REACTIONS OF FREE RADICALS ASSOCIATED WITH LOW TEMPERATURE OXIDATION OF PARAFFINS BY E. R. BELL, J. H. RALEY, F. F. RUST, F. H. SEUBOLD AND W. E. VAUGHAN Received 5th February, 1951 An attem2t has b-en made to elucidate the elementary, free radical reactions which oscur in the low temperature oxidation of paraffin hydrocarbons. The problem has b-en approached through a study of the reactions of radical and mdecular sp3cies which are known or postulated to be intermediates in the oxidations. Foi this purpose a variety of aliphatic peroxides has been used both to provide the free radicals desired and to allow examination of the be- haviours of the peroxides themselves. An oxidation mechanism which describes the reaction sequence from the initial union of an alkyl radical and oxygen t o the formation of the primary stable molecular species has been postulated, and experimental evidence for each step has been obtained.Of particular import- ance in this scheme are the processes by which alkylperoxy radicals are con- verted to alkoxy radicals which are the immediate precursors of the primary stable oxidation products. The detailed mechanism of the low temperature oxidation of paraffin hydrocarbons has been a matter of speculation for many years.l With this material as a basis, we have attacked the problem by a study of the reactions of both known and postulated peroxidic intermediates. From this study a self-consistent mechanism has been evolved, and each step has been verified experimentally. The major contribution of this work is an explanation of the processes involved in the conversion of alkyl- peroxy radicals, through alkoxy radicals, to the primary stable products of the oxidation.The further transformations of these primary products may well take place in an analogous manner, but these possibilities are not within the scope of the present study. This picture of the low temperature oxidation process is illustrated in Fig. IA and IB, each step of which will be discussed in more detail in later sections of this paper. In summary, however, alkyl radicals pro- duced in the initiation process (Step I) react with oxygen quantitatively to yield alkylperoxy radicals (Step 2). These may then be converted to alkoxy radicals by (i) abstraction of a hydrogen atom to form a hydro- peroxide (Step 3A), followed by decomposition to alkoxy and hydroxyl (Step 4A) ; (ii) reaction with one another to yield alkoxy radicals and oxygen directly (Step 3B) ; or (iii) reaction with an alkyl radical (possible only a t very low oxygen concentrations) to yield a dialkyl peroxide (Step 3C) which will decompose to alkoxy radicals (Step 4C).Attack on a hydroperoxide to regenerate an alkylperoxy radical (step 3A’) has also been demonstrated. The alkoxy radicals are the immediate pre- cursors of the first stable oxidation products. By decomposition or free radical attack aldehydes or ketones are formed ; by hydrogen atom abstraction alcohols are produced ; and by association with alkyl radicals ethers may be formed in small amounts. It should be emphasized that the conclusions have been based on a study of the reactions of postulated intermediates, rather than on 1 For pertinent references see Lewis and Elbe, Combustion, Flames and Ex- plosions of Gases (Cambridge, 1938) ; Jost, Explosion and Combustion Processes in Gases (McGraw-Hill, 1946) ; Faraduy SOG.Discussions, 1946. 242BELL, RALEY, RUST, SEUBOLD AND VAUGHAN 243 observation of actual hydrocarbon oxidation, and demonstrate the value of this method of approach to the solution of mechanistic problems. Step 2. Formation of Alkylperoxy Radicals by the Reaction of Alkyl Radicals with Oxygen.