H. W. MELVILLE, J. C. ROBB AND R . C. TUTTON 163 STRUCTURAL FACTORS GOVERNING THE REACTIVITIES OF a -METHYLENIC GROUPS TOWARDS ACTIVE FREE RADICALS BY E. C. KOOYMAN Received 5th February, 195 I Twenty-two hydrocarbons of varying structures are compared with regard to reactivity towards trichloromethyl radicals, using their retardation of carbon tetrachloride addition to cetene as a basis. Reactivities are expressed as a semiempirical constant, which shows correlation with other numerical con- stants bearing on hydrogen abstraction reactions at a-methylenic groups. Structural influences on a-methylenic reactivities are discussed on the basis of degree and type of substitution a t the reactive group. The higher reactivity of compounds in which the a-methylenic group forms part of a ring system as compared with analogous open-chain compounds is tentatively attributed to an entropy effect.The results tend to suggest that hydrogen abstraction by free radicals involving hydrocarbons is largely governed by the heats of reaction.REACTIVITIES OF a-METHYLENIC GROUPS a-Methylenic groups * are often preferential points of free-radical attack resulting in hydrogen abstraction, the removal of these atoms being facilitated by the low dissociation energies of the corresponding carbon-hydrogen bonds ; these are associated with the appreciable reson- ance energies of the allyl- or benzyl-type radicals formed. These reactions are of primary importance in the autoxidation of olefins, of rubber and the fatty oils, and of alkylated aromatics.Further typical examples are the bromination by means of N-bromosuccinimide, as discovered by Ziegler, as well as dehydrogenation reactions occurring under the influence of sulphur, selenium and quinones.1 Recently, a hydrogen abstraction reaction of the above type was found to be responsible for chain termination in the benzoyl peroxide initiated addition of carbon tetrachloride to olefins : X’ + R-CH,-CHzCH, __t XH + R-CH-CH=CH,. . (I) I In this equation X’ denotes an active radical, i.e. a radical capable of continuing the chain reaction, e.g. by abstracting a chlorine atom from carbon tetrachloride to produce a new CC1, radical. The stable allyl-type radicals formed in (I) fail to do so ; in experiments with cyclohexene as the olefin, they were detected in the form of their dimer 3 : 3’-dicyclo-hexenyl.I n line with this, foreign compounds capable of giving stable radicals by reactions analogous to (I) were found to exert a marked retarding influence on the carbon tetrachloride addition reaction. The present paper describes a simple method for the quantitative comparison of re- activities of this type ; it is based on the magnitude of the above retarding effects. the propagation sequence of the carbon tetrachloride addition to olefins may be represented in a manner analogous to that obtaining for the “ab- normal ’’ addition of hydrogen bromide to olefins : CCl,’ + R-CH,-CH=CH, --f R-CH,-CH-CH,-CCl, . (2) (trichloroalkyl radical) I Outline of method.-According to Kharasch, Urry and Jensen R-CH,-CH-CH,-CCI , + CCl, + R-CH,-CHC1-CH,-CC1 , + CCl ,’.(3) I It might be expected, therefore, that the chain-carrying trichloromethyl radical and the trichloroalkyl radical would play a predominant part in the termination step (I), forming chloroform and a trichloroalkane respectively. However, detailed analyses 2h 9 of the products formed when reacting a * a-Methylenic is employed here to indicate the positions adjacent to one or two double bonds or aromatic nuclei ; i t may apply to substituted or unsubstituted methyl or methylene groups (-CH,) having at least one hydrogen atom. 1 E.g. Waters in The Chemistry of Free Radicals, 2nd edn. (Clarendon Press, Oxford, 1948), pp. 168, 241-3, 255-7. (a) Kooyman, Rec. t m v . chim., 1950, 69,492 ; (b) Kooyman and Farenhorst, Rec. trav. chim., 1951 (in press).Kharasch et al., J . Amer. Chem. Soc., 1947, 69, 1100. p It should be