GENERAL DISCUSSION 308 GENERAL DISCUSSION Prof. W. A. Noyes Jr. (Rochester) said Recent experiments by Mr. Martin in our laboratory indicate that methyl radicals from the photo- lysis of mercury dimethyl upon reaction with oxygen at low pressures give CO and CO,. Not enough results were available when I left to allow calculation of rate constants although the rates are in approximate agreement with the acetone results. Some uncertainty exists concerning the nature of the primary process in mercury dimethyl so that many more experiments may be necessary (only two had been made before I left) before quantitative conclusions can be drawn. Other results have been obtained with acetone-oxygen mixtures by Messrs. Marcotte and Durbetaki at intensities about 10 times those pre- viously used both in the presence and absence of mercury vapour.It is concluded that mercury vapour does not affect the results by more than the experimental error. At these high intensities in the absence of oxygen at 120Othe acetyl radical seems to be sufficiently stable to permit appreci- able reaction of the type . CH + COCH = CH,COCH and possibly - (1) - (2) Thus a t this temperature the yield of CO + CO may be as much as 2COCH = (COCH,) . six times the yield of CO in the absence of oxygen. At zooo however the rate of decomposition of COCH is sufficiently rapid so that the results at these higher intensities agree with those previously found. These experiments are being performed with a view to permitting an analysis of products other than CO CO, CH and C,H,.Mr. Marcotte has also made experiments at 225OC. In general the results are similar to those obtained at lower temperatures but i t is evident is setting in. Thus CO + CO may have yields of 3-6 or more. This fact that at this temperature one has reached a point that chain propagation may limit the temperature range over which i t will be possible to deter- mine the activation energy for HCO decomposition. The fact that the CO is less sensitive to oxygen pressure and is higher than would be ex- pected from the low temperature results may be due to hydrogen abstrac- tion by radicals other than methyl. Dr. Finkelstein has preliminary results on the decomposition of diethyl ketone in the presence of oxygen.At 25°C the yield of CO falls as the pressure of oxygen increases indicating that the COC,H radical reacts with oxygen to give something other than CO at this temperature. In this respect i t seems to resemble the acetyl radical. However the quantum yield of oxygen disappearance is still rising at the highest oxygen pressure used (0.18 mm.) and has reached a value of 6. Complete oxidation of I molecule of ketone to CO and H,O would require 7-5 molecules of oxygen. Since small amounts of C hydrocarbons are found either more than I molecule of ketone disappears per quantum (doubtful a t 25OC) or the oxidation is proceeding quite completely. Further conclusions are not warranted at this time. At 200' C the yield of CO + CO seems to reach a constant value of 3 to 3.2 with increase in oxygen pressure the CO being about 2.3 and the CO about 0.8.However the quantum yield of oxygen disappearance has reached 7 at 0.78 mm. oxygen pressure and shows no sign of reaching a constant value. C hydrocarbons are still formed to the extent of about 0.3 mole per quantum even under these conditions. The total yield of CO + CO is about the same as for acetone thus in- It would be premature to give a mechanism based on these results. dicating that either a CH radical or a C,H radical yields either CO or CO when i t reacts with oxygen. However since the CO yield falls with increase in oxygen pressure for acetone and rises under similar circum- GENERAL DISCUSSION equation Two primary mechanisms suggested themselves namely * * CH,I-+ h9 CH +- I CH + 0 -f CH,O + OH e + HCOOH +- HCOOH.+ HCO + OH* occurs in preference to HCO + RH + HCOH + R HCO -f H + CO. or 3 09 stances for diethyl ketone the explanation for the two phenomena must have important differences. Dr. R. Spence (Harwell) said I would like to recall some experiments by Dr. Bates and myself on the oxidation of methyl radicals produced by the photodissociation of gaseous methyl iodide a t room temperatures. The products formaldehyde methylal water and iodine were identified and determined quantitatively and the reaction was shown to follow the qCH,I + 2 0 % -f (HCHO) + (CH,O),CH + H,O + P I . (1) - ( 2 ) and CH + 0 -+ CH,O,. The absence of any inert gas effect which might have been expected with mechanism (2) seemed to favour the hydroxyl mechanism (I) but Blaedel Ogg and Leighton failed to detect the OH radical by its ab- sorption spectrum and therefore proposed that a mechanism of type (2) is operative.Prof. Noyes’s results at high temperatures suggest that the fonnyl radical is produced from the interaction of methyl radicals and oxygen but the above results indicate fairly clearly that i t is not formed at room temperatures. This discussion has drawn attention to the fact that after twenty years we are still unable to give precise answers to such fundamental questions as how long such entities as 0 and CH can stay together on collision and what type of stabilizing collisions are subsequently required. Dr. N. Uri (Munchester) said The HCO radical was postulated by Marcotte and Noyes as an intermediate product in the photolysis of acetone in the presence of 0,.It appears to me very important to learn more about this radical which may also play an important role in the process of photo-synthesis. I consider that formaldehyde may be formed from formic acid by an electron transfer i.e. HCO + R . H 3 HCOH + R where the bond R H has a relatively small bond dissociation energy. On the other hand i t is postulated that HCO decomposes into H + CO. In this connection i t would be interesting to have more data on the formation of formaldehyde in the experiments discussed by Prof. Noyes so as to ascertain under what conditions Prof. A. R. Ubbelohde (Belfast) said It may be useful to reconsider to what extent the recombination of two radicals as in CH + CH -f C,H H+O,+HO - (B) requires a ternary collision to stabilize the product.Why do the energy- rich products not fly apart again almost at once as they would in simple atomic collisions such as H + 0 -f HO ? In the theory put forward in 1935 for the formation of peroxide radicals * R + 0 -f R02* (A) (1) . (2) . . R02* + RH + ROOH + R . I attributed the temporary stabilization of the RO complex to “ quantum smudging ” notable for R > C,H, but more rigorous conditions should apply when R is small as in reaction (B) and especially (A). Bates and Spence J . Amer. Chem. SOC. 1932 53 1689. Blaedel Ogg and Leighton J . Amer. Chem. SOC. 1942 64 2499.3 10 GENERAL DISCUSSION Prof. S . W. Benson (California) said Prof. Ubbelohde has raised a question with regard to the allowability of two-body recombinations such as have been employed by the present authors. I believe that this question can be answered with considerable more certainty than has heretofore been the case. A recent theory of Dr. N. B. Slater has shown that to a first approximation the rate of unimolecular decomposition of an activated complex with energy E in excess of a critical energy E* needed for decomposition is given by k(E) = (1 - f>"-l where Y lies in the range of normal frequencies (1012-1014 sec.-l) and n is the number of normal co-ordinates required for the description of the energy molecule decomposition.- E* + kT The so average that for any n v and with E* energy we can in calculate excess of the E* mean has life t ( c ) = I / K ( E ) . If we use mean values such as E*/kT - 40 ; v = 1013 sec.-l we find (with Slater) that t(E) w 10-l3 sec. if n = I ; t ( ~ ) E I O - ~ sec. if for n = I) are very much larger for low temperatures. Thus for (except e*/kT n = - 5 140 and at t ( ~ ) room E 10-4 temperature sec. if n = and 10. if we These set mean n = 5 life-times the mean ethane life of an excited molecule will be I O - ~ sec. These results imply that for the reverse reactions the recombinations of free radicals complexes may be formed having mean lives sufficiently long to assure even at very low pressures quite efficient stabilization by collisional deactivation.This will be true for the high-temperature pyrolysis reactions and will be more pronounced for photochemical re- actions where temperatures are considerably lower. A reaction to form the radical HO is one that is close to the permissible limits. Here n < 3 ; E* > 45 kcal. and the average complex formed at 130O C has a mean life t ( ~ ) < v(kT/E*)2 w 3 x I O - ~ sec. which is of the order of collision times a t S.T.P. At temperatures of 500' C to 600° C or higher this mean life is much shorter and i t is probable that most of the HO complexes decompose before they can be stabilized. In the more complex methyl radical the theory gives a reasonable accounting of the high collision efficiency which has been observed for their recombination.It will not take many degrees of freedom in a com- plex with such a high critical energy to give i t a very long mean life. Finally we may say that recombinations will manifest their three-body character when the complex has a low critical energy the temperature is high and the combining radicals are not complex. Dr. C. A. McDowell (Liverpool University) (communicated) I should like to mention that Dr. Thomas and I have found that in the presence of a large excess of oxygen the acetyl radical is quite stable up to about zoo0 c. Dr. A. D. Walsh (Leeds) said Were any analyses for formaldehyde carried out by Noyes and Marcotte ? A t the higher temperatures normally used for the study of methane oxidation it was difficult not to suppose the reaction of CH radicals and 0 to produce HCHO took place.Dr. Peter Gray (Cambridge) (communicated) I am very interested to read of these experiments on the peroxides and in particular of the role of the alkoxy radical RO. Peroxides are encountered as an important feature of some of the chemiluminescent (cool-flame) processes in hydro- carbon oxidation and the alkoxy radicals are presumably also present. Such RO radicals are also important in the thermal decompositions of the alkyl nitrates and nitrites. The initial step is the fission of the molecule to give the alkoxy radical and the appropriate oxide of nitrogen. Both these processes may be accompanied by a luminescence.41 Methyl McDonell and Thomas J .Chem. Soc. 1949 2208 2217. Gray and Y o f f e Proc. Roy. SOC. A 1949 200 114 fi Gray xg51 unpublished work. GENERAL DISCUSSION nitrite gives nitric oxide which may not be expected to oxidize the CH ,O- . Thus this chemiluminescence observed with nitrates nitrites and in the oxidation by nitrogen dioxide of the simple alcohols may be a property of the alkoxy radical itself. Furthermore from experiments on the nitrates values may be ob- tained under favourable circumstances for the heats of formation of the alkoxy radicals. The heats of formation of the other species involved and the energy required to break the 0-N bond are needed the latter may be identified with the energy of activation derived from measure- ments of initial rates.Such values may be employed both as a check on other measurements and as a means of calculating provisional values of heats of formation of peroxides of which the activation energies of decomposition but not the heats of combustion are known. The heats of formation of the methoxy and ethoxy radicals may be obtained from the following data on nitrate and nitrite esters. (i) CH,ONO,(g) = CH,O + NO ; Eact METHOXY COMPOUNDS Qj = 29.4 (ii) CH,ONO(g) = CH,O + NO ; Eact = 39-5 kcal. - 8.1 Qf(CH30) = - 2 , - 20.9 Qj(CH,O) = = 36'4 + 1.2 8 , - 8.1 Qf(C,H,O) = + z'o, The formation of each radical from its elements is numerically small and errors in any of the primary quantities appear magnified.The values (- 0.4 f 1.2) kcal./mole and (2 f 2) kcal./mole might be assigned to the CH,O and C,H,O radicals respectively. Qf = 16-7 ETHOXY COMPOUNDS CZH~ONO,(~) = C&',50 + NO ; Eact Qj = 33.4 Sir Alfred Egerton (ImperiaE College) said I can allay Dr. Stevenson's anxiety about the measurements of heat of formation of peroxide which are mentioned in the preliminary draft of the paper by myself Emte and Minkoff. The values found though the subject of careful measure- ments are unconvincing; for instance the heats of formation of the peroxides are in most cases greater than those of the corresponding more stable substances. It is not easy to determine accurately the heats of combustion of these explosive peroxides. We intend to redetermine them and to withdraw for the present the last section of our paper referring to the heats of formation of peroxides.Dr. A. G. Gaydon (Imperial College) said I should like to point out that the values for the dissociation energies of H, 0 and OH taken from my book are for oo K not 298OK. Fortunately the errors about 0.8 kcal./mole nearly cancel out for the equations as used. Dr. A. D. Jenkins (Courtaulds Ltd Maidenhead) (communicated) Vaughan ei al. report that vapour-phase pyrolysis of methyl-tert.