-The present investigation has been coil- cerned solely with the propagation steps of the oxidation chain and no attempt has been made to specify the mode of initiation. It is assumed that the reaction begins by the production of an alkyl free radical by some means (Step I ) , and the union of this radical with molecular oxygen is the step which will introduce this discussion.The combination of an alkyl radical and oxygen to form an alkyl- peroxy radical is now generally accepted. Some information on the rate of this bimolecular process for the methyl radical is available from earlier studies, and additional data are reported below. A recalculation of data on the photo-oxidation of methyl iodide 3 has shown that methyl radicals combine with oxygen at about 1/8ooth of the rate at which they react with iodine. Also, the activation e n e r a for the oxygen combination has been estimated as I or z k ~ a l . ~ In the pressnt study, observations on the pyrolysis of di-tert.-butyl peroxide, a source of free methyl, in the presence of oxygen afford a comparison with the rates of formation of ethane and methane, the normal pyrolysis p r o d ~ c t s .~ As shown in Table I, in all experiments in which the available oxygen was not ex- hausted, hydrocarbon formation after the oxygen was admitted was vanishingly small. In view of this suppression, it is concluded that the only significant reaction of methyl under these conditions is a union with oxygen. CH,O, + CH, + CH,OH + H,CO is Considered less likely since, to conform to the data, a very high specific rate, relative to other methyl radical reactions, would have to be ascribed to this process. TABLE I.-EFFECT OF OXYGEN ON C2H, AND CH, FORMATION IN THE A possible alternative, the occurrence of the reaction DECOMPOSITION OF DI-tett.-BUTYL PEROXIDE Oxygen remaining .- Total CH, found as CH, + C,H, Total CH, released * . . CH, released before O2 added * . . 1 5 9 - 8 O 147.20 122.1° (m.1 (mm.1 (mm.) 34 I7 I 2 I 0 340 310 238 284 81 28 19 I 84 24 I9 2 < 0'2 251 I 22 Step 3. Formation of Alkyl Hydroperoxides by the Reaction of Alkylperoxy Radicals with Hydrogen Atom Donors .-This reaction is well known from experiments carried out in both the liquid and gas phases. In the former category typical examples are the oxidations of tetralin 6 and cumene,' in which the hydrocarbon itself serves as the hydrogen atom donor : Blaedel, Ogg and Leighton, J. Amer. Chern. SOC., 1942, 64, 2500. Bates and Spence, J. Amer. Chena. SOC., 1931, 53, 1689.Van Tiggelen, Ann. Mines Belg., 1941, 43, 117-44. 6 Raley, Porter, Rust and Vaughan, J. Amer. Chem. SOC., 1951, 73, 15. 6 Hartmann and Seiberth, Helv. cAim. Acta, 1932, 15, 1390. 7 Hock and Lang, Ber., 1944, 77, 257.244 OXIDATION O F PARAFFINS The oxidation of isobutane,s in which hydrogen bromide performs the function of donor, is a pertinent example of hydroperoxide formation in the gas phase : CH3 CH, 1 I I CH, I CH, CH,COO* + HBr -+ CH,COOH + Br' In the three cases cited the hydroperoxides may be isolated in nearly quantitative yield. (Part I of z Parts) X H i STEP 1 8. STEP4A STEP 2 R a R02H + R- a SfEP4C 0 2 32 1 Ror STEP 38 R02. t ' + 0 2 2 ROD ROD FIG. IB.-A generalized mechanism of the low temperature oxidation of paraffin Formation of Alkylperoxy Radicals by the Reaction of Alkyl Hydroperoxides with Free Radicals.-Although the abstraction of the hydrogen atom bonded to oxygen in a hydroperoxide has been postulated in the literature,@ the reaction was first verified in the vapour phase reaction of tert.-butyl deuteroperoxide with di-tert.-butyl and di- tert.-amyl peroxides, carried out in a flow system at 1 g 5 O .l ~ Methyl and ethyl radicals, derived respectively from the two dialkyl peroxides, attacked the deuteroperoxide to yield deuteromethane and deutero- ethane : CH, . (or C,H,-) + (CH,) ,COOD -+ CH,D (or C,H,D) + (CH,) ,COO. As will be explained in a subsequent section, methoxy and ethoxy radicals were also produced in the respective rcactions. Methanol4 was isolated from the reaction of di-tert.-butyl peroxide with tert.-butyl deuteroperoxide, hydrocarbons.Step 3A'. * Bell, Dickey, Raley, Rust and Vaughan, Ind. Eng. Chem., 1949, 41, 2597. Robertson and Waters, Trans. Faraduy SOC., 1946, 0, 201-210. l o Seubold, Rust and Vaughan, J . Amer. Chem. SOC., I951,73, 18.BELL, RALEY, RUST, SEUBOLD AND VAUGHAN 245 presumably by the attack of methoxy radicals on the deuteroperoxide : Although it is considered unlikely, exchange in the vapour phase between the deuteroperoxide and methanol cannot be excluded as a source of methanol-d. Formation of Dialkyl Peroxides by the Combination of Alkylperoxy and Alkyl Radicals.