noted that the above analyses were only possible because of the relatively short overall kinetic chain lengths of the carbon tetrachloride addition. With normal a-olefins, about z yo of peroxide (based on the olefin) is required 2 * to give complete conversion of the olefin. However, 3 : 3-dimethylbutene-I, CH, I H,G-C-CH==CH, LH3 which does not contain any or-methylenic hydrogen atoms, is completely con- verted under the influence of as little as 0.15 yo. Here, the termination step must be of a differentE. C. KOOYMAN 165 threefold excess of carbon tetrachloride with olefin in the presence of benzoyl peroxide revealed the formation of nearly two molecules of chloroform per molecule of peroxide decomposed, whereas the amounts of trichloroalkanes were relatively unimportant (a few per cent based on peroxide).The other hydrides XH expected on the basis of eqn. (I) such as benzene and benzoic acid, which are formed from the phenyl and benzoate radicals arising from the peroxide, were also present in minor amounts. From the above data it was concluded that the trichloromethyl radicals furnish the main contribution to chain termination. Apparently, re- action ( 3 ) proceeds quite readily under the reaction conditions employed, giving the trichloroalkyl radicals little opportunity for reaction according to (I). MILLMOLES OF A ao 6 0 - 5 0 - 4 0 - SLOPE 55‘4 (BELOW 75% CONVERSION) MEAN DEVIATION 0.5 MlLLlMOLES OF BENZOYL PmOXlOE DECOMPOSED FIG.ca carbon tetrachloride bound to cetene (A) as a function of benzoyl (Initial molar ratio CCl,!cetene 3-00 ; temp. g ~ + &+ O C.) The overall kinetics of the system carbon tetrachloride + cetene (molar ratio 3.0) + benzoyl peroxide were investigated a t 918 f *‘C.* A linear correlation was found between the amount A of carbon tetra- chloride bound to olefin and the amount of peroxide decomposed AP, irrespective of the intake of peroxide : This empirical relation holds up to olefin conversions as high as 80 yo ; it may be readily explained when assuming (2) to be irreversible. In that case, all trichloroalkyl radicals eventually give products containing one molecule of carbon tetrachloride; the very small amounts of tri- chloroalkanes (below 5 yo based on peroxide, or less than 0.1 yo based on A ) may be neglected.This argument is not affected by telomerization reactions (addition of trichloroalkyl radicals to olefin rather than reaction with CCl,) as the telomers also contain one molecule of CC1,. peroxide decomposed. A P = k A . . - (4 Combination of (I) and ( 2 ) gives In this equation T designates the trichloromethyl radicals eliminated by * This temperature was chosen because it is easily maintained in the course of experiments with molar ratio 3.0, the heat of reaction being carried off by the gzntly refluxing carbon tetrachloride.I 66 REACTIVITIES OF a-METHYLENIC GROUPS ( I ) , Kt and K, are the rate constants for (I) and (2), respectively ; 0 repre- sents the olefin. All quantities are expressed in moles or millimoles.When only the CC1, radicals are operative in (I), (b) may be integrated to give (u). However, K (eqn. ( a ) ) was found slightly to increase (from 0.0180 to 0.0196) when decreasing the initial molar ratio of halide and cetene to 2 . 0 ; it showed a decrease to 0.0168 when using the ratio 4-2. This suggests that K contains small contributions from other components besides K (eqn. (b)) ; for example, the trichloroalkyl radicals will give increasing contributions to termination as the excess of carbon tetrachloride decreases. Assuming only the trichloromethyl radicals to be operative in chain termination reactions with retarders, the influence of retarders may be expressed by the following equations : or AP' dA - = K , J (0). . R Here, T' denotes the trichloromethyl radicals eliminated by the retarder. The amount of retarder R was much larger than that of the peroxide decomposed and was therefore taken as a constant.AP' represents the amount of peroxide consumed by the retarder ; in connection with (b), it equals the difference between the amounts of peroxide decomposed and the amount of peroxide which would have sufficed to effect the same extent of addition A in the absence of retarder. K, constitutes our measure of or-methylenic reactivity, k , applying to the same olefin (cetene) in all experiments. 