-amyl peroxide (I) gives an 8 % yield of methyl ethyl ether which they ascribe to vapour phase combination of a methoxy radical with an ethyl radical. The scheme of decomposition of this peroxide will be substantially CH / CH,-O-O-G-CH -+ CHSO + O-C-CH = 39-5 kcal.CH / Gray and Yoffe J. Chem. SOC. 1950 3180. 312 GENERAL DISCUSSION bond Instead of the combination reaction suggested above i t is quite possible that the ether could be formed by reaction between the ethyl radical and the substrate by attack at the -0-O- CH3 CH3 CZHS + CH300C-CH3 -+ CH30CzHs + -0-C-CH3 / \ C2HS CZHS C2HS / \ Whilst no data are available for estimating the activation energy of this process from experimental results one can make a comparison with E for the alternative hydrogen abstraction reaction CH C,H + CH,OOC-CH / -+ C,H + radical Thus the general reaction \ in the following way. In the former reaction an -0-0- bond is broken to form an 0-C bond so that the exothermicity will be of the order (go-55)=35 kcal.and we can represent this process by the diagram FIG. I. In the second case the reaction is nearly thermoneutral and the bond broken is a C-H bond. Thus the corresponding diagram will be Fig 2. The crucial factors are the shapes and positions of the two repulsion curves but if they are not very different in the two cases the activation energy of the former process will almost certainly be lower than t h a t of the latter and therefore less than approx. 13 kcal. If the decomposition of this peroxide followed a chain type of reaction the chains would be very short if 8 % of the products resulted from the termination process so that the methyl ethyl ether may be at least partly formed by the reaction proposed here.R + R’OOR” -+ R’OR + R”0 or R”0R + R’O may occur in diaIkyl peroxide decompositions where R results from R’O -f R + CH,O or R”Q 3 R + CH,O. GENERAL DISCUSSION 313 It may be noted that nitrates or nitrites might undergo a similar reaction even more readily since the 0-N bond has a dissociation energy of the order 38 kcal. and the reaction will be even more exothermic than the corresponding peroxide reaction. Dr. L. Bateman and Mrs. Hilda Hughes (Welwyn Garden City) (communicated) The decomposition of cyclohexene hydroperoxide in solution in the temperature range 60-100’ C exhibits rather different features to those reported by Bell ei! al. and also by other workers for other hydroperoxides.718 Our investigations in this field are still pro- ceeding but certain results appear to warrant consideration here.In benzene solution the reaction is almost exactly second order. The non-formation of diphenyl and phenol indicates that the solvent is inert under our conditions. Cyclohexenone and R0,-double bond addition compounds appear to be the main organic products. Each mole of peroxide decomposing liberates nearly 0.5 mole of water and much less oxygen (- 0.1 mole). The Arrhenius activation energy is 26 kcal./mole. We conclude that the measured rate refers to the one-step bimolecular decomposition RO2-H + HO-OR + RO2- + H2O + RO-. - (1) Radical Radical t C C H5 H5 + H - - - - Radical I \ C H~ + CN 00ccc>15 /C H . . 3 Inactive products ’C H i FIG.2. In cyclohexene solution the decomposition proceeds faster (roughly five-fold) and the order is reduced to 1-7-1.8. These characteristics the isolation of cyclohexenol as the major product,O and a linear relationship between [R02H] /rate and I /[RO,H] are consistent with a chain reaction comprising (I) as initiation step and RO- + R’H + ROH + R’- R’- + ROJ3 + R’OH + RO- zRO-1 2R’- i ’ . - (4 * (3) * * (4) (5) (6) wheze in this case R and R’ are identical and the reasonable assumption 10 is made that kz = A&,. In the presence of stearic acid the reaction is greatly accelerated and becomeF of the first-order with respect to the peroxide. These facts correlate strikingly with the catalytic action of the hydro- peroxide on the oxidation of the parent olefin with molecular oxygen R’- + RO- Farkas and Passaglia J .Amer. Chem. SOC. 1950 72 3333. Kharasch Fono and Nudenberg. J . Org. Chem. 1951 16 113. Farmer and Sundralingam J . Chem. SOC. 1942 121. lo Cf. Bateman Gee Morris and Watson this Discussion. GENERAL DISCUSSION * roxidation = const. d([RO2HI2) 3 I4 Normally the relationship is obeyed ; in the presence of stearic acid this changes to (7) . (8) . Reaction (3) has not been considered by Bell et al. and an indication of its existence in their work viz. the formation of methanol4 from CMe,O,D and Me radicals has been otherwise explained. However we believe the following experiment provides strong evidence for its occur- rence. Cyclohexene hydroperoxide was decomposed in the substituted I 4-diene ethyl linoleate (R’H).R’ is thus a mesomeric pentadienyl radical which if formed will react preferentially to give a conjugated I 3-diene derivative. In fact conjugated diene units were produced they were located only in an ethyl hydroxy-linoleate and the formation of this substance was greatly in excess of that expected to result either directly or indirectly via radical-radical interaction. The alternative reaction . H j ( 7 . H !n R’- R’- + ROZH -+ RO2- + R’H has possibly been shown to occur by Bell et al. at 195’ in the gas phase reaction of Me and Et radicals with CMe,O,D. It should be noted how- ever that even in these systems the reaction is roughly 10 kcal. endo- thermic and this figure is increased to about 30 kcal.when R’H is an olefin and allylic resonance energy has to be supplied. At lower temper- atures and especially in unsaturated solvents (8) must be quite unim- portant compared with ( 3 ) which is about 30 kcal. exothermic (if Do-o = 40 kcal.). Reaction (3) is envisaged as a radical substitution reaction (SRz-to extend the Hughes-Ingold terminology) formally analogous to (I) cf. (-y with RO,-I-H + O-:-OR + 0-\--OR ui t-)! where the arrows denote one-electron displacements. The corresponding ionic processes involving heterolytic bond scission can obviously be re- garded likewise. We do not suggest that the decomposition mechanism now advanced is necessarily applicable to other hydroperoxides under different con- ditions-it is quite evident that analogy can be especially misleading in this field.Thus the effect of increasing temperature may be particularly critical in facilitating a unimolecular S,I relative to an S,Z decomposi- t ‘on because the increasing tendency of the comparatively weak 0-0 bond to rupture will be roughly paralleled by a decrease in the molecular association in the hydroperoxide (evidenced by hydrogen bonding) which is undoubtedly precursory to reaction (I). Nevertheless we do suggest that the relevant conclusions of Bell et al. require careful reconsideration in particular the postulation of unimolecular dissociation of tert.-butyl hydroperoxide in solution at temperatures not greatly above 100’ C. Not only are the rate data at 150’ C given in Table I1 of these authors’ paper consistent with our scheme (they correspond to a reaction order of 1-55) but Bolland and Morris 11 have made the very significant observation that this hydroperoxide in common with numerous primary secondary and tertiary allylic hydroperoxides catalyzes olefin oxidations in accordance with eqn.(7). Dr. W. E. Vaughan (EmeryvilZe California) (communicaded) The questions raised by Dr. Bateman and Mrs. Hughes concerning our paper seem to u s to be well answered by the experimental evidence presented namely the isolation of methane-d and ethane-d and of ethyl tert.-butyl l1 See Quart. Rev. 1949 3 I. GENERAL DISCUSSION 31 5 peroxide from the appropriate reactions. These actual product isolations in substantial yields provide definite proof for the reactions R' + (CH,),COOD -+ RD + (CH,),COO* R.+ (CH,),COO' -f (CH,),COOR . and and they are not otherwise explained by the mechanism proposed by Dr. Bateman and Mrs. Hughes. R* + - CHdH-CH2-CHSH- . It is our opinion that reaction (I) is not endothermic as claimed by Bateman and Hughes but rather is exothermic to the extent of about 20 kcal./mole when R is methyl. No quantitative evidence is available but qualj tative comparison of ted-butyl hydroperoxide as a hydrogen atom donor with cyclohexene and formaldehyde leads us to the conclusion that Do-,(ROOH) is of the order of 80 kcal./mole rather than 105 or 118 kcal./mole as observed in certain alcohols and in water. Further proof for the chain decomposition we have proposed may be obtained from a comparison of the rates of decomposition a t 140~ of tert.- butyl hydroperoxide in a reactive solvent n-octane and in an inert medium chlorobenzene.In the latter solvent reaction (3) cannot intervene R'* + ROOH -f R'OH + RO' because decomposition leads to quantitative yields of tert.