed and sealed in a I-cm. square this experiment was 0.156. Pyrex tube which served both as kinetics vial and cui-ette for the spectrophotometer; thus each kinetics The rate constants obtained a t each temperature run involved only one sample vial. The run was conare fairly reproducible and do not change with a definite ducted a t reactant concentrations such t h a t the abtrend when the ratio of initial concentrations of stable sorbance due to the radical fell in the measurable radical to peroxide (Zo,'Po)is changed. This, along range. For each temperature, control samples, conwith the fact t h a t eq. 1 is obeyed with fair accuracy in each run through about three half-lives, is strong ( 8 ) G. hl Coppinger, J . A m Chem. S O L .79, , ,501 (1937). ((1) h l S. Kharasch and B. S. Joshi, J . O r p Cheni., 22, 1435 (1957), evidence that the reaction is zero order in the stable (10) P. 1) Bartlett and T . Funahashi, J . A m Chem. SOL., 84, 2596 radical, i . e . , there is no significant second-order reaction (1962) between the peroxide and the stable radicals under the (11) F I ) . Greene, U'. Adam, and J E Cantiill, Ibid.,83, 3-161 119G1) conditions used. Also, since the efficiencies do not (12) P. 1). Bartlett, B. A . Gontarev, and H. Sakurai, ibid., 84, 3101 (1962). change with any definite trend with increasing Zo Po, (13) C. F. Koelsch, ibid.,79, 4439 (19571, 64,474-1 (1932). -1
ROBERTC. LAMBAND
916
the stable radicals must scavenge essentially all of the reactive radicals which escape cage reactions. The results of the BDPA experiments (at 45') agree extremely well with those performed with galvinoxyl, especially when i t is considered t h a t the initial stable radical concentration (2,) is greater by a factor of 43 in the galvinoxyl experiments (series 4, 8, and 9, Table I ) . The results of the two runs with D P P H (series 5) are not significantly different from those obtained with galvinoxyl and BDPA when one takes into consideration the high rate of peroxide decomposition which is being measured by the method used. Some of our early products study work seemed to indicate t h a t a rather large yield of cyclohexanecarboxylic acid (0.4 mole,'mole of peroxide) is obtained when cyclohexaneformyl peroxide is decomposed in the presence of excess galvinoxyl. More recent work has caused us to revise the yield drastically downward (see Table 111), b u t in the interim the effect of added cyclohexanecarboxylic acid on the rate and efficiency in the decomposition of the peroxide was investigated. l 4 No effect could be detected under the conditions used (compare series 8 and 9, Table I ) .
CYCLOHEXANEFORMYL PEROXIDEI N BENZENECONTAISING EXCESS GALVINOXTL"
Product
Cyclohexanecarboxylic acid Cyclohexyl cyclohexanecarboxylate Bicyclohexyl Cyclohexene Cyclohexane Radical yield
GRADYPACIFICI
Vol. 86
in Table I1 is only O . 1 5 G j somewhat lower than the average value obtained a t the same temperature using lower reactant concentrations. T h e experiments involving cyclohexaneformyl peroxide and stable radicals in benzene are all essentially in agreement, both in respect to rates and free radical efficiencies, although the reactant concentrations are varied over a thousandfold range; and, a t given levels of 2 0 and Po, the ratio 20,Po is varied in some cases by a factor of two. The data are therefore valid, even if, in some cases, the experimental accuracy is not superb. A plot of log k d / T vs. 1;'T using the data in Table I shows a slightly S-shaped curve about the least-squares line. The parameters obtained are AH* = 20.2 + 0.5 kcal. and A S * = -10.5 e.u. Although this high negative value of AS* would not be in disagreement with a mechanism of peroxide decomposition in benzene involving rearrangements which proceed through cyclic transition states, other considerations lead us to doubt, if not entirely reject, such a mechanism. The main consideration which suggests a purely radical mechanism for the decomposition in benzene concerns the slight temperature dependence of the free radical efficiencies. The expression f = k ' ( k k') can be rearranged to give k,'k' = f / ( 1 - f). Combining this result with the transition state equation leads to eq. 3
+
TABLE I11 SOME PRODGCTS OF THE DECOMPOSITION OF
Yield (mo1e:mole peroxide)
JAS.