1% may be determined from the experimental plot of 1/(0) as a function of A ; K, constitutes the slope of the plot AP'/(R) against 18. Both the amount of peroxide and the amount of retarder were varied in order to obtain a number of points throughout the range of olefin conversions from 10 to 60 yo M.In using eqn. (d), the possible influence of the reaction products of the allyl-type radicals formed in (I), e.g. dimers or disproportionation products, had to be neglected. However, they are only forming a small proportion of the total olefin present, except at very high conversions of olefin. Experimental Materials .-Analytical grade carbon tetrachloride was employed in all experiments. Benzoyl peroxide was purified by dissolving in chloroform and precipitating with methanol ; the purity of the samples thus obtained varied between 97.5 yo and 99.0 yo. The preparations and properties of the hydro- carbons employed are indicated on opposite page. Procedure.*-(I) The determination of A may be illustrated by the following example.25 ml. of carbon tetrachloride (259 mmoles) and 25 ml. of cetene (87 mmoles, 19-50 8.) were pipetted into a flask containing 250 mg. (97-5 % pure) of benzoyl peroxide (I-01 mmoles) and a boiling chip. The mixture was rapidly heated to 91' C and then placed in an oil bath maintained between 93' and 9 4 O C. In this manner, the reaction temperature was kept a t g1-92OC for two hours. Solid carbon dioxide was periodically added in order to avoid the entrance of air. The mixture was then cooled to room temperature and made up to IOO ml. in a volumetric flask by means of the required amount of carbon tetrachloride. 5 ml. of this solution contained 4-50 mg. of peroxide, indicating that 1534 mg. (0.634 mmole) of peroxide had been decomposed.The remaining solution was then evaporated in a weighed flask at reduced pressure below 70°C ; the last traces of carbon tetrachloride were removed by increasing the temperature and * Cf. ref. (zb). The carbon tetrachloride refluxed very slowly.E. C. KOOYMAN 167 PROPERTIES AND PREPARATIONS OF HYDROCARBONS EMPLOYED Compound n-Decane . . . iso-Octane (2 : 2 : 4-tri- methylpentane . Decalin (cis) . . Decalin(truns) . . Benzene . . . Toluene . . . p-Xylene . . . m-Xylene . . . Ethylbenzene . . Cumene . . . Mesitylene . . tent.-Butyl benzene . Allylbenzene . . Tetralin . . . Diphenylmethane . Dibenzyl . . . meso-2 : 3-Diphenyl- Triphenylmethane . butane Cetene n-Heptene-3 . . 2 : 2 : 5 : 5-Tetramethyl- hexene-3 Cyclohexene .. Fluorene . . . g : lo-Dihydroanthra- cene . . . Physical Constants t 1.p. (" C/mm. Hg) Found 173 99 I93 185 80 I11 138 138 I35 152-3 x63-4 I 68 158-60 IOo/ZO II3lI - - - 170137 96.0 122 83 - - Lit. I74 99 I93 185 80 138 I39 136 152-3 165 169 I11 158-60 207 261-2 - - - 155115 96.0 125 83 I - Found 1.4121 1.3916 1.4828 1.4700 1.5010 1.4966 1.4956 1'4970 1.4959 1.4913 1.4990 2.4927 1.5136 1.5464 - - - - 1.4411 1'4050 1'4113 1'4734 - - Lit. 1'41203 1.3916 1.4828 1.4702 1~50110 1.49682 1.49580 1.49715 1-49580 1.4913 1.4990 1'4925 1.5143 1.5462 - - - - 1.4411 1.4049 1'4115 1.4727 - M.p. (" C) Founc Preparation Commercial sample * * 1, >, Fractionation of a commer- cial sample through a 30 plate column * Commercial sample * * * * * I , s, IS 22 I 1 1 , 25 ,, I , I , Org.Synth., 2, 341 * Standard procedures Commercial sample * Meyer and Wurster (Ber., Standard procedures $ Cf. Siside and NozakiJ J . Amer. Chem. SOC., 1948, Org. Synth., 1, 548 $ * 1, 2 ) 187336,964 $ 70,776 Standard procedures Howard et al., J . Res. Nal. Bur. Stand., 1947, 