-butyl alcohol and oxygen. Moreover in this solvent the rate is some 40 times that in n-octane. (A) . Bateman and Hughes have acknowledged the difficulties in drawing analogies between decompositions of various peroxides under differing conditions. However the decomposition of cyclohexene hydroperoxide in an unsaturated solvent can also be interpreted by our mechanism which has the evidence of actual isolation of a stable peroxide intermediate.For ethyl linoleate this intermediate would correspond to (B) ; -+ RH + -CH=CHL:CHLCHCCH- OOR ROO' + (A) -+ -CHdH-CHdH-CH- I (B) 0. I (B) -+ KO' + -CH=CH-CHeCH-CH- - (1) ' (4 * (3) OH (4) (5) (6) (7) (C) (C) + ROOH (or RH) -+ ROO' (or R.) + -CH=CH-CH=CH-CH- The allylic stabilization of (A) would increase the possibility for the postulated association reaction (2) at the expense of metathetical reactions such as (I) or ( 3 ) . I The decomposition of cyclohexene hydroperoxide in cyclohexene can also be interpreted by a strictly analogous mechanism. However we think that in an inert solvent the reactions of the hydroperoxide group are greatly complicated by the presence of reactive methylene groups and the double bond in cyclohexene hydroperoxide.Although our in- formation offers nothing either pro or con concerning reaction (3) we see no objection to it on purely theoretical grounds and have in fact indicated that the processes proposed in our paper do not exclude other possible competitive transformations which may be proceeding simul- taneously and undetected. Dr. W. A. Waters (Oxford) said Whilst we all realize the great value of the kinetic studies of olefin autoxidation carried out by members of the British Rubber Producers' Research Laboratory I would suggest that there will be little value in increasing the precision and detail of GENERAL DISCUSSION their present types of kinetic work beyond the stage which it has now reached.Kinetic studies of chain reactions give velocity equations which are dependent very much on one particular chain-ending process e.g. 2R' 3 products or R' + ROa' +- products (dimers often unspecified). When however organic chemists start looking for these diagnostic compounds in autoxidized mixtures they seldom find them and sometimes can show that the postulated dimers are far too reactive to persist in the final oxidation product. For example in our current study of the effects of phenols on benzaldehyde autoxida- tion 12 Mr. C. Wickham-Jones and I deduced kinetically that the chain- termination product in mixtures containing +-cresol was a dimer of a mesomeric phenoxy radical (ArO*)2. We have failed as yet to isolate any of the known dimers of the tolyloxy radical from our reaction product and have found that the material which is eventually formed must be more complex.Moreover by adding the known dimers separately to our reacting system we have shown that they have a comparable reactivity to the p-cresol and so could not persist throughout the whole autoxidation. With 2 6-xylen-1-01 which gives similar kinetics the obtainable reaction product is not the phenol dimer but its further oxidation product. the corresponding diphenoquinone ; even this is an autoxidation retarder which therefore must be converted to still another product. One important point that kinetic investigators should remember is that the chain-terminating reaction which will be revealed by their reaction velocity measurements is the one which most rapidly removes active free radicals from the system.None of a whole series of slower radical-removing processes or of secondary reactions involving products which initially are present in very low concentration may be able to influence the reaction velocity to a discernible extent unless the experi- mental conditions are especially designed so as to detect one of them. " Initial velocity I' measurements in particular are of little help in the diagnosis of the chemical identities of products of chain termination. In reactions involving hydroperoxides it is important to remember that if the system becomes appreciably acidic or basic then an ionic decomposition of the hydroperoxide may set in to give products which are often quite different to that of its homolytic fission.The homolytic fission seems to proceed + 'OH RlR2R3G-O-OH -+ RlR,R3C-O' L RlR2C==0 + 'R3 where the fission rules for the breakdown of the ROO radical are in general those given by Walsh.lS The acid-base catalyzed reaction however proceeds RlR2R3C-0-0-H ways. + H+ -* where R is the group that most easily carries with it an electron pair. These different products influence the subsequent reactions in different la J . Chem. SOC. 1951 812. l3 Trans. Faraday SOC. 1946 42 269. GENERAL DISCUSSION 317 Dr. A. J. Harding (Cambridge) said A good approximation to the kinetics of hydrocarbon oxidation may be derived without postulating intermediates of definite types such as peroxide or aldehyde.It is only necessary to assume that the chains consist of links which are alternately reacting with oxygen and with the hydrocarbon that the chains are terminated by destruction of the radicals which would otherwise react with hydrocarbon and that the chains are initiated by the breakdown of an intermediate which they themselves produce. The use of special features such as the ratio pmax/Apmax becomes essential if the finer points of the kinetics are to be evaluated. One of these points is the type of degenerate branching. Dr. Mulcahy has described the type in which the reaction fails to reach an infinite rate (i.e. ignition) because the re- actants are consumed before the intermediate concentration becomes sufficiently great.Another type of degenerate branching has been suggested for the oxidation of hydrocarbons at higher temperatures. In this second type the intermediate (e.g. aldehyde or peroxide) which generates chains is itself oxidized by the radicals i t produces. Since the production of intermediate is proportional to the first power of its concentration and the destruction is proportional to the second power a limiting concentration of intermediate is reached. This means that even if no reactants are consumed in the process the reaction attains a limiting finite velocity. A distinction between the two types of degenerate branching is readily made. In the first case addition of intermediate at the beginning of the reaction will give an increase of maximum rate as well as a decrease of induction period.In the second addition of intermediate will reduce the induction period but will produce no change in the reaction velocity until so much intermediate has been added that the induction period has been completely eliminated. The oxidation of ethylene at temperatures in the region of 400" C with formaldehyde as intermediate shows the behaviour expected of the second t41~e.l~ A derivation of the variation of pmax/Apmax with hydrocarbon con- centration for the second type of degenerate branching leads to a relation- ship approximately the same as that found by Dr. Mulcahy for the first type. This relationship does not therefore provide a clue to the type of degenerate branching although i t does give valuable support to the thesis that degenerate branching is taking place.l4 Dr. Peter Gray (Cambridge) (communicated) May the fact that the propylene oxidation studied by Mulcahy appears to occur through a peroxide which may react in such a way as not to produce active radicals be associated with the characteristics of propylene as a chain-breaking agent ? (In this case the constant C might include a term corresponding to the propylene concentration.) Dr. M. F. R. Mulcahy (Melbourne) (communicated) Dr. Gray's sug- gestion would require the chain-terminating effect of propylene to be specific towards the radicals produced immediately on decomposition of the peroxide. The participation of the hydrocarbon in a chain-terminating reaction in the main cycle would affect the second rather than the third term in eqn.(4) (tending to make B independent of [RH] under conditions of excess oxygen). However i t now appears that the significance of the value of C derived from the experimental results is somewhat complex. Some very recent experiments carried out in this laboratory by Mr. Ridge with propylene at 288" C have shown that the intercept of the pmax/Apmax against [RH] curve on the pmax/Apmax axis is dependent on the surface conditions and may become positive. It seems therefore that the simple interpretation of this intercept (C) given in my paper is in need of some amplifications. 1 4 Harding Thesis (Cambridge 1948). Norrish X V I Int. Coll. C.N.R.S. (Paris 1948) p. 16. Harding and Norrish (in preparation).GENERAL DISCUSSION Dr. E. J. Harris (London) (communicated) In 1935 l5 i t was suggested that compounds known to promote the branched chain reaction in the slow oxidation of hydrocarbons such as ethyl nitrite and ethyl peroxide did so by providing -OR radicals and the converse process a condensa- tion of -OR to peroxide seemed not unlikely. In 1938 Neumann and Tutakin l6 showed that small additions of diethyl peroxide to butane + oxygen mixtures would cause the appearance of a flame similar to the cool flame seen during the oxidation reaction. when induced in hydrocarbon + oxygen mixtures. The result flame does not inevitably promote hydrocarbon oxidation even quent reaction. This is explicable on the basis of radical formation. 318 To put the hypothesis of the intervention of peroxide on a frrmer basis two approaches are possible namely the physical and chemical investiga- tion of the slow oxidation and the study of the peroxides themselves under appropriate conditions.In 1936 Sir Alfred Egerton suggested the latter method as a profitable one. Before mentioning a few relevant properties of peroxides i t will be useful to state the analytical evidence concerning peroxides isolated from oxidation reactions. At 320-270' propane and butane can be made to yleld considerable quantities of hydrogen peroxide making up 10 yo of the total condensate. The H,O combines with two molecules of form- aldehyde one of the other main products and in one experiment l7 0.5 g. of the compound was isolated.If however the reaction is carried out when the walls of the vessel are slightly contaminated with alkaline material the reaction no longer leads to the formation of the peroxide though all the other products are nearly the same. This and other evidence suggested that the H,O,-formaldehyde compound was only formed after condensation. The fact that H,OB can be isolated indicates that radicals -OH or -0,H intervene in the slow oxidation the sensi- tivity to the state of the wall may mean that hydrogen peroxide is only formed when collisions involving the wall permit it. The analytical problem of showing the presence of an alkyl peroxide is of course com- plicated by the presence of a large amount of H202. Results of the study of the alkyl peroxides 18 19 2o and their influence on slow oxidation reactions are consistent with the following (a) They may be formed in traces as by-products but chemical evidence is lacking.(b) The alkyl hydrogen peroxides like H,O, are very sensitive to the state of the surface being decomposed by traces of alkali. They will ignite giving a diffuse blue luminescence when introduced with oxygen into a vessel at 250-3ooo. The luminosity resembles that sometimes seen in the hydrocarbon oxidation and is general rather than flame-like. The dialkyl peroxides will decompose with a blue flame in certain conditions of temperature pressure and gas mixture. Below the critical limits they decompose homogeneously. The artificial blue depends upon the surface and the temperature.( d ) Traces of the peroxides reduce the induction period of the slow oxidation of hydrocarbons without altering the rate of the subse- The small quantities of peroxide which are needed to produce visible effects make i t not surprising that they have not been isolated. Only 0.007 cm. of dipropyl peroxide will give a flash with air at 270° and 0-01 cm. 16 Egerton Smith and Ubbelohde PhiE. Trans. Roy. SOC. 1935 243 433. 16 Neumann and Tutakin Acta physicochim. 1938 9 861. l7 Harris and Egerton Chem. Rev. 1937 21 287. 1s Harris and Egerton Proc. Roy. SOC. A 1938 168 I. l9 Harris Proc. Roy. SOC. A 1939 173 126. 20 H a r k Proc. Roy. SOC. A 1940 175 254. GENERAL DISCUSSION 3 I 9 of ethyl hydrogen peroxide can be seen to luminesce when oxygen is present.In spite of the explosive properties of the peroxides the alkyl hydrogen peroxides like H,O,,21 can survive quite high temperatures if the vessel has a low surface/volume ratio and particularly if a carrier gas is present. Thus at 180' only 5 yo ethyl hydrogen peroxide was decomposed when passed with N through a tube at 180' (contact time 4 sec.). A t 320° z yo survived. Under somewhat similar conditions ( I sec. contact time) 85 yo H,O sometimes survived. Post-war work on hydrocarbon oxidation (e.g. Hinshelwood 22 Mulcahy 23) appears to be in accord with the scheme outlined by Egerton and Harris 24 in 1938 in which i t was proposed that the initial step was peroxide forma- tion followed by splitting to give radicals -OR -OH and -O,H which carry on the main reaction.Analogous schemes were outlined by Ubbel~hde,,~ Pease 28 and Lewis and v Elbe.,' Egerton and Harris however did expressly suggest (cf. their Table 11) the reaction z -OH -f H,O as a chain-terminating one and this together with a reaction of -0,H with a hydrogen compound would explain hydrogen peroxide formation. Organic peroxides as such were only invoked in the initiation of the reaction and this agrees with the fact that surfaces active in decomposing the alkyl hydrogen peroxides (e.g. salt-coated ones which under the experimental conditions become alkaline) are those which give rise to long induction periods for the slow oxidation reaction. Dr. G. J. Minkoff (ImperiaE College) said I should like to raise two points in connection with Dr.Mulcahy's paper. First with regard to the evidence of Badin which is quoted in support of the formation of hydroperoxides in oxidation Badin observed a line at 11.4 p in the infra- red spectrum of tert.