Yield ( e q u i v C-6 groups,'mole peroxide)
0.12 .40
0.12
80 08 28 .07
.01 .28 .07
2f
=
0 34
Total 1 69' Taken from the experiments Zo = 0.06 M; Po = 0.12 .If. This is equivalent t o 81.5',; of the C-8 a t 45' in Table 11. groups.
The rates of decomposition obtained by measuring the rates of fading of excess stable radicals in benzene (Table I) have been validated by independent measurement (Table 11). The rate was determined by measuring the disappearance of the peroxide carbonyl band. both in the presence and absence of galvinoxyl (runs 14 and 15), and, in addition, the rate was measured by iodometric titration (run 16). Of these, the poorest agreement is found in run 15, in which the half-life obtained (54.3 niin.) a t 3S.6' differs by 2070 from the average half-life obtained (45.5 min.) by measuring the fading of galvinoxyl (series 1 I , Table I ) . In Table 11, we have also reported the results of an experiment in which the rate was determined by measuring the fading of galvinoxyl, but in which both 20 and Po are considerably greater than were used in the experiments in Table I (run 17 in Table 11). The observed rate is 21% greater than the average rate obtained a t 45' a t lower concentrations (series 4. 8 , and 9 in Table I ) , and it is also greater than the rate measured by infrared analysis a t nearly the same reactant concentrations (run 14, Table 11). The reason for the discrepancy is not clear, but it cannot be ascribed to a direct reaction between galvinoxyl and the peroxide a t higher reactant concentrations, for such a reaction would destroy more galvinoxyl per mole of peroxide than would be destroyed in the reaction a t lower 20 and Po. Thus a direct reaction between galvinoxyl and peroxide would increase both the rate and the apparent efficiency, yet the efficiency obtained in run 17 ( 1 4 ) H H a r t and R . A . Cipriani, .I. A m . Chem Soc , 84, 3897 (1962)
where A H d * - A H c * = Ead- Eacis the difference between the enthalpies or energies of activation of the diffusion of radicals from the solvent cage and the reaction of radicals with each other therein, and the intercept of the equation gives the difference in the activation entropies of the two processes. Least-squares adjustment of efficiencies obtained with BDP.1 and galvinoxyl in Table I to eq. 3 gives the values Ead- Eac= 2.3 + 0.2 kcal. and A&* - ASc* = +4.2 + 0.7 e.u. Herk, Feld, and Szwarc have treated data analogously using a quite different system.l5 I t was found t h a t in the photolysis of azomethane in isooctane between 6-95' the only products are methane, ethane, and nitrogen, and all the ethane is formed in the cage reaction of methyl radicals. The yield of methyl radicals per mole of azomethane decomposed is given by the quantity CH4, Nz. In terms of the semantics used in this paper, .f = CH4/2?Jz,since the theoretical yield of methyl radicals from azomethane is 2 moles/ mole of reactant. It follows that f I ( 1 - f) = (CHI ' N2) ' ( 2 - CH4,'Nz). Using the data obtained in the most accurate series of experiments, Herk, Feld, and Szwarc plotted the logarithm of the latter value ns. 1,"T and obtained Ead- Eac= 2 kcal. ; since Eac = 0 for the cage recombination of methyl radicals, it follows that 2 kcal. is an estimate of the activation energy for diffusion of methyl radicals from the solvent cage in isooctane. We have recalculated this value somewhat more precisely from their data*deterrnined in the series of experiments using 3BOO-A.light, and obtained Ead- Eac= 2.17 + 0.14 kcal., and A&* - A s c * = f 4 . 9 + O,4e.u.lfi Herk, Feld, and Szwarc called attention to the similarity in the value of Ea1- Eac= 2.2 kcal., and the activation energy obtained from the temperature coefficient of the viscosity of isooctane, 2.1 kcal. Similarly, for the decomposition of cyclohexaneformyl peroxide (15) I,. Herk, M . Feld, and M . Szwarc, i b i d . . 89, 2448 ( 1 9 6 1 ) (16) I t is nut surprising t h a t this difference i n activation entropies is positive, both f o r t h e decomposition of cyclohexaneformyl peroxide and f o r azomethane. I t is prohable t h a t AS,,*will he positive for any initiator. and A S c * should always be negative: hence t h e difference must he positive.