-As mentioned in Step 3A’, the re- action of tert.-butyl deuteroperoxide with di-tert.-amyl peroxide in the (Part I1 of z Parts) CH,O’ + (CH,) ,COOD -+ CH,OD + (CH,) ,COO.Step 3C. If RO* is o primoty alkoxy radicol. R‘C H 2 0. STEP R& CH 2~ R’CH~OH+R~ If R0.k o secondory olkoxy rodicol. R:CHO- STEP RltR’CHO R;CHOH+R* R;CO+RH R;CO+ROH Sy y;;P RbGO+Ri R;COH+R FIG. IA.-A generalized mechanism of the low temperature oxidation of paraffin hydrocarbons. gas phase at 195” yielded ethane-d and tert.-butylperoxy radicals. In addition, a small yield of ethyl tert.-butyl peroxide was also obtained. To determine the importance of this process, the same reaction was carried out at I ~ o ” , at which point 65 yo of the input dialkyl peroxide decomposed in the 2-min. residence time. On the assumptions that the rate of decomposition of ethyl tert.-butyl peroxide was the same as that di-tert.-amyl peroxide, and that every successful collision of an ethyl with a tert.-butylperoxy radical yielded ethyl tert.-butyl peroxide, the maximum yield of the mixed peroxide would be 23 yo of the input di- tert.-amyl peroxide.This calculation assumed that residence times of the two dialkyl peroxides were the same. Since, in reality, the ethyl derivative is formed throughout the reactor, the yield should be somewhat higher. Actua.lly, a 30 yo yield was obtained, indicating that the most important mode of disappearance of alkylperoxy radicals in the absence of oxygen was combination with alkyl radicals : (CH,) ,COO* + C&,* -+ (CH,),COOC,H,.246 OXIDATION O F PARAFFINS The decomposition of the mixed dialkyl peroxide has been reported l1 and the data will be summarized subsequently. The general reaction has also been confirmed by the isolation of methyl tert.-butyl peroxide in low yield from the reaction of tert.-butyl hydroper- oxide with di-tert.-butyl peroxide at 195~.The Formation of Alkoxy Radicals by the Interaction of Alkylperoxy Radicals.-In order to simulate more closely the con- ditions of an oxidation, oxygen more than sufficient to react with all alkyl radicals produced was included in the tert. -butyl hydroperoxide + di-tert.-amyl peroxide system, which, as shown in the preceding section, gives in the absence of oxygen a 30 yo yield of ethyl tert.-butyl peroxide at 180'. While those products typical of alkoxy radical reactions, e.g. ethanol, methanol, and minor amounts of acetic and formic acids and carbon monoxide (see Table 111), were isolated in good yields, no trace of either ethyl tevt.-butyl ar diethyl peroxide could be detected.I n addition, no normal products of the reactions of methyl or ethyl radicals, such as methane, ethane, or butane, could be found. From the evidence presented for Step 2, it would appear that all alkyl radicals were converted to peroxy radicals and that the latter reacted according to the following overall equation ROO. + R'OO. -+ RO. + R'O. + 0, to yield the corresponding alkoxy radicals. The same conclusion was drawn from a study of the oxidation of methyl radicals.s Further evidence for this transformation was derived from the decomposition of tert.-butyl hydroperoxide in various solvents.la The pertinent example is the case in which chlorobenzene served as the solvent.It is apparently highly resistant to radical attack and a very rapid chain decomposition of the hydroperoxide ensued at 140°, yielding tert.-butyl alcohol and oxygen almost quantitatively by the mechanism : Step 3B. (CH,) ,COOH -+ (CH,) ,Coo + 'OH (CH,) ,COO + (CH,) ,COOH -4 (CH,) ,CO HO' } 2(CH3) ,COO. + 2(CH3) ,COO + 0,. The mechanistic details of peroxy radical interaction cannot as yet be specified. The Thermal Dissociation of terf.-Butyl Hydroperoxide.- The decomposition of this hydroperoxide has been studied qualitatively in the gas-phase and quantitatively in the liquid phase.12 In the former case it was passed through an open Pyrex reactor with a molar excess of cyclohexene as a hydrogen atom donor at 260'. In addition to the ex- pected products (acetone, tert.