-BuOOH; he concluded that since H,O also has a line at 11*4p this line is characteristic of the -0OH grouping. This is unsound for several reasons; in many observations of the infra-red spectrum of freshly prepared ten!.-BuOOH I have only once found a line at 11-4p and that was in an impure sample ; the position of the line was otherwise found to be at 11.6~. Di-tert.-butyl peroxide does have a line at 11*4p so that clearly the line cannot be characteristic of -0OH ; in fact as was pointed out before i t may be connected with both the -04- group and with the tevt.-butyl group.The line which Badin found may well have belonged to H,O,. The other point is connected with the attempt t o draw up a mathe- matical expression for the rate of pressure change. This is based on the proportionality observed between peroxide concentration and the rate of pressure change. In the derivation in terms of initial concentrations the tacit assumption is made that P is the hydroperoxide of the original hydrocarbon. However this assumption may not be correct because in the degradation of a long cha'n paraffin several steps must occur all prob- ably involving peroxides. In the calculations made by Bolland this ob- jection does not arise since only one peroxide is involved. Another complicating factor is that the different peroxides formed (i.e.primary secondary peracids etc.) will react to different extents with the potassium iodide reagent; thus the maximum peroxide concentration may be masked by the lack of reaction with KI of some of the peroxide present. Dr. W. A. Waters (Oxford) (communicated) Sir Cyril Hinshelwood's review of the influence of substituents on the rate of hydrocarbon oxida- tion can be given an alternative interpretation which is equally consistent ar Harris Trans. Faraday Soc. 1.948 9 764. 2 Hinshelwood Faraday SOC. Dascussaons 1947. z8 Mulcahy Trans. Furuduy Soc. 1949. 45 575. ,4 Egerton and Harris Proc. I8me Congr. Chim. Ind. (Nancy 1938). 2s Ubbelohde and Egerton Proc. Boy. SOC. A 1935 152 354. 26Pease J .Amer. Chem. SOL 1929 51 1839 et seq. 27 Lewis and v. Elbe J . Amer. Chem. SOC. 1937 59 976. 320 GENERAL DISCUSSION with the theories of general polarity. The substituents which he finds to increase oxidation rates are also those which promote the attack of methyl radicals on C-H groups. Kharasch and his colleagues for in- stance have reported that methyl radicals from the decomposition of diacetyl peroxide preferentially attack C-H bonds vicinal to C-C1 and also CH-CO- CH-0- CH-CO-OMe but do not attack CH groups. From studies of tert.-butyl peroxide it now appears that R-0' radicals (e.g. Me,C-O.) have not quite the reactivity of alkyl radicals such as methyl and are even more selective in the same sense in their reactivities towards C-H groups.28 I would therefore suggest that the reaction which immediately follows Sir Cyril's chain initiation process R-0-0-X + R-0' + '0-X i.e.R-0' + H-C- -+ R-0-H / \ (iii) R-M + M 3 R-M-M' groups to build up high polymers. + 'G- / \ may be much slower and much more selective than the subsequent stages of the gas-phase oxidation involving R' and R-0-0. radicals. It may well be rate-determining for the whole breakdown of the oxidized compound. A similar state of affairs is well known in polymerization chemistry where the sequence (ii) R' + Monomer -+ Hydrocarbon type radical R-M' ; (i) Catalyst (eg. benzoyl peroxide) -+ Radicals R' ; etc. (fast chain growth) results in a polymerization kinetically dependent upon the rates of both processes (i) and (ii) as for instance in the benzoyl peroxide catalyzed polymerizations of both styrene and vinyl acetate.In polymer chemistry again the stability of CH in comparison with A H z - is strongly marked as for instance in the ready chain transfer to -CHz-CH=CH- and the converse tendency of compounds with CH,-C=C In connection with later states of hydrocarbon oxidation i t may be noted that in the Dyson Perrins Laboratory we have recently shown that in the liquid phase the breakdown (1) Alkyl-CO' + AllrJtl' + CO occurs quite readily a t temperatures as low as SO-IOO~ C. This is con- firmatory evidence for the view that in hydrocarbon oxidation at elevated temperatures only the initial stages of the reaction sequence are rate- determining.Dr. A. D. Walsh (Leeds) said It is helpful in considering the kinetics of oxidation reactions of hydrocarbons in the gas phase to write the step that produces the alkyl peroxide radical in the form R + 0 = RO,*. Commonly the * is omitted but its inclusion serves to remind us that in the first instance the RO radical is energy-rich. This is important for at least the following reasons. (i) It makes it less likely that the reverse reaction to (I) will be neglected in considering the full kinetic scheme. (ii) The excess energy is presumably distributed over various vibrational degrees of freedom. The radical is therefore to be thought of as vigorously twisting turning and generally distorting itself far more than does a " normal " radical.This means that subsequent reactions which axe rather more complicated than are usually found in gas-phase oxidation processes are rendered more pIausible than they would otherwise be. Cf. also Farmer and Moore J . Ckem. Soc. 1951 131. 321 GENERAL DISCUSSION Those gas phase chain reaction steps which are best substantiated commonly' fall into one of 4 classes 29 ( a ) those in which one linkage is broken e.g. CH,CH3 = CH + CH CH + NO = CH,NO* . - (2) ( b ) those in which one linkage is formed e.g. (I) above the reverse of (2) and * (3) (G) those in which one linkage is formed and one is broken e.g. abstraction of an atom (particularly H) from a molecule by a radical as in * OH + RH = H,O + R; If this is re-written as . RCH,CHO + 0 = RCHO + HCHO CH + 0 = HCHO + OH.CH3 + 0 = [C,HO,"] = HCHO + OH. (4) ( d ) energy transfer reactions e.g. reactions involved in the quenching of fluorescence A * + B = A + B * . As illustrations of these four types it is instructive to look at the paper by Bell Raley Rust Seubold and Vaughan. The great merit of that paper is that some experimental evidence for each step postulated has been obtained. Consideration of these steps shows that they all con- form (or in one case can readily be slightly altered to conform) to the above types. - ( 5 ) In other words the steps occurring most commonly appear to be the simplest. This should not be surprising for in postulating a reaction step i t is vital to consider just how the reactants could be converted into the supposed products; and such consideration makes it difficult not to conclude that more complicated steps than (a) t o (a) are likely to have unfavourable steric factors or activation energies.30 If a postulated gas- phase reaction step does not conform to one of the 4 types one cannot say it is impossible-but one can and should demand that the evidence for it be unusually strong unless it is to be dismissed as mere speculation.This is incidentally a serious criticism of many of the reaction steps postulated by certain authors. To take but one example the supposed reaction involves the simultaneous breaking of two linkages and the forming of two linkages. Unless particularly strong evidence in support of this was forthcoming-and such has certainly not yet been produced-this postulated step can be taken as very unplausible.In the special case of a reaction such as (I) however i t is not unplaus- ible to suppose a subsequent reaction for RO,* more complicated than types (a) to (d). An example is the reaction it is split into simpler steps and the fact that the second involves the breaking of two linkages and the forming of one does not appear un- plausible. Another example might be a somewhat complicated isomer- ization of RO,*. (iii) Even if the reaction supposed subsequent to (I) conforms to (a) to (d) inserting the * in (I) helps one to remember that the activation energy' for the following reaction need not be as great as would be the case for " normal " RO,.Some or all of the activation energy may be provided by the energy produced by (I). The step RO + RH = ROOH + R - (6) 29 Cf. Chamberlain and Walsh Rev. Inst. Frangais du Pe'trole 1949 4 307. 3 0 Cf. Ubbelohde Rev. Inst. Frangais du Pe'trole 1949 4 315. 31 Norrish this Discussion. L GENERAL DISCUSSION HO,* + NO = EIO + NO -/- 0 322 although probably exothermic when RH is an olefin is probably endo- thermic when RH is a paraffin.32 The activation energy required for (6) in gas phase reactions however need not be as great as this endothermicity. The idea is closely similar to that of (ii). The energy of formation of RO, carried over into a subsequent reaction reduces any improbability of that subsequent step. To realize this is to recognize that the process of gas-phase peroxide formation may’ be subject to inhibition by inert gases.The greater the dilution of the reactants the greater the chance that collisions (and reactions of type (a)) will take so much energy from RO,* that the probability of (6) is seriously reduced. Peroxide formation is involved in cool flame phenomena. To test whether inhibition by inert gases of cool-flame formation can occur however one needs to choose conditions carefully’. At temperatures near the lower temperature limit of spontaneous cool-flame formation the dominant chain-ending reaction is usually a surface destruction of radicals and the inhibition may be swamped by a greater promotion viz. the hindering of radical diffusion to the walls. Along the upper temperature limit of cool flame formation however the dominant chain-ending reaction is a gas phase one (in most cases the decomposition of a bulky alkyl radical to an olefin and a small alkyl radical 33 3ii).At such temperatures for diethyl ether (Fig. 3 of ref. (33)) and for propane (Fig. 11 of ref. (35)) a small inhibition by’ inert gases is observed. It seems probable that this inhibition represents the expected effect of reduction of the energy content of RO,* though of course i t has to be remembered that cool-flame formation3(j and pro- pagationa7 involve a thermal factor and inert gases may also inhibit purely by virtue of their thermal properties. Finally i t is important to be clear about the nature of the excess energy carried by RO,*.This is surely vibrational in nature. For the analogous HO,* radical however the following reaction has been supposed to occur 38 the HO,* radical transferring the whole of its energy to the NO,. Such a complete transfer of energy seems only likely for electronic energy. Yet the effects of inert gases (M) on the second limit of the €-I + 0 ex- plosion are usually supposed to require the transfer of energy from HO,* (i.e. H + 0,) to M ; and the different gases that are effective make i t difficult to suppose other than vibrational energy is concerned. Some confusion seems therefore to exist. Dr. C. A. McDowell (Liverpool) (communicated) I was interested to read Dr. Walsh’s remarks on the existence of RO,* radicals i.e. RO radicals endowed with excess energy.Similar views were expressed by Dr. Thomas and myself in our paper on the inhibitory effect of nitrogen peroxide on the gas-phase oxidation of acetaldehyde. 