38, 1 2 2, 365 § Standard procedures 3 Commercial sample $ Wieland, Ber., xgm, 45,492 * Repeatedly shaken with 96 "I,, sulphuric acid until the acid layer remained nearly colourless, washed t Most of the literature data are from the Handbook of Ckmistry and Physics, 31st edn. ; those for the $Recrystallized from absolute ethanol. by further reduction of pressure. Finally, the residue was heated for 10 min. a t I I O - I I ~ O C a t 0-1-1 mm. Hg ; its weight then amounted t o 23.83 g.; this was not changed by another 5 min. heating. In calculating A a correction was used for volatile peroxide decomposition products amounting to 50 % of the peroxide converted.2b Thus the original reaction product contained (23.83 x 20/1g - 19-50 - 0.09 - 0.08 =) 5-42 g. of carbon tetrachloride bound to olefin, or 35-2 mmole. Both this procedure and ( 2 ) were checked by experiments without peroxide ; under the experimental conditions used, all the unreacted cetene remained behind. Chlorine deter- minations in the residues tallied very satisfactorily with the results obtained by direct weighing. ( 2 ) Determinations of A in the experiments with retarders were made in the same way, taking account of the volatility of the retarders. The latter were mostly either sufficiently volatile to be removed along with carbon tetrachloride, or remained behind.(Tetralin gave some difficulties owing to its intermediate with water, dried with calcium chloride and distilled over sodium. alkylated benzenes were taken from Francis, Chem. Rev., 1948, 42, 126. Distilled over sodium.168 REACT IV IT IES O F a-METHY LEN IC GROUPS volatility.) Methods (I) and (2) required some practice ; in order to ensure reproducible results all manipulations had to be carried out in exactly the same way. Duplicate experiments yielded A values hardly ever differing by more than 0.5, A ranging from 8 to 45 in procedure (2) and from 5 to 75 in &IN MILLMOLES-I 0.110 0.100 0.090 I 0.080 a070 0.050 4040 0.030 a020 0.010 QOOO A (IN MILLIMOLES) FIG. z.-1/(0) as a function of A (0, = 87 millimoles) .procedure (I). (3) Determination of plot of 1/(0) as a function of A was made by determining the uncon- verted olefin (bromine addition method according to McIlhiney), taking aliquot samples of the original reaction products. This plot (Fig. 2) was then graphically integrated ; for convenience, the results were plotted as Jg = f ( A ) . Results The mean K , values obtained J% by plotting AP’/(R) against have been recorded in Tables I, I1 and 111. Figures placed immediately after the name of the retarder refer to the numbers of experiments made ; figures after f indicate the largest deviations from the mean values. For the unretarded addi- tion to cetene, the mean deviation from the mean value amounted to not more than I % (Fig.I). basis of eqn. ( d ) . Fig. 3 shows a number of plots obtained on the lo’. A PyR (IN MILLIUOLES ) 6.50 r 600- 550 - WO. - 4 5 6 - do0 - 3.50 - 300- 2.50 - 2.00 - ,1.50 - la0 - f* FIG. 3.-K, values from equation (d).E. C. KOOYMAN TABLE I.--K, VALUES OF COMPOUNDS HAVING NO WMETHYLENIC HYDROGEN ATOMS Compound n-Decane (2) . * . iso-Octane (I) . (2 : z : q-trimethylpentane) . Decalin (trans) (2) . Benzene (I) . ted-Butylbenzene (I) . z : z : 5 : 5-Tetramethylhexene-3 (2) Decalin (cis) (2) . . . Kr x 102 0'0 0'0 0.08 f 0.02 0.17 & 0.05 0.03 0'1 0'0 TABLE II.-INFLUENCE OF TYPES AND DEGREES OF SUBSTITUTION AT THE Q-METHYLENIC GROUPS Compound Benzene (I) . Toluene (4) . Ethylbenzene (4) . Cumene (3) . tert.