-butyl alcohol, methane, methanol, carbon monoxide and water), a 13 yo yield of cyclohexanol (based on decomposed peroxide) was found.The most reasonable process by which this alcohol could be formed is a dissociation of the hydroperoxide at the 0-0 bond followed by addition of the hydroxyl radical to the olefin : Step 4A. (CH,),COOH + (CH,),CO* + HO. The possibility of unimolecular fission of the 0-0 bond was con- firmed by a study of the kinetics of the decomposition in n-octane in the range 150-180'. The reaction proved to be a combination of uni- molecular and chain processes, the observed " first order I' rate constant l1 Rust, Seubold and Vaughan, J . Amer. Chew. SOC., 1950, 72, 338. IZ Unpublished data from this Laboratory.BELL, RALEY, RUST, SEUBOLD AND VAUGHAN 247 increasing markedly with the initial concentration of the peroxide.The values of these constants at the various temperatures and concentrations employed are presented in Table 11. The true values of the first order rate constants were obtained by extrapolation to zero concentration on a TABLE II.-RATE CONSTANTS FOR THE DECOMPOSITION OF tert.-BUTYL HYDROPEROXIDE I N %?-OCTANE T, O K 423'43 423'39 423.15 422'98 433'08 433'19 433.13 433'54 433'01 442'92 442'97 443'14 442.92 453.01 452'96 453'03 422'96 45"96 452'99 Ik 0.04 - 10skI (sec.-1) extrap. 0.8, 2'57 6-99 18.2 plot of hobs against [ROOH],. From these extrapolated ra.te constants the activation energy for the process was calculated to be 39.0 f 0.6 kcsl./mole, in good agreement with the value calculated from thermo- chemical data as follows : z tert.-C,H, + zH + 2 ~ s o - C ~ H ~ , H,O(z) H2 + 0.50, H, + 2H 2 0 + 0, z iso-C4Hl, + 130, + 8C0, + 1oH,O(Z) gH,O(Z) + 8C0, + (tert.-C,H,O), + 11-50, (tert.-C4HgO), -+ z tert.-C,H,O z tert.-C4Hg + 2 0 -+ z tert.-C4Hg0 Thus Do-o (tert.-C,H,O) = 84.0 kcal. /mole tert.-C,H, + 0 + tert.-C,H,O H f O - t O H 0, + 2 0 ~ s o - C ~ H ~ , -+ tert.-C,H, + H 4CO, + 5H,O(Z) --f iso-C4Hl, + 6.50, 5-50, + tert.-C,H,OOH 3 4C02 + 5H,O(Z) tert.-C4HgOOH 3 tert.-C,H,O + OH Thus Do+ (tert.-C4HgOOH) = 38.5 kcal./mole. A EL=, kcal./mole - 172-l~ + 103.2 la - I 368.5 l5 - II7.2l4 + 67'5 + 1279.9 IS + 39'1 l7 - 168.0 - 84.0 - 101-l~ + 86*13 + 117.2 l4 + 6 8 4 ~ 3 ~ ~ - 664.0 13 + 38.5 l3 Baughan and Polanyi, Trans.Faruday Soc., 1943, 39, 19. l4 Gaydon, Dissociation Energies (Chapman and Hall Ltd., London, 1947). l6 Calculated from heat of combustion (Rossini, J . Res. Nut. Bur. Stand., l* Calculated from heat of combustion (Raley, Rust and Vaughan, J . Amer. 1935, 15, 357). Chem. SOC., 1948, 70, 88). 17 See ref. (16).OXIDATION OF PARAFFINS From the experimental activation energy and the various extrapolated rate constants, the entropy of activation for the unimolecular fission of the 0-0 bond is calculated to be 9-7 f 0.3 cal./mole deg. Interpretation of this entropy change in comparison with that observed for di-tert.-butyl peroxide, 14.5 cal. /mole deg.,ls is straightforward. The greater rigidity of the dialkyl peroxide is evident from a study of models, so that a greater increase in entropy on dissociation would be expected.Considerable interaction between the tert.-butyl group and the peroxidic hydrogen atom in the hydroperoxide is indicated, however, by the relatively high value of the entropy term observed for that decomposition. It is much more difficult to explain the coincidence of the energies of activation for the two decompositions when it is considered that the 0-0 bond dissociation energy in hydrogen peroxide is 52 kcal. /mole? Probably at least two opposing factors are operative, one inductive, the other perhaps steric. No definite answer to the problem can be made at this time. In summary, the overall mechanism of the decomposition of tert.- butyl hydroperoxide may be portrayed as : (CH,),COOH -+ (CH,),CO* + HO- (CH,),CO* --f (CH,),CO + CH,.X- + (CH,),COOH + XH -+- (CH,),COO. Re + (CH,),COO- --f (CH,),COOR (CH,) ,COOR -+ (CH,) ,CO- + ROO 2(CH3),COO* + 2(CH3),C0. + 0, wherein X = HO-, RO-, (CH,),CO*, CH,. and R = CH,* or a radical derived from the solvent. Further transformations of the alkoxy radicals will be discussed in a subsequent section. The Formation of Alkoxy Radicals by the Thermal De- composition of Dialkyl Peroxides .