39 In considering possible subsequent reactions of RO,* radicals one has to be extremely careful for the excess vibrational energy with which these radicals are endowed may or may not persist throughout the numerous collisions which these molecules may make before they undergo reaction. If the excess of vibrational energy does not persist then ob- viously little is to be gained by ma’ntaining the above notation. Numerous 32 Walsh J . Chem. SOC. 1948 339. 33 Chamberlain and Walsh 3rd S y m p . Combustion Flame and Explosion Phe- nomena (Williams and Wilkins Baltimore 1949). 34 Walsh in course of publication.35 Malherbe and Walsh Trans. Favaday SOC. 1950 46 835. 36 Malherbe and Walsh Trans. Faraday SOC. 1950 46 824. 37 Spence and Townend 3rd S y m p . Combustion Flame and Explosion Phe- nomena (Williams and Wilkins Baltimore 1949). 38 Dainton and Norrish Proc. Roy. Soc. A 1941. 177 395. 3% McDowell and Thomas J . Chem. Soc. 1950 1462. GENERAL DISCUSSION 323 cases are however known where vibrational energy does persist through- out a time interval during which a molecule may make as many as 5 x 104 collisions before one quantum of vibrational energy is dissipated. Theoretically there is no reason why vibrational energy should not be transferred from one molecule to another provided their vibrational levels are sufficiently close together to permit resonance to occur.Such cases are well known and it should perhaps be pointed out that it is not necessary that all the excess vibrational energy should be transferred from one molecule to another ; in fact it seems probable from elementary theoretical considerations that the excess energy is more likely to be transferred in small amounts in successive quanta. One other type of transfer which must be borne in mind is the transfer of vibrational energy from one molecule into translational energy of another. This is most easily under- stood in the case of the transfer of vibrational energy from vibrationally excited molecules to rare gas atoms. This latter type of transfer probably also occurs with light molecules. RCH,CHO + O=RCHO + H,CO was originally postulated as a source Prof.R. G. W. Norrish. (Cambridge) (communicated) The reaction of excited formaldehyde to account for the luminosity of the cool flame. In view of the very low luminosity observed (ca. I quantum per 10 mole- cules of hydrocarbon reacting) its probability in comparison with the other radical reactions would be very low in accordance with Dr. Walsh’s expressed view. However I have as will be observed in my paper dissociated myself from this reaction except as a possible component in the dark blue flame and even then the suggestion must not be taken too literally for the reaction may occur in two stages RCH,CHO + 0 = RCH CHO + OH RCH . CHO + OH = RCHO + H,CO. In any case I am not fully able to agree with limitations which Dr.Walsh would lay down for chain processes. Not enough is yet known about the possible configurations in the transition state. I agree with his remarks about RO,*. They conform to the views we have already expressed with reference to HO ; see for example the footnote in my paper or Axford and Norrish on the oxidation of formal- deh~de.4~ Prof. A. R. Ubbelohde and Mr. Small (Belfast) (communicated) The suggestion that RO,* is in the first instance “ energy-rich ” and is more effective in reaction ( 2 ) above if i t can use this energy before i t is randomized by collisions etc. has an interesting experimental corollary. It has recently been observed 41 that molecular hydrogen inhibits certain reaction chains in both aldehyde and hydrocarbon oxidations much more markedly than molecular nitrogen.Tentatively this may be attributed to the efficiency with which hydrogen can effect the transfer between internal and translational molecular energy in collisions. RO,* + H + H,* + RO,. In (pseudo) unimolecular reactions the special efficiency of hydrogen collisions in the reverse process translational + internal is well known and from the principle of microscopic reversibility collision process (3) would also be expected. Dr. A. D. Walsh (Leeds) (communicated) Prof. Sir Cyril Hinshelwood refers to my suggestion that bonds between strongly electronegative elements should be strengthened by electron-repelling groups. As originally put forward this was based upon the possibility of increas- ing the overlap of the two atomic orbitals concerned in a single bond 40 Axford and Norrish Proc.Roy. SOC. A 1948 192 518. 41 Small and Ubbelohde J . Chem. Soc. 1950 723 and unpublished results. L* 324 GENERAL DISCUSSION between the elements. It is perhaps worth pointing out however that the suggestion could also be based upon the possibility of reducing lone pair- lone pair repulsion between the 0 atoms by attaching electron-repelling groups. The interaction between lone pair electrons on a halogen atom and electrons of a neighbouring system is greatest for the most electro- negative halogen (F) and least for I.43 A similar statement applies to interaction involving lone pair electrons on a Group VI eIeme11t.4~ The interaction between lone pairs on the adjacent 0 atoms of an -0-O- group is expected to be repulsive (witness e.g.the skew nature of the H,O molecule). Attaching electron-repelling groups to an 0-0 group will reduce the effective electronegativity’ of each 0 atom and so be expected to reduce the lone pair-lone pair repulsion i.e. be expected to increase the 0-0 bond strength. Dr. N. Uri (Manchester) said In the primary step relating to the oxidation of hydrocarbons as postulated by’ Prof. Norrish viz. RH + 0 -+ R + HO the endothermicity and activation energy are likely to be of the order of 40-70 kcal. if the dissociation energy of the HO radical into H + O Z ( D ~ o 2 ) is as low as 36 kcal. This latter value is obtained 4 4 9 45 from the electron affinity (in solution) of the HO radical which is in turn evaluated by inter- polation from the energies corresponding to the absorption maxima of various ferric ion pairs.While some unrecognized error may cause our quoted value of D H O ~ to be a few kcal. low it would be difficult to account from hydrocarbon chemistry for a value of D H O ~ as high as 65 k~a1.,4~ unless i t also assumed that the activation energy of the first step of the oxidation of hydrocarbons is much larger than its endothermicity. Dr. A. J. B. Robertson (King’s College London) said The heat evolved in the formation of HO may be determined by the method outlined by’ Stevenson in this Discussion. I find that hydrogen peroxide is dissoci- ated by electron impact to give the HO,+ ion with a small probability.The appearance potential is 16.1 eV. The ionization potential of H,O is found to be 12.1 elr. The ionization of O, HO and H,O very prob- ably involves the removal of a weakly anti-bonding or non-bonding electron located approximately on an oxygen atom. In these circum- stances we may provisionally estimate the ionization potential of the HO free radical as 12.2 eV intermediate between that of 0 and H,O, and i t may be less. This assumption gives 46 kcal. for the heat evolved in the formation of HO from H and O, subject to an experimental uncer- tainty a t present of about g kcal. If the ionization potential of HO is lower than 12-2 eV the formation of HO is even less exothermic. Dr. R. Spence (Harwell) said Prof. Norrish has raised many inter- esting points.I have always thought that the phenomena of com- bustion are so numerous and varied that general mechanisms should not be pressed too far unless all the well-established facts have been taken into account. For instance I do not think that the isolation of aldehydes as products of the reaction is necessarily evidence that they occur as essential intermediates. This was one of the important lessons to be learned from the early work of Bone who tried to establish the hydroxylation theory of combustion by the identification and separation of a set of products which were regarded as intermediates. Formalde- hyde for instance is a product of the oxidation of acetylene at 3 o 0 - 3 ~ 0 ~ C.4’ It is known that its rate of reaction with oxygen at these temperatures la See Baker and Hopkins J .Chem. SOC. 1949 1089. 43 Unpublished work of Baker and Barrett. 44 Evans and Uri Trans. Faraday SOC. 1949 45 224. 45 Evans Hush and Uri (in the course of publication). 46 Walsh J . Chem. SOC. 1948 331. 47 Spence and Kjstiakowsky J . Amer. Chem. SOC. 1930 52 4846. 48 Spence J. Chem. SOC. 1936 652. GENERAL DISCUSSION 32 5 is quite slow even at moderate partial pressures. Thus ordinary formaldehyde cannot be an intermediate in the combustion of acetylene ; i t can only be an intermediate if i t reacts whilst still in the excited state and i t is important as Dr. Walsh has mentioned earlier to recognize this in the mechanism. Then as regards the OH radical I have referred earlier in this dis- cussion to the work of Dr.Bates and myself 49 on the reaction between methyl radicals and oxygen and to the possibility of the formation of OH radicals or of CH,O in the primary step. More recently Blaedel Ogg and Leighton 5 O failed to detect the OH radical in this reaction by optical methods so that if present its concentration must be very small. Another piece of evidence which is not easy to fit into a simple radical chain theory was obtained by my wife 51 when working in Prof. Townend's laboratory. The propagation of a cool flame through a vertical tube at room temperature containing acetaldehyde ether hexane or heptane and oxygen or air is determined purely by thermal considerations. Free radical chain effects usually associated with the vessel wall and with the addition of inert gases were not observed.The addition of methyl radicals or of iodine produced no measurable effect ; the only indication that a chain mechanism might be operating was the inhibitory effect of additions of NO,. Another experimental result which is not easy to reconcile with a mechanism involving oxygen atoms is the homogeneous slow combustion of formaldehyde in reaction vessels of only I mm. diam.48 In this case the chain carriers must be relatively unaffected by the wall. The surface oxidation only predominates when the vessel is packed with powdered glass. The " dark blue " flame mentioned by Prof. Norrish is to be identified with the " blue" flame of Townend and co-workers 51 which follows in the path of the " cool " flame.Prof. R. G. W. Norrish (Cambridge) (communicated) There can be no question that aldehydes are formed as intermediates in hydrocarbon oxidation because they are readily observed the question a t issue is whether they are to be identified as the moderately stable intermediate Iesponsible for the delayed branching. I have given the reasons in my paper for supposing that they are particularly in the second paragraph I do not think that more weighty evidence has yet been produced for any other intermediate product playing this role. It must be remembered that the process of degenerate branching also takes place at temperatures where peroxides are completely unstable and unobservable and that Semenov's hypothesis demands a finite measureable concentration.The possible origin of aldehydes via a transitional peroxide is not ruled out on this account e.g. RCH + 0 -+ [RCH,O] + RCHO + OH as I have indicated in my paper. I do not feel that the remarks about the oxidation of acetylene are relevant. My paper did not concern this reaction but i t may be men- tioned that the rate of oxidation of formaldehyde at 286" C is appreciable (for partial pressures of formaldehyde and oxygen of IOO mm. in a vessel of 23-6 mm. diam. and volume 80 cm. the rate of reaction as measured by pressure change is 1-9 mm./rnir~.).