-Butylbenzene (I) . p-Xylene (4) . m-Xylene (3) .Mesitylene (4) . Toluene (4) Kt x 102 0.03 0.42 f 0.01 1-28 f 0-ob 1-75 f 0'1 0'1 0.95 f 0-04 0.84 f 0.04 1-29 f 0.08 0.42 f 0.02 Compound Diphenylmethane (6) . Triphenylmethane * (5) Fluorene (4) . Cetene (35) . n-Heptene-3 (6) . Allylbenzene (6) . . Dibenzyl (5) . meso-z : 3-Diphenyl- butane (2) . KT x 102 3'35 f 0.2 7 z t I 4'9 0.3 I 2 f I 1-3 f 0.1 1-13 f 0.05 * This compound failed to give a satisfactory straight line. i This value was estimated as follows. Allylbenzene and carbon tetra- chloride (molar ratio 1/3) were reacted with varying amounts of benzoyl peroxide (temp.. g ~ + " C). The plot AP = f ( A ) showed a nearly straight line, some deviation towards lower A values being observed a t increasing conversions of allylbenzene. Assuming the ratio of the rates of addition of the CC1, radical (K, in eqn.( b ) ) to allylbenzene and to n-octene to be 0.7 (as roughly determined by Kharasch and Sage) and the k , for n-octene and cetene to be equal, this gives 0.7 X 0.17 = 12 X I O - ~ for K, (allylbenzene). This value obviously constitutes merely a very crude approximation. The slope amounted to 0.17. TABLE III.-INFLUENCE OF CYCLIC STRUCTURES ~~ ~~ Cyclic Compound K, x 102 Analogous Acyclic Compound Cyclohexene (8) . Tetralin (9) . g : 10-Dihydro anthracene (7) . n-Heptene-3 (4) . Ethylbenzene (4). Diphenyl- methane (6) . KT x 13 I Discussion The significance of K,.-Inasmuch as the plots obtained satisfy eqn. ( d ) the largest deviation from the mean slopes being about 5-6 yo, whereas the mean deviations are about z yo, both the experiments with and without retarder may be interpreted in terms of a single radical operative ICharasch and Sage, J. Org.Chem., 1949, 14, 537.1 70 REACTIVITIES OF a-METHYLENIC GROUPS both in termination and addition. No analytical evidence is available, however, that the trichloromethyl radicals alsc predominate in the experi- ments with retarders. In the above sense, our K , values constitute a measure of a-methylenic reactivity ; it is realized that small contributions from other than the trichloromethyl radicals are incorporated in K, in a manner not indicated by the experimental data, As only retardations are measured, reactions of active radicals not leading to termination, i.e. to stable radicals incapable of continuing the chains, escape observation.For example, paraffins might be attacked to a small extent, giving alkyl radicals, which would in turn attack carbon tetrachloride to regenerate trichloromethyl radicals. Triphenylmethane appeared to be the only compound showing a systematic deviation, viz. towards decreasing K, values at increasing 1% values. This is tentatively attributed to the special properties of the triphenylmethyl radical, the dimerization of which is known to proceed towards an equilibrium ; because of the very high steady state equilibrium concentration (" high " compared with the concentrations of the other stable radicals) chain-starting reactions by these radicals might occur. Structural factors affecting &.-The values of K, found for the different compounds investigated have been arranged in Tables I, I1 and I11 in order to show that ( a ) (cf.Table I) : little or no retardation occurs by compounds not containing a-methylenic hydrogen atoms, 2 : 3-diphenylbutane (Table 11) forming the only exception ; (b) (cf. Table 11) : a-methylenic reactivity is increased by. substitution at the a-methylenic groups, methyl, phenyl and vinyl having " loosening " efficiencies increasing in the order mentioned, whereas the first substituent has more influence than the second ; ( G ) (cf. Table 111) ; a-methylenic groups forming part of a six-mem- bered ring are more reactive than the corresponding open-chain compounds. Conclusion ( a ) is limited to the classes of hydrocarbons investigated, polycyclic aromatics having marked retarding effect^,^ which are probably due to the capturing of free radicals to form new, stable radicals, rather than to hydrogen abstractions.Thus, naphthalene was found to act as a retarder, whereas Kharasch and Dannley6 have shown both the I- and 2-naphthyl radical formed by decomposition of the corresponding dinaphthoyl peroxides to be quite reactive, abstracting a chlorine atom from CCI, with formation of the two corresponding chloronaphthalenes. The formation of naphthyl radicals