-This reaction was early proposed 2 o but unequivocal proof awaited the study of the decomposition of di- tert.-butyl peroxide in various solvents capable of donating a hydrogen atom. At 1 2 5 O , for example, in cumene,21 tri-n-butylaminc,21 and benzal- dehyde,22 tert.-butyl alcohol is formed in yields of 80, 95 and IOO yo re- spectively, the remainder being accounted for as acetone. (CH,),CO- + RH + (CH,) ,COH + Re.The radicals formed by attack on the solvent produced a dimer in the case of cumene and gave more complex association products in the other two solvents. Step 5. The Reactions of Alkoxy Radicals.-A study of the vapour phase decomposition of a series of alkyl tert.-butyl peroxides at 195' in the presence of a molar excess of cyclohexene as a hydrogen atom donor provided a means of comparing the stabilities of the alkoxy radicals generated.1' The results of this survey are summarized in Table 111, which shows the order of decreasing stability to be Step 4C. (CH,) ,COOC(CH,) , + 2(CH3) ,COO CH,O > C2H,0 > n-C,H,O > iso-C,H,O > iso-C,H,O = tert.-C,H,O The high yields of tert.-butyl alcohol in the decompositions of the n- and isobutyl and tert.-butyl peroxides suggest the possibility of intramolec- ular reaction to yield the appropriate butyraldehyde and tert.-butyl alcohol without the formation of the corresponding free butoxy radicals.The additional tert.-butyl alcohol is not due to the presence of a relatively l8 See ref. (16). l9 GiguCre, Can. J . Res., 1950, 28, I?. 2o Rieche, AZkyZperoxide und Ozonzde (Theodor Steinkopff, Leipzig, 1931). 21 Raley, Rust and Vaughan, J . Amer. Chem. SOC., 1948, 70, 1336. 22 Rust, Seubold and Vaughan, J . Amer. Ckem. SOC., 1948, 70, 3258. P- 30.BELL, RALEY, RUST, SEUBOLD AND VAUGHAN 249 high concmtration of butyraldehyde, since decomposition of ethyl tert.- butyl percxide with a.molar excess of isobutyraldehyde at 195" gave only an 11.2 yo yield of tert.-butyl alcohol. TABLE III.-REACTIONS OF (CH,) ,COOR WITH CYCLO~EXENE AT 1g5O RO Yield in moles/roo moles of Peroxide Reacted Products Derived from RO By H Abstraction 76 CH,OH 65 CH,CH,OH 30 CH,CH,CH,CH,OH 19 (CH,),CHOH 6 (CH,),CHCH20H 8 (CH,),COH By H Loss 8 CH,O 5 co 8 CH,CHO scot 10 CH,CH,CH,CH( 16 CO t 11 (CH,),CO 11 (CH,),CHCHO I4 CO t By De- composition - * 10 CH,O I C O t 26 CH,O 3COt 35 CH,CHO 23 cot 61 CH,O 6COt 94 (CH3)2CC Products Derived from t&.-BuO By Sta- bilization (CHa)&OH 1 0 I2 2 0 6 22 8 By Decom- position (CHa)&O S2 80 69 82 67 94 * Decomposition of CH,O* according to CH,O* -+ CH,O + H* cannot exceed 0-2 mole/Ioo moles of peroxide reacted, the amount of hydrogen produced. t Carbon monoxide is formed by decomposition of aldehydes following free radical attack. The assumed distribution is based on the stability of formaldehyde as observed in the decomposition of ethyl tert.-butyl peroxide with no additive present . If the relative stability of an alkoxy radical is defined as the ratio of the moles of alcohol produced to the sum of that quantity and the moles of decomposition products, the alkoxy radicals tested are ranked a3 follows : CH,O', 1-00 ; C,H,O' 0.86 ; n-C,H,O*, 0.51 ; iso-C,H,O', 0.25 ; iso-C,H,O' 0.08 ; tert.-C,H,O', 0.08. In addition to decomposing and participating in hydrogen atom transfer reactions, certain alkoxy radicals may associate with . copresent alkyl radicals to form ethers. Thus, a small (8 yo) yield of methyl ethyl ether was prcduced by the union of methoxy and ethyl radical: as supplied by the vapour phase decomposition of methyl tert.-amyl peroxide.23 Still lower yields would be expected with the other, less stable, alkoxy radicals. The authors are pleased to acknowledge the assistance of Messrs. D. 0. Collamer, Jr., and L. M. Porter on portions of the experimental work. Shell Development Company, California. Emeryville, 8, 23 Raley and Collamer (in press).