~ Further i t is highly probable that the formaldehyde product in the acetylene oxidation is excited and thus more highly reactive. However I do not wish to commit myself to any view concerning acetylene oxidation in the present remarks.49 Bates and Spence J . Amer. Chem. SOC. 1932 53 1689. 50 Blaedel Ogg and Leighton J . Amer. Chem. SOC. 1942 64 2499. 51 Kate Spence Thesis (Leeds 1945). 52 Axford and Norrish Proc. Roy. SOC. A 1948 192 518. 326 GENERAL DISCUSSION As shown by Dr. Reagh and myself,5s the surface effect in the slow oxidation of hydrocarbons only becomes apparent at a certain limiting diameter of about 5-10 mm. The theory of slow branching requires that the reaction shall be entirely inhibited a t a finite diameter when the net branching factor becomes less than zero by the increase of the surface deactivation. This was found to be the case for methane ethane propane acetylene ethylene and propylene.We should not expect the cool flame to be affected until a limiting diameter of tube was reached. Finally with regard to the oxidation of formaldehyde this shows none of the criteria of a degenerate branched reaction and the results have been explained by us in terns of a straight chain mechanism. We should not expect the dramatic results observed with hydrocarbons on decreasing the diameter. There is however a competition in chain ending between the volume and surface reactions in which the surface effect only becomes predominant a t very small diameters ( < I mm.) corresponding to packed vessels. Nothing here appears to me to be inconsistent with our inter- pretation of the oxidation of formaldehyde as a straight chain reaction. With regard to the “ dark blue ” flame i t will be seen that I have drawn attention to the similar flames in acetaldehyde observed by Townend and his co-workers in my paper.Dr. G . Porter (Cambridge) (communicated) As far as I am aware no short-lived free radical has ever been detected by its absorption spectrum in a chemical reaction at normal temperatures nor in an ordinary photo- chemical reaction despite numerous attempts with systems where radical mechanisms are known to be operative. I think it important to bear this in mind when assessing negative evidence of the kind mentioned by Dr. Spence though the workers quoted used a particularly sensitive method and their results certainly suggest a short lifetime of I O - ~ sec. or less of the OH radical if it was present.By using the method of flash photolpsis54 very high instantaneous concentrations of intermediates can be obtained and I have been able to observe the OH radical in absorption at high intensity in two systems which almost certainly involve the reaction between methyl and oxygen viz. the photochemical oxidation of acetone and the chlorine photo- sensitized oxidation of methane. Unfortunately even this positive result does not enable one to conclude that OH is formed in the primary reaction of methyl with oxygen rather than in the subsequent steps though i t is hoped that studies of the kinetics of OH appearance will make this dis- tinction possible. Dr. L. Bateman (Welzuyn Garden C i t y ) (communicated) Norrish’s contention that peroxides do not absorb light and are therefore photo- chemically inert in the wavelength range 3000-4000 A is misleading and his arguments as presented have not the force claimed.Admittedly peroxides ranging from about 0.5-10 at 3000 to 0-01-1 at 3650 A but such absorption is weak E of typical saturated and allylic hydro- and di- Gee and I have presented quantitative Fvidence 55 that the very strong catalytic action of irradiation a t 3650 A on the autoxidation of liquid olefins originates in the photolysis of the hydroperoxide. This catalysis certainly extends to wavelengths as high as 4000 A and possibly higher.56 The quantum efficiency of the photolysis at 3650 is of the order of 0.1. However simple carbonyl compounds are even less absorbing in this spectral region (for acetone E~~~~ N o - o o ~ ) while photolysis is of com- parable efficiency to the peroxide (Norrish Crane and Saltmarsh 5 7 report + = 0-2 at 3150 A).53 Norrish and Reagh Proc. Roy. SOC. A 1940 176 429. 54 Porter Proc. Roy. SOC. A 1950 200 284. 55 Bateman and Gee Proc. Boy. SOC. A 1948 195 376. 56 Bateman Trans. Faraday Soc. 1946 42 266. 57 Norrish Crane and Saltmarsh J . CAem. SUG 1934 1456. 327 GENERAL DISCUSSION Prof. R. G. W. Norrish (Cambridge) (communicated) I cannot sub- scribe to the cogency of Dr. Bateman’s criticism with acetaldehyde for example at 3400 the extinction coefficient is 1.0 whereas for tert.-butyl hydroperoxide and also for hydrogen peroxide the extinction coefficient is 0.1. In addition formaldehyde has an absorption coefficient of 0.07 a t 3560 while the peroxides have fallen to negligible values.I would point out that the peroxide with which he was concerned namely cyclo- hexene hydr~peroxide,~~ is one that would be expected to have consider- able absorption owing to the unsaturated nature of the compound. The combined effect of the double bond and the peroxide group will un- doubtedly cause light to be absorbed more strongly at longer wavelengths than in the compound where the peroxide group is the only chromophore. Add to this the fact that the intermediate concentration of aldehydes in the oxidation of methane and ethylene and in higher hydrocarbons at the higher oxidation temperatures is of the order of millimetres while the concentration of peroxides is undetectable analytically and I think i t must be agreed that the photochemical effect is to be ascribed to the aldehyde.In addition Mr. Booth in this laboratory has recently been studying the photolysis of tert.-butyl peroxide using a solution of 2 yo by volume in n-hexane. He finds no photolysis whatsoever when all wave- lengths below 3350 are filtered out. but considerable decomposition when only those wavelengths above 3000 are removed. Dr. C. F. Cullis (Imperial College) (communicated) Prof. Norrish has suggested that the large structural effect encountered in the oxida- tion of the normal paraffins is simply attributable to the greater number of points of attack in the longer molecules Thus he argues that n-octane is more readily oxidized than n-pentane since there are more CH groups in the former molecule and the removal of hydrogen atoms from such groups is considerably easier than from terminal methyl groups.On this view the increase in oxidation rate with chain length should be roughly proportional to the number of CH2 groups whereas in fact the variation is much more marked.59 In any case an extension of this argu- ment fails to explain qualitatively the facts relating to the oxidation of branched chain paraffins. It is well known in hydrocarbon chemistry that the order of reactivity of C-H bonds towards radical attack decreases in the order 3” > zo > IO and Rice has shown that at 3ooOC the chances of attack at primary secondary and tertiary carbon atoms are in the ratio I 3 33.60 For the isomeric hexanes for example i t is possible to calculate the relative probabilities of hydrogen atom removal from the molecules concerned by multiplying the numbers of such atoms attached to primary secondary and tertiary carbon atoms by Rice’s factors and summing over the whole molecule.If the ease of removal of a hydrogen atom from a hydrocarbon molecule is the main factor controlling the ease of oxidation as Prof. Norrish’s argument implies the figures in the final column but one of the following Table should be measures of the oxidizability of the compounds concerned. On this basis therefore the order of ease of combustion of the isomeric hexanes would be V > I1 = I11 > I > IV whereas the experimentally determined relative oxidation rates (given in the final column of the Table) show that the order is I > I1 > I11 > IV > V.59 Prof.R. G. W. Norrish (Cawbridge) (communicated) My remarks about the relative ease of oxidation if hydrocarbons are intended to indicate that other things being equal the ease of oxidation per carbon atom will increase from methane and ethane which will have values 58 Bateman and Gee PYOC. Roy. SOC. A 1948 195 376. Cullis and Hinshelwood Faraday Soc. Discussions 1947 2 I 17 ; Cullis and Mulcahy Revue l’lnst. Franpais Pe‘trole 1949 4 283. 60 Rice and Rice The Aliphatic Free Radicals (Johns Hopkins I935) p. 100. 328 characteristic of the CH,.group to a value characteristic of the CH group for an infinitely long chain.Such comparisons as are made by Dr. Cullis in his criticism of my statement appear to me to have little relevance because the bases of comparison are so doubtful. Firstly his comparison of the rates of oxidation of hydrocarbons is based on the rates of pressure changes the reactants being at comparable pressures in all cases. This does not take account of the great increase in " carbon concentration " as we pass to higher hydrocarbons which would have the effect of weighting his results strongly in favour of the higher hydrocarbons. Secondly the comparisons could not all be made at the same temperature and the accumulation of intermediate products greatly' depends on the tem- perature as well as the nature of the hydrocarbon. For example in Compound I CH .CH . CH . CH . CH . CH (I) CH CH,. C H . CH . CH . CH CH CH,. CH,. CH . CH,. CH CH CH, C . CH,. CH I CH CH,CH I I I I CH,. CH . C H . CH one experiment of Dr. Patnaik the oxidation of butane at 270' C using a butane oxygen ratio of 1-5 and a total pressure of butane of 76.1 mm. intermediate products such as aldehydes and alcohols accounted for over 25 mm. in a total pressure change of 60 mm. It is not legitimate to com- pare the rates of oxidation of hydrocarbons by pressure changes in view of this great variability of end-products and the only certain way is to measure the rate of disappearance of hydrocarbon by analytical technique. This has not yet been done. Finally Rice's figures refer to the relative reactivity of CH, CH and CH respectively with alkyl groups whereas in comparing the rates of oxidation one is concerned with their reactivity towards oxygen or OH.There is no reason why there should be any close quantitative parallelism in the two cases which will undoubtedly be differently conditioned by temperature and steric hindrance. Such calculations as have been made on this basis can have little quantitative value until much more data have been accumulated and until some more uniform basis of comparison of reaction rates has been devised. Dr. D. W. G. Style (London) said When illuminated with Schumann ultra-violet light diethyl peroxide and ethyl nitrate emits as a fluorescence the same band system which extends from approximately 3300 %L to 5000 A.The emiss:ons from methyl chloroformate and methyl nitrite are also identical but different from that from the first two substances. 78 GENERAL DISCUSSION TABLE I1 Relative Probabilities of H Atom Removal Total telative telative Proba- Oxida- ilitr of tion Rate From IO From 2' I Atom From 3" C Atoms C Atoms Z A toms :emoval 1580 4 x 1 30 8 x 3 54 560 (11) 9 x 1 4 x 3 60 54 9 x 1 4 x 3 (111) I 2 18 I 2 x I (IV) 2 x 3 - I I 2 x I (V) GENERAL DISCUSSION 3 29 I t is probable that the emitters of the two band systems are respectively C,H,O. and CH,O-. Have these spectra been observed with other sources particularly cool flames ? Dr. Peter Gray (Cambridge) (communicated) It is very interesting to see the spectra of molecules containing the alkoxyl group shown by Dr.Style and to learn that the emitter is likely to be the alkoxy radical itself. It is hoped soon to examine a system in which the methoxy (CH,O-) group thermally produced may be an emitter viz. the lumin- escent thermal decomposition of methyl nitrate. The initial decom- position of methyl nitrate gives nitrogen dioxide and methoxy groups and is accompanied by a blue luminescence. The same behaviour is displayed by methyl nitrite. The light may be due either to CH,O- or to excited formaldehyde CH,O* and now both spectra are known i t may be possible to identify the source of this luminescence. Sir Alfred Egerton (Imperial College) (communicated) The emission bands which Dr. Styles has found and ascribed to the C,H,O radical in the spectrum of glow in diethylperoxide are interesting.When Dr. Harris was working on the decomposition of diethylperoxide in my laboratory' we looked for bands in the glow but a t that time had not the means for photographs of long exposure. It is possible these bands have some relation to the pronounced absorption bands obtained in the com- bustion of hydrocarbons which Pidgeon and I reported in 1933. Dr. C. A. McDowell (Liverpool Univevsity) (communicated) With re- gard to the oxidation of acetaldehyde Dr. Thomas Mr. Farmer and I have shown fairly conclusively by inhibition studies G 1 ~ 6 2 that the thermal reaction u p to about 140'C is a chain reaction in which two radicals the acetyl and peracetyl radicals play the predominant part.Dr. G. J. Minkoff (Imperial College) said The heats of combustion presented in our paper clearly have some peculiarities since the values of (-0-0-) which can be derived are of the order of 100-140 kcal. We intend to repeat thf measurements a t some future time but have mentioned the results already obtained as the work was done most care- fully and as the values are self-consistent within themselves (cf. Qc for members of series and for isomers). We have also recorded the refractive indices and melting points which we obtained so as to have as many published data on this subject as possible. These should be of use in determining the purities of future samples though we would stress the great advantages of determining the infra-red spectra of the peroxides during the preparations ; as the standard methods of purification (e.g.fractional distillation) are often inapplicable i t is a great help to know what particular impurities have to be removed. A brief point is that the latent heats of evaporation show high Trouton coefficients for all the hydroperoxides; this agrees with infra-red studies of the OH fre- quencies in suggesting that the compounds are strongly associated. DeteIminations of the heat of formation and of D(-0-0-) should thus if possible make allowance for the heat of dimerization etc. involved. Finally I should like to explain the use of the terms " acetyl propionyl and butyryl hydroperoxides " to describe the corresponding peracids. Mr. Everett and I have measured the (classical) dissociation constants in aqueous solutions of these compounds and of H,O, Me0,H and EtO,H and have found that the peracids are intermediate in acid strength between the corresponding hydroperoxides and carboxylic acids.In fact the increase over the hydroperoxide can be shown to be approx- imately accounted for by inductive and polarizability effects. We there- fore suggest naming the peracids " acyl hydroperoxides " to avoid the implication of a great acid strength sometimes inferred from the term McDowell and Thomas J . Chem. SOC. 1949,2208 2217 ; 1950 1462 ; Trans. Faraday SOC. 1950 46 1030. G 2 McDowell and Farmer unpublished work. GENERAL DISCUSSION 3 30 peracid (cf. perchloric and permanganic acids). Further if the diacyl peroxides are referred to as such there will be less chance of confusion with the corresponding acyl hydroperoxides (cf.dibenzyl peroxide benzoyl peroxide and perbenzoic acid). Dr. W. E. Vaughan (Emeryville California) (communicated) In the course of our own studies on organic peroxides we have obtained certain physical data on tert.-butyl hydroperoxide and di-tert.-butyl peroxide pounds were better than gg yo pure we do not wish to claim superior which differ from those just presented. Although we believe our com- accuracy for our data but wish rather to present them for the record in the hope that the differences may be resolved in the future. tert.-Butyl hydroperoxide 1.4010 5'5 I 15-0 log10 P (mm.1 I 1-5 kcal./mole 65 4.2 kcal./mole +alp . K p . "C . B.p. "C . Vapour pressure tained Di-lert.-butyl Penoxide 1.3890 - 40.0 I 11'0 9.6 kcal. /mole 12 75.0 kcal. /mole 0.0005 . Heat of vaporization . . Heat of combustion . Dr. M. Magat (Paris) said Leadbeater 63 has recently prepared in our laboratory some carefully purified peroxides and measured their refractive indices and Raman spectra. The following results were ob- Diethyl peroxide nb5 = 1.3720 f 0-0005 Di-a-hydroxydiethyl peroxide nh6 = 1.4265 Ethyl-a-hydroxyhydroperoxide n:4 = 1.4150 f 0-0010. For the diethyl peroxide the agreement with the refractive index given by Egerton et al. is excellent. The characteristic Raman frequency for the 0-0 bond found in all peroxides investigated by Leadbeater as well as in dibenzylperoxide and in H,O is located at 880 -& 3 cm.-l.I would like also to call attention to the danger of working with very pure crystallized peroxides even with those classified as relatively stable a spontaneous or induced rupture of the crystal has led to an extremely serious explosion when monohydroxydiethyl peroxide was purified by' crystallization. Dr. N. S . Wooding (Courtaulds Ltd. Coventry) (communicated) Cobalt acetate in concentrations of IO- to 10-3 M has been shown to be an effective catalyst for the oxidation of trimethylethylene in solution by gaseous oxygen. Recently i t has been found that the autoxidation of cellulose under alkaline conditions was also catalyzed by cobalt acetate.The effect of other metal ions was investigated and some were found to catalyze while others retarded the autoxidation. Such behaviour has been reported elsewhere.65 However manganese salts were found to behave both as catalysts and as retarders depending upon the concentra- tion of salt used. A possible explanation of this effect has been suggested in terms of the mechanism of metal ion catalysis postulated in this paper and else~here.~42 O 5 I would like to ask Prof. Eawn if he has found any evidence for negative catalysis by metallic cations in the autoxidation 63 Leadbeater Cow@. rend. 1950 230 829. 64 Entwistle Cole and ?Tooding Textile Res. J. 1948 19 527 609. 65 George and Robertson Tram. Faraday SOC. 1946 42 217. GENERAL DISCUSSION 33 1 of trimethylethylene or any of the other systems he has investigated since such phenomena if the suggested explanation is correct should be observable in other systems under the appropriate conditions.Prof. C. E. H. Bawn (Liverpool) (communicated) Under the experi- mental conditions so far studied the condition in which the rate of oxida- tion is independent of the catalyst concentration as observed by George and Robertson with saturated hydrocarbons has not been attained. There is no evidence at present that the catalyst terminates chains by a process such as Solvent 258 (-4 I <IO % decrease'in I conc. in 3 hr. 66 RO + Co++ -+ R0,- + Co+++. Dr. C. F. H. Tipper (Edinburgh) said It is nearly always assumed when considering the oxidation of hydrocarbons in solution that the hydroperoxides initially formed decompose to give free radicals.However under certain conditions for example in solvents of high dielectric constant or especially in the presence of acids as Dr. Waters has stressed heterolytic fission of the 0-0 bond may occur to give ions. The evidence for this has so far been mainly' organic,66 and so I would like to report the results of some kinetic measurements on the decomposition of decalin hydroperoxide in various solvents. The overall decomposition was found to be first order in all cases at any rate at low concentrations ( < I O - ~ mole/l.). The rates at 130' C and overall activation energies of decomposition in different solvents are shown below.TABLE I11 First Order Constant Time of Half Cnange Chlorobenzene . . o-Dichlorobenzene . Ethylene glycol . . @in.-1) Very slow 0*0105 0.002 68 0.0332 0.359 30,250 29,800 16,700 22,140 21 2 I Acetic Acetic acid acidlwater i 40% H,O by volume J . The addition of water to the ethylene glycol or the acetic acid thus presumably greatly increasing the dielectric constant of the solvent increased the rate of decomposition considerably. In the case of the acetic acid addition of 2 yo by volume of water had little effect but further addition up to about 30 yo by volume caused a large increase in the rate. Above 30 yo increase in the water content had no effect. Also in chlorobenzene solution a t I 15.5' C no decomposition was detectable over a period of hours but if 1-5 mole yo of acetic acid was added the time of half decomposition fell to 64 min.These results would seem to show that in the first three solvents de- composition of the hydroperoxide molecules into radicals is taking place possibly followed by a chain decomposition but that with a change in conditions an ionic decomposition can occur very readily'. Water and organic acids are very often stable products of oxidation of hydrocarbons in solution and thus as the reaction proceeds it is possible that ionic decomposition of the hydroperoxide formed might become important. Prof. Bawn (Liverpool) said IXr. S. F. Mellish and I have observed that the stable free radical aa-diphenyl p-picryl hy'drazyl reacts rapidly with the radicals of the type RO.and this provides a simple and convenient method for studying the rates of dissociation of peroxides into radicals 66 For example Robertson and V'aters J . Chem. Soc. 1948 1577. Bartlett and Cotman J . Amer. Chem. SOC. 1950 72 3095. Kharasch Fono and Nudenberg J . Org. Chem. Igjo 15 748; 1951 16 113 128. GENERAL DISCUSSION 3 32 (ROOR -f RO. + RO.). The vividly coloured radical which gives stable solutions in a wide range of organic solvents undergoes a sharp colour change on reaction with radicals which may be measured in a simple colorimeter. This method has been used to measure the rate of dissocj- ation of polymerization in;t'ators such as peroxide and azonitriles. Dr. W. A. Waters (Oxford) said A similar instance to the decomposi- tion of decalin hydroperoxide quoted by Dr.Tipper is that of tetralin hydroperoxide which was reported some time ago by Robertson and This afforded a good example of a reaction which appeared from kinetic study to be much more simple than i t really was. The results reported by Dr. Henderson show the extent to which very minor products or impurities can influence the course of autoxida- t:on. In the autoxidation of a related compound dibenzyl ether (Ph . CH,),O which has recently' been studied by Mr. Wickham-Jones and myself we have been able to show that irregularities in the oxidation are due to the formation of a trace of a phenolic by-product. Dibenzyl ether gives a stable peroxide and in the main the uncatalyzed rate of oxygen absorption is independent of the concentration of the peroxide formed.Homolytic dissociation of the peroxide to give more chain- starting free radicals does not therefore play a major role in determining the oxygen uptake rate. However the autoxidation is self-retarding and we have been able to show that as the peroxide of dibenzyl ether decomposes there is gradually formed just enough phenolic material to give a positive indophenol reaction. We ascribe the gradual retardation of the autoxidation to the formation of this phenol and have noted that when our reaction vessel was packed with chopped glass wool there was less peroxide decomposition less formation of phenol and less retardation of the autoxidation. Here the decisive factor seems to be surface catalysis of the mode of the secondary reaction-the peroxide decomposition- yet i t significantly influences the whole autoxidation process.Sir Alfred Egerton (Imperial College) (communicated) With refer- ence to Dr. Henderson's paper i t is well known that reproducible results are not obtained in hydrocarbon oxidation until the surface of the vessel has been conditioned by previous experiments ; the wall catalyst obtained in the liquid-phase oxidaton of ethylbenzene seems to be in line with this effect. Prof. J. P. Wibaut (Amsterdam) said In collaboration with Dr. A. Strang 6 8 we have carried out an investigation into the oxidation in the liquid phase by molecular oxygen of a number of normal alkanes with 8 to zz carbon atoms and of some branched octanes.When cobalt stearate is used as a catalyst the oxidation proceeds at a measurable rate. It has been found that the first stage of the oxidation reaction consists in the formation of a hydroperoxide. z 5-Dimethylhexane is slowly oxidized by molecular oxygen a t zoo C and a crystalline dihydro- peroxide melting a t 106.5~ C is formed CH3 CH CH3 CH I I + 2 0 2 4 H,C-C-CH2-CH2-C-CH3 I H I H3C-C-CHz-CH2-C-CH3 H I 0 0 I 1 0 0 I H Ir The presence of small quantities of peroxides can also be detected in samples of other saturated hydrocarbons which have been kept for a long time for instance in 3 4-dimethylhexane z 5-dimethylhexane 3-methylheptane n-nonane n-hexadecane methylcyclohexane.The primarily formed peroxide is decomposed under the influence of the cobalt 67 J . Chem. Soc. 1948 1578. 