therefore cannot lead to chain termination. It is interesting to note that the tertiary hydrogen atom in iso-octane appears to react quite slowly ; otherwise the formation of tertiary radicals would lead to chain termination ; the latter type of radicals has been found to be incapable of abstracting a chlorine atom from carbon tetra- chloride except when the free valence occurs at a bridgehead.7 The small difference found between the K , values of the two decalins is, admittedly, at the limits of our experimental accuracy. However, cis-decaljn has a somewhat higher heat of combustion (2-3 kcal./mole) and is more readily oxidized than the trans isomer ; this would suggest that the radicals formed from both isomers, when removing the hydrogen 5 Kooyman et al.(to be published shortly). 6 Kharasch and Dannley, J. Org. Chem., 1945, 10, 406. 7 Kharasch ef al., J . Amer. Chem. SOC., 1942, 64, 1621 ; 1943, 65, 2428. Criegee, Ber., 1944, 77, 22.E. C. KOOYMAN atom at the g position, are identical and that their formation is more facile with cis-decalin than with the trans compound, in agreement with our result.The very low value found for z : 3-diphenylbutane seems to form an exception, as this compound contains two a-methy'lenic hydrogen atoms. Moreover, its value could be expected to be higher than that found for dibenzyl. However, model considerations showed that the bulky methyl groups probably cause steric hindrance ; in line with this, dibxzyl is readily dehydrogenated by chloranil to give stilbcne, whereas 2 : 3-diphenylbutane fails to react.* With 2 : 2 : 5 : 5-tetramethyl hexene-3, one might have expected some influence of addition of free radicals to the double bond; here too, steric hindrance will prevent the addition. Moreover, this olefin even failed to add on bromotrichloromethane, a halide which is known9 to react quite readily with olefins having the double bond in non-terminal positions. The very low values found with benzene and with tert.-butylbenzene, besides showing the effect of absence of a-methylenic hydrogen atoms, also indicate that-if any reaction occurs at all-the products formed with trichloromethyl radicals are sufficiently active to continue the chain.With regard to ( b ) , the influences of substitution on reactivity as expressed by this rule would seem to be analogous to the influence on bond strengths. Few data are, however, available as regards the values of the dissociation energies of the C-H bonds in most of the retarders used. Table IV gives a number of dissociation energies recently pub- lished by Roberts and Skinner 1 0 which illustrates the influences of sub- stitution on C-H bond energies in simple hydrocarbons.TABLE IV.-DISSOCIATION ENERGIES OF C-H BONDS Compound I l- Do-€I kcal . I02 97'5 90.8 86-5 77'5 78 I02 Difference with Respect to DC-= in Methane (kcal.) 0 4'5 15'5 24'5 24 11'2 0 Similarly, the dissociation constants of a number of extremely weak acids l1 show the same trends as the corresponding K , values (Table V). Apparently, it makes little difference that the pK values apply to the formation of a negative ion, whereas K, applies to the formation of a free radical. The methylated benzenes are seen to possess nearly equal reactivities when taking account of the number of reactive H atoms. Recent bond energy data obtained by Szwarc and by Szwarc and Sheon l2 are in line with this result (Table VI). These results suggest that little or no inter- action occurs between methyl groups in the meta isomers, whereas a * Private communication by Dr.N. Dost of this laboratory. lo Roberts and Skinner, Trans. Faraday Soc., 1949, 45, 339. l1 (a) Conant and Wheland, J . Amer. Chem. Soc., 1932, 54, 1212 ; l2 Szwarc, J . Chem. Physics, 1948, 16, 128; Kharasch, et al., J . Amer. Chem. SOC., 1947, 69, 1105. ( b ) Gilman, (c) McEwen, J . Amer. Chem. SOC., 1936, Szwarc and Shcon, J . Chem. J . Amer. Chem. SOC., 1938, 60, 2336; 58, 1124. Physics, 1950, 