March 5 , 1964
CYCLOHEXANEFORMYL PEROXIDE WITH STABLE RADICALS
in benzene, our value for the quantity Ead - Eac = 2.3 f 0.2 kcal., is identical with the activation energy obtained from the temperature coefficient of the viscosity of benzene, 2.4 kcal. l7 Here, then, are two quite different initiator systems in which the free radical efficiencies increase with decreasing solvent viscosity in quite an analogous way. Since i t is certain that the photolysis of azomethane proceeds by a free radical path only (that is, there is no rearrangement giving ethane), the similarity in the behavior of the two systems suggests t h a t the decomposition of cyclohexaneformyl peroxide in benzene is probably free of rearrangement reactions. The hypothesis developed here is fundamentally the same as that of DeTar and Weis, who have argued convincingly that in the decomposition of d-phenylvaleryl peroxide in carbon tetrachloride, the slight dependence of product yields upon temperature indicates t h a t the decomposition is probably all homolytic. In t h a t system, the free radical efficiency is half the yield of d-phenylbutyl chloride. From the average values of the latter obtained a t 55 and 7 i 0 , one may calculate Ead - Eac = 2.1 kcal., and A&* - A s c * = + 5 , 5 e . u . ; the difference in Ea's should be compared with a value of 2.8 kcal. obtained from the temperature coefficient of the viscosity of carbon tetrachloride. The absence of rearrangement reactions in the cyclohexaneformyl peroxide-benzene system has not, of course, been proved. The point being made is that unless log f / ( l - f) in any such system shows significantly more temperature dependence than log 117 for the solvent, the homolysis mechanism should suffice in the absence of convincing independent evidence to the contrary. The products study (Table 111) is not as quantitative as one would like; 84.5% of the C-6 groups have been accounted for, when the %f free radicals which become attached t o galvinoxyl during the decomposition are added to the yields of volatile products. Apparently, the four usual cage reactions are operative : (a) recombination and disproportionation of acyloxy and alkyl radicals, forming ester, acid, and cyclohexene ; and (b) recombination and disproportionation reactions of alkyl radicals forming dimer, cyclohexane, and cyclohexene. The most important of these is the recombination of acyloxy radicals with alkyl radicals, since ester is the predominant product. It is probable that the average activation energy for these cage reactions i. near zero. Finally, it should be remarked that kinetics experiments of the type described here have been attempted with several systems, using, in most cases, benzene as the solvent. Experiments with the remarkably stable red liquid radical, di-t-butyl nitroxide, I y gave unrealistically high rates in the decompositions of cyclohexaneformyl peroxide in benzene at 45', and in the decomposition of 6-heptenoyl peroxide a t T i 0 in toluene. BDPX is also very stable in benzene a t higher temperatures, and has been applied successfully to the decompositions of a series of p-substituted y-benzylidenebutyryl peroxides in benzene in the 3 1i O o range.?O Galvinoxyl apparently undergoes some spontaneous decomposition in benzene a t 60' and above when heated for several hours. T h e runs in series 6 in Table I were possible because the half-life of the per(17) A Weissberger, Ed , "Organic Solvents, Vol V I I , Technique of Organic Chemistry," 2nd Ed , Interscience Publishers, I n c . , New York, s.Y , p 24 (18) I ) . F IIeTar and C Weis, J A m C h e m Soc , '79, 3041 (1957). (IS) A . K . Hoffman and A T Henderson, i b ' d . , 83, 1671 (1961). (20) Unpublished work, P h 1) Dissertation, Leonard P Spadafino, August, 1963
917
oxide a t 60' is less than 5 min. ; the reaction was thus complete in 45 min., during which time no fading could be detected in the control sample. Experimental*' Galvinoxyl was prepared b y t h e method of Coppinger,8 a n d purification was accomplished by t h e method described by Bartlett and Funahashi,Io except t h a t our final recrystallization was from ethanol rather than carbon tetrach11)ride. BDPA was prepared by methods described by Koelsch,~3using t h e modification described b y Solar arid Lindquist for the preparation of a,?-bis-(dipheny1ene)-0-phenylallylalcohol . 2 2 Cyclohexaneformyl peroxide was prepared by adding the acid chloride t o a slight excess of sodium peroxide in an ice-waterether slurry. T h e peroxide was recrystallized twice from methanol; t h e methanol was removed each time b y decantation. Finally, benzene portions were a d d e d , then subsequently removed b y evacuation in order t o remove the last traces of methanol. Although this peroxide is obtained in well defined crystals when recrystallized from methanol (using a Dry Ice-acetone b a t h ) , t h e peroxide decomposes rapidly in the melt a t room temperature. Therefore, as soon as the methanol was removed, t h e peroxide was dissolved in benzene, and the concentrated solution was frozen and stored a t - T O " . Kinetics Runs.-The benzene used in t h e kinetics experiments was of reagent grade, and was freshly distilled just prior t o use. After distillation, t h e benzene was flushed vigorously with carbon dioxide b y immersing a carbon dioxide inlet under t h e surface of t h e benzene. The gas flow r a t e was adjusted so a s t o cause vigorous agitation of t h e benzene. T h e benzene was flushed thusly for about 5 min. T h e peroxide stock solution was prepared from t h e concent r a t e d solution described above. T h e initial peroxide concentration was determined b y iodometric titration of three aliquots of t h e peroxide stock solution. A stock solution of t h e stable radical in benzene was then prepared, and aliquots of the two stock solutions were mixed a s desired, then sufficient benzene was added t o give a n initial absorbance due t o t h e stable radical which fell in a desirable range. Kinetics vials were prepared from 1-cm. square Pyrex tubing.23 Each vial was equipped with a flat b o t t o m , and was connected male joint. T h e circular t u b e leading to a t t h e t o p to a 10/30 t h e joint was previously constricted. The vials were filled through t h e constriction with t h e kinetics sample with t h e aid of a hypodermic syringe equipped with a long needle. The vials were then attached t o a vacuum manifold. The contents of t h e vials were then frozen, t h e system was degassed, t h e solution was melted, t h e system was flushed with carbon dioxide, and the cycle was repeated. The vials were then sealed off under vacuum a t the previously prepared constriction. The vials were wrapped with aluminum foil prior t o degassing. The wrapping remained on t h e vials during t h e course of t h e kinetics r u n , except during spectrophotometric analysis. The b a t h temperatures were controlled t o 0.02"; temperatures were measured with 0.1' S B S standardized thermometers. I n order t o take readings, t h e vials were removed from t h e bath a t intervals and cooled with ice water t o 0 " ; t h e bath oil was removed with organic solvents and t h e spectrophotometic reading taken. The vial was wrapped again with foil and returncd t o the bath as soon as possible. Kinetics runs in which infrared or iodometric analyses were used were performed by standard methods. Isopropyl alcohol was used a s titration solvent in the iodometric runs. Product Analyses.-Products were analL-zed by v.p.c. methods completely analogous t o those used for the decompositions of 6-heptenoyl peroxide in toluene described in paper I1 in this was I used for the analysis of cyclohexane and series.' Column .cyclohexene; c ~ ) l u m nB was used for bic?-clohexyl and ester, The cyclohesanecarboxyIic acid was extracted from a decomposition mixture with dilute sodium bicarbonate a n d , after isolation, it was estimated by titration with standard sodium hydroxide ~ 0 1 tion.
Acknowledgment.-Ii'e are indebted to the Petroleum Research Fund for support; to the University of ( 2 1 ) A line-operated Beckman Model I I U spectrophotometer, equipped with a specially built box-like cover for the cell chamber, was used f o r measuring stable radical concentrations spectrophotometrically A per^ kin-Elmer Model 421 infrared spectrophotometer was used in kinetics runs A Beckman G C - 2 gas chroma. determined by infrared analysis (Tahle 1 1 ) tograph was used in the products study (Table 111). (22) S . L Solar and R hl I i n d q u i s t , J . A m C h r m .Voc , 81, 4286 (1960). (23) This tubing, obtained from the Fischer and Porter C o . , Warminster, P a . , had two opposite sides frosted, the other two sides were polished.
~ -
918
A. WILLIAMJOHNSON, VICTORJ. HRUBY,AND
Georgia Department of Experimental Statistics for the statistical adjustment of the d a t a ; and to Dr.
[CONTRIBUTIOX FROM THE CHEMISTRY
Chemistry of Ylids.
X.
JOSEPH
L. WILLIAMS
Vol. 86
Leonard P. Spadafino, who prepared our sample of BDPA.
DEPARTMENT, UXIVERSITY
OF XORTH
DAKOTA,GRAND FORKS, N D.]
Diphenylsulfonium Alkylides-A Synthesis of Epoxides1
B Y A. WILLIAMJOHNSON, VICTORJ. HRUBY, AND
JOSEPH
Stereoselective
L. WILLIAMS
RECEIVED SEPTEMBER 16, 1963 T h e preparation and properties of diphenylsulfonium benzylide (8) and butylide ( 9 ) are described. Both \-lids can be dissociated into carbenes and Dherivl sulfide T h e benzylide 8 reacts stereoselectively with benzaldehydes to form tvans-stilbene oxides.
.
.
Introduction From a study of the properties of phosphorus ylids2 i t became obvious that ylids containing heteroatoms other than phosphorus should be capable of existence. The major requirement appeared to be the ability of the heteroatom to stabirize the ylid carbanion, usually by overlap of the free electron pair with the vacant, low-energy d-orbitals of the heteroatom. In instances where this was not possible, such as many nitrogen y l i d ~ the , ~ substances were capable of only fleeting existence. A t an early date Ingold and Jessop,4 although apparently not using this line of reasoning, prepared a sulfur ylid, 9-dimethylsulfonium fluorenylide (1). In more recent years Johnsonsa and Wittigsb have investigated the chemistry of ylids containing group V heteroatoms other than phosphorus. Some time ago we undertook to investigate the physical and chemical properties of several new ylids and chose first to examine the chemistry of sulfonium ylids. We were particularly anxious to determine how these ylids would react with carbonyl compounds under the conditions of the well-known M'ittig reaction.6 Since all ylids are nucleophiles, albeit of varying strength, we expected they would undergo a t least the first step of the n'ittig reaction, i.e. the attack of the ylid carbanion on the carbonyl carbon to form a betaine intermediate (e.g., 2 ) . Whether or not the betaine would dissociate into an olefin (e.g., 3) and an oxide of the heteroatom group ( e . g . , 4) was expected to depend on the bond energies involved. In our original work on the chemistry of sulfonium ylids7 we showed that 9-diinethylsulfoniuni fluorenylide (1) reacted with benzaldehydes to form a mixture of the benzalfluorene oxide ( 6 , R = C6H4X)and a rearranged alcohol (7, R = CsH4X). The mechanism shown in Chart 1 was tentatively proposed to account for the observed products. This mechanism was based on the relative carbonyl reactivity sequence, analogy with other sulfonium salt displacements, and (1) (a) For paper I X in t h i s series see A . W . Johnson, Chem. I n d ( L o n d o n ) , Ill9 (1963) ( b ) T a k e n in p a r t from the thesis of X'ictor J . Hruby presented to t h e Graduate School of the University of N o r t h 1)akota in partial fulfillment of t h e requirements for the degree . ~ f Master uf Science. ( c ) Part of this work w a s originally presented a t the 142nd A u g u s t , 1062 Sationai hIeeting of the American Chemical Society. 1)ivision of Organic Chemistry, Atlantic C i t y , S . J , September 10, 1 9 6 2 , Another portion w a s announced in a preliminary c o m m u n i c a t i o n , J A m . C h e m S o r , 84, 3386 (1962) (d) W e g r a t e f u l l y acknowledge t h e financial support of t h e N a t i o n a l Science Foundation (Grants S o G-l7:345 and GP-l:321) ( 2 ) A \T, J o h n s o n a n d R B l . a C o u n t , ? ' r i r n h r d i o i i , 9, 130 (1960) (3) G U'ittig and W Tochtermann, Chum Rei , 94, 1692 (1961) (4) C K . Ingold and J A Jessop, J C h r m Soc , 7 1 3 (1930) (.5) (a) A W J o h n s o n , J Oig ( ' h e m , 26, 18X (1960), ( b j M. C Henry a n d G Wittig, J A m . C k e m Soc , 82, 503 (1960) ( 6 ) S Trippett, "Advances i n Organic C h e m i s t r y , " Vol I , R A R a p h a e l , E. C Taylor. and H W y n b e r g . Ed , Interscience Publishers, I n c , S e w York, N Y , 1960. p. 83 (7) A W . Johnson and R . B L a c o u n t , Chrm I n d . ( I m n d o n ) . 1440 (19.58): .I A m C k r m . Soc , 83, 417 (1961)
the analogy with the known Somnielet rearrangement of the original ylid 1. Epoxide formation, rather than olefin formation as in the Wittig reaction, was not completely unexpected on the basis of the difference in bond dissociation energies of dimethyl sulfoxide (89 kcal./niole)sa and triphenylphosphine oxide (128 kcal./ mole)8b and the known reluctance of hydroxide to attack trivalent sulfur b u t not tetravalent p h o s p h o r ~ s . ~ On the basis of this work we predicted that most, if not all, sulfonium ylids would form epoxides upon reaction with aldehydes and ketones. This expectation has been verified in the much later work reported by Corey, et al., with dimethylsulfoxonium methylidelOa and dimethylsulfonium methylidelOh and the more recent reports by Franzen and co-workers using phenylmethylsulfonium methylide. IOc To develop our original reaction between sulfonium ylids and carbonyl compounds into one of practical value for the synthesis of epoxides, two modifications would be required. First, the reactivity of the ylids would have to be increased since the fluorenylide (1) CHARTI
I
+S\
/
CH3 CH3 1
1
RCHO
r
R 7
t
-
CH3 R
P
6
\ CH I
R 6
( 8 ) (a) H. Mackle, Teltahedron, 19, 11.59 (1963); ( b ) A F Bedfvrd a n d C T Mortimer, J Chem SOL.,1622 (1960) (9) C. K Ingold, el d , i h i d . , ,531, 5 3 3 (1933). ( I O ) (a) E J Corey and M C h a y k o v s k y , J A m . Chem. Soc , 84, 867 (1962); (b) i h t d . , 84, 3782 (1962); (c) V . Franzen and H . E IWiessen, T e l r a h r d v o n r.cilevs. 661 ( 1 9 6 2 ) ; Chem R e v , 96, 1881 '1968)