68 W'ibaut and Strang Proc. Kon. Neder. Akad. Wet. B 1951 54 ( z ) 101. 333 GENERAL DISCUSSION ions ; the radicals thus formed start a chain mechanism so that the oxida- tion reaction proceeds ROOH + Co++ -+ RO' + OH- + Co+++ ROOH ~2 ROO- + H+ Co+++ + ROO- -+ Co++ + ROO. We ascertained what products are formed by the oxidation of 2 5- dimethylhexane ; with reference to these reaction products a reaction scheme can be drawn up. The characteristic feature of the reaction scheme for the catalytic oxidation in the liquid phase of z 5-dimethyl- hexane and of other branched hydrocarbons is that the chain mechanism is initiated by an alkoxy radical.We ascertained the maximum rate of oxidation of normal alkanes with 8 g 10 12 14 16 18 20 and 22 carbon atoms. The experiments were carried out at 110.4" C and with 0.112 mmoles cobalt stearate per 61.7 mmoles hydrocarbon. There is a linear relation between the number of carbon atoms and the maximum rate of oxidation from C,,H, to C Z z H 4 6 . This can be explained by assuming that all the secondary carbon atoms have an equal chance of reaction. Some branched alkanes oxidized much more easily than the isomers with normal chains ; we ascertained the maximum rate of oxidation at 78.1" C measured in ml. 0 per 61.4 mmole hydrocarbon per hour (catalyst 70 mg. cobalt stearate) to be . 11.0 - 35'0 3-methylheptane z 5-dimethylhexane z 2 4-trimethy'lpentane .0-0 a-methylheptane 3 4-dimethy'lhexane . 3-methyl-3-ethylpentane . 0-0 . . 3'0 The oxidation begins at a tertiary C-H bond from which a hydro- peroxide group is formed. A considerable quantity' of acetone is formed during the oxidation of 2-methylheptane methylethylketone being formed during the oxidation of 3-methylheptane. The fact that z z 4-trimethylpentane (iso-octane) is not oxidizable under the conditions chosen can be explained by steric hindrance. The quaternary carbon atom which has three methyl groups screens the tertiary carbon atom to such a degree that an oxygen atom cannot approach the tertiary hydrogen atom to within the distance pertaining in the transition state. Dr. M. F. R.Mulcahy (Melbourne) (counrtzunicated) Several con- tributors 89-7 have discussed the liquid-phase oxidation of hydrocarbons catalyzed by decomposing peroxide or metallic catalysts in terms of the propagation mechanism ROZ- + RH + ROOH + R-. It may perhaps be of interest to recall that there is evidence that . . 5.0 Roy. SOC. A 1946 183 337. R- + 0 + ROZ- in the absence of catalysts (and of light) formation of hydroperoxide may occur by' a different mechanism. This was shown by the work of George and Robertson 73 on the " thermal " oxidation of tetralin. A similar result has recently been found by Mr. Watt in this laboratory with the uncatalyzed oxidation of benzaldehyde,' the kinetic behaviour being analogous to that of the tetra'in reaction.In the presence of benzoy'l peroxide however a reaction of the type shown above is initiated and is (additionally) superimposed on the uncatalyzed reaction. 69 Bateman Gee Morris and Watson this Discussion. 7O Bau-n Pennington and Tipper this Discussion. 71 Brook and Matthews this Discussion. 72 Mulcahy this Discussion. 73 George and Robertson Proc. Roy. SOC. A 1946 183 309 ; George Proc. 74 Mulcahy and Watt Nature (in press). GENERAL DISCUSSION 334 Dr. N. Uri (Munchester) said The liquid-phase reactions in which a wall effect is observed are few and therefore remarkable. There are ody four such cases known to me (a) the experiments relating to the oxidation of liquid ethyl benzene described by G. M. Henderson in this Discussion; (b) my own findings in the catalytic decomposition of hydrogen peroxide under certain experimental conditions ; 75 (G) the observations made by Dain and Kachan 7 G in the photochemical oxidation of water by ceric ions.It is not unlikely that the wall effect in this case leads to a recombination of OH radicals and a subsequent instantaneous decomposition of hydrogen peroxide by’ ceric ion. (d) In our work on the photo-initiated free radical polymerization of vinyl compounds in aqueous solution77 we made the observation that distance < rmm. from the wall the polymer is produced (by’ recombina- under conditions when practically all the free radicals are formed a t a tion of active endings) exclusively on the wall; none appears to be SUS- pended in solution.No doubt these experiments require some co-ordination and in this connection i t is interesting to note that all these reactions involve free radicals as intermediates and the process effected by the wall is con- sidered to be a termination process which may in some cases lead to a new type of chain reaction as reported by Henderson. In the oxidation of lubricating oils the effect of iron or copper as single catalysts was studied by Brook and Matthews. It is well known that in the catalytic decomposition of hydrogen peroxide the joint action of iron and copper as co-catalysts is much more than additive. It would therefore be interesting to study those effects in the oxidation of hydrocarbons particularly if peroxides are postulated as intermediate products.Dr. G . M. Henderson (Blackley) (communicated) The effects of walls on liquid-phase chain reactions are fairly’ well indicated in past work and to restrict it to two examples on oxidations excluding all references to solid catalysts the most relevant papers are those of Stephens 78 working with cyclohexene and Medvedev and Podyapol~kaya,~~ working with tetra- lin where many effects akin to the present work were noted and the effects of walls mentioned only with very different explanations. The novel feature of ethyl benzene to which we wished to draw attention as a convenient expet imental medium is that the two hydroperoxide forming mechanisms seem more clearly distinguishable and mutually exclusive. With very fresh and pure ethyl benzene we presume that the original wall termination process is without further visible effect but that in an older sample or after sufficient of the initial peroxide has further decom- posed some of the breakdown products possibly acids act as haptens (to borrow a term from immunology) and a new wall termination process arises which gives rise to visible effects on the rate of reaction.One can generalize that in other similar reactions that the two possible re- actions might then run concurrently or that the second stage might never even arise. Sir Alfred Egerton (Imperial College) (communicated) This paper refers particularly to the inhibition of the oxidation of lubricating oils I would like to draw the authors’ attention to a paper by Hanson and 75 Uri J .Physic. Chem. 1949 53 1070. 76 Dain and Kachan A C.S. Abstr. 1949 43 7349. 77 Unpublished observations. 78 Stephens J . Amer. Chew. Soc. 1936 58 219. 7g Medvedev and Podyapolskaya J . Physic. Chew. U.S.S.R. 1939 12 79. tiENERAL DISCUSSION 335 myself 8o on " Nitrogen oxides in internal combustion engine gases " in which the promoting action of the nitrogen oxides on the oxidation of lubricating oils was investigated and a subsequent paper entitled " In- fluence of catalysis on oil oxidation " 81 iron oxide was found to inhibit the nitrogen oxide catalyzed reaction. Dr. J. B. Matthews and Mr. J. H. T. Brook (Thornton) (communi- cated) In connection with Dr. Uri's suggestion that the study of mixed iron and copper catalysts would be of interest in the oxidation of hydro- carbons] results have been given in the paper showing that in the inhibited reaction the two catalysts are additive in their effect on the initial rate of reaction.In the uninhibited reaction] however the measured values of t$ are greater than the added effects of the individual catalysts thus lending support to the inclusion of peroxides as intermediate products. Values of t$ for iron and mixed iron-copper catalysts are given in Table 11 I 0 c s WEiCHT FRACTION OF HYDROCARBON FIG. I. of the paper and values of 4 for copper can be calculated from the data in Table I together with values of the intercepts with the oxygen absorbed axis of the asymptotes to the curves in Fig. 3 using the equation = ( A + B)t - (B/4) to describe the asymptote.Mr. J. H. T. Brook ( T ~ C Y M ~ O O ~ Z ) said I would like to add some further information on the kinetics of the iron-catalyzed inhibited oxidation. Using 50 p.p.m. of iron added as ferric stearate at 1 5 0 O C and using diphenyl as the inert solvent the dependency of the reaction upon the oil concentration was found to be of the form (as Fig. I above) [RHI cc 220 + 330 [RH]' No immediate explanation of the difference in kinetics between the iron and copper-catalyzed reactions is apparent. ' 0 PYOC. ROY. SOC. A 1937 153 90. 81 Symposium on Engine Wear Inst. Mech. Eng. 1937. 308 GENERAL DISCUSSION GENERAL DISCUSSION Prof. W. A. Noyes Jr. (Rochester) said Recent experiments by Mr. Martin in our laboratory indicate that methyl radicals from the photo-lysis of mercury dimethyl upon reaction with oxygen at low pressures give CO and CO,.Not enough results were available when I left to allow calculation of rate constants although the rates are in approximate agreement with the acetone results. Some uncertainty exists concerning the nature of the primary process in mercury dimethyl so that many more experiments may be necessary (only two had been made before I left) before quantitative conclusions can be drawn. Other results have been obtained with acetone-oxygen mixtures by Messrs. Marcotte and Durbetaki at intensities about 10 times those pre-viously used both in the presence and absence of mercury vapour. It is concluded that mercury vapour does not affect the results by more than the experimental error.At these high intensities in the absence of oxygen at 120Othe acetyl radical seems to be sufficiently stable to permit appreci-able reaction of the type and possibly CH + COCH = CH,COCH . - (1) 2COCH = (COCH,) . - (2) Thus a t this temperature the yield of CO + CO may be as much as six times the yield of CO in the absence of oxygen. At zooo however, the rate of decomposition of COCH is sufficiently rapid so that the results at these