18, 237.172 REACTIVITIES OF a-METHYLENIC GROUPS loosening of hydrogen atoms in methyl groups occurs in the ortho and para isomers. Dibenzyl forms an interesting case, its K; being only 0.3, whereas ethylbenzene has 0.65.As pointed out by Szwarc lS both the shortening of the central C--C link (to 1-48 A) and thermal data indicate that this bond possesses some double bond character. The fact that the ct-methylenic hydrogen atoms are somewhat more tightly bound may be related to the higher Do-= in ethene as compared with that in ethane. The relatively small difference in K,.' observed between triphenyl- methane and dipheny'lmethane (7 and 1-68, respectively) suggests that the C-H bond energies are rather similar. This is in line with current conceptions that th2 high stability of the triphenylmethyl radical as compared with the diphenylmethyl and benzyl radical is largely due to steric factors, the resonance energy of the triphenylmethyl radical being not much higher than that of the others, as the three bulky phenyl groups cannot lie in one plane.14 TABLE V.-DISSOCIATION CONSTANTS AND ct-METHYLENIC REACTIVITIES (pK VALUES GIVEN BY CONANT AND WHELAND~~ COMPARED WITH K,) I I Fluorene .Triphenylmethane . Diphenylmethane . Cumene . 25 33 35 37 47 7 3'35 1-75 TABLE VI.-DISSOCIATION ENERGIES AND K, VALUES FOR METHYLATEII BENZENES Compound Toluene . m-Xylene . o-Xylene . p-Xy lene Mesitylene . Dissociation Energy of C H Bonds in Methyl Groups (kcal./mole) 77'3 77'1 74.8 76.2 - Kr ( x 102) 0.42 0.86 0.95 1'29 1.02 K,'(x 14) (= Kr per Reactive H Atom) 0.140 0.143 0.1 70 0.158 0.143 With regard to ( c ) , this higher reactivity of the cyclic hydrocarbons may be caused partly by a difference in the entropy of activation.Thus, Price and Hammett 16 interpreted the higher entropies of activation for the reaction between semicarbazide and cyclic ketones as compared with those for the acyclic ketones in the following manner. The cyclic ketones will lose fewer degrees of freedom upon entering the (rigid) transition state than the acyclic ketones ; consequently, the activation entropy will be more strongly positive for the former class of compounds. Whereas the analogous pairs tetralin + ethylbenzene and cyclo- hexene + n-heptene-3 have K,' values differing by a factor of about 24 (1.8 and 0.65; 2.8 and 1.2, resp.), the difference between dihydro- anthracene and diphenylmethane (81 and 1-68) is much larger. Dihydro- l3 Szwarc, Faraday SOC. Discussions, 1947, p. 39. l4 Faraday SOC.Discussions, 1947. pp. 70, 71. l6 Price and Hammett, J . Amer. Chem. SOC., 1941, 63, 2387 ; Hammett et al., 1943, 65, 1824.E. C. KOOYMAN I 7 3 anthracene is obviously an extremely “ rigid ” compound ; the removal of a hydrogen atom from the 9-position can hardly be expected to pro- duce a radical of much higher resonance energy than diphenylmethyl. The high reactivity of fluorene may be caused mainly by the low dis- sociation energy of the C-H bonds in the methylene group, the fluorenyl radical possessing an even higher resonance energy than the diphenyl- methyl radical. Probably, the rigidity of the fluorene molecule is of secondary importance. Correlations between the K, values and rate constants of other reactions involving a-methylenic groups .-CHAIN TRANSFER CONSTANTS IN THE POLYMERIZATION O F STYRENE.-Gregg and Mayo l6 recently described chain transfer experiments involving styrene ; several of the solvents used by these authors were also studied by us.Inasmuch as alkylated benzenes can also react by adding on a radical, which enters Kr x 10’ 105 x CHAIN TRANSFER CONSTANTS FIG. 4.--K, values as a function of chain transfer constants for styrene polymerization (temp. 914 C ) . into the chain transfer constant, but not into K,, we have corrected the values given by Gregg and Mayo for the small contributions arising from this type of reaction ; it was assumed that this contribution would be the same for all compounds containing a phenyl group. The resulting values were then recalculated for our reaction temperature 919°C.The values thus obtained were plotted against the corresponding K, values It is seen that both constants are very nearly on a straight line.* As the chain transfer constant represents the ratio between rate of transfer with solvent for a growing radical and rate of addition to styrene monomer, it is a measure of a-methylenic reactivity in those cases where other types of reactions are playing a minor part. (Fig. 4)- Gregg and Mayo, ref. (13), p. 328. * The values for fluorene (G. and M., about 1100 x I O - ~ , K, about 47 x I O - ~ ) do not fit into this picture, the K , value expected on the basis of a linear relation- ship being about 115 x I O - ~ .I 74 REACTIVITIES O F a-METHYLENIC GROUPS OXIDATION RATES.-An extensive series of oxidation experiments with The vital propaga- unsaturates was recently summarized by Bolland.’‘ tion step in these oxidations may be schematically represented by RO,’ + RH -+ R0,H + R’.* (4) The rate constant for these steps ( k , in Bolland’s nomenclature) was calculated for various types of olefinic compounds ; among other things, the results led the author t o the formulation of a number of rules con- cerning a-methylenic reactivity. These rules, as well as the general trends in the numerical values obtained, were quite similar to ours. Thus, a fair agreement is observed in the relative reactivities of the radicals CCl 3, substituted benzyl and RO, towards different a-methylenic groups. Whereas both CCl,’ and RO,’ are fairly reactive as well as rather electrophilic in character, the substituted benzyl radical is relatively stable ; its more rapid addition to monomers such as maleic anhydride as compared with styrene itself tends to suggest that it possesses a nucleo- philic character.On the other hand, heats of reaction have been shown in several in- stances to be closely related to the activation energies when comparing a series of related exothermic reactions. Thus, Butler and Polanyi suggested a linear relation between these two quantities ; in reactions involving hydrogen abstraction at a-methylenic groups, the differences in the heats of reaction are equal to the differences in the resonance energies of the stable radicals formed. From this point of view, the above relation- ships between the K,, values and the other rate constants are not sur- prising; they are the more significant inasmuch as they apply to con- stants of differing orders of magnitude. From the above, it may be inferred that polar properties are relatively unimportant in hydrogen abstraction reactions, the bond strengths of the C-H links involved being more decisive. This inference is admittedly based on a limited number of data ; no pertinent numerical data are available as regards the influence of strongly polar groups on a-methylenic reactivity. Anyway, it stands in contrast to the strong influence exerted by polarity factors on the addition of free radicals to double bonds. As the above-described retardation method is both a simple and a rapid one, it might serve as a yardstick for investigating a-methylenic reactivities even in mixtures and, probably, for compounds in which the “ loosening ” of the hydrogen atoms is caused by other groups, such as in ketones. An extension of the method to highly reactive retarders, which requires the more reactive system styrene + bromotrichloro- methane, will be reported on later.lS The author wishes to express his thanks to the Management of the N.V. de Bataafsche Petroleum Maatschappij for permission to publish this communication. Koninklijke /Shell-Laboratory, A msterdam. l7 Bolland, Quart. Rev., 1949, 3, I . Butler and Polanyi, Trans. Faraday SOC., 1943, 39, 19 ; cf. Evans and Walling and Mayo, ref. (13), p. 295 ; Mayo Polanyi, Trans. Faraday SOC., 1938, 34, T I . e l al., ref. (13), p. 285. 19 Cf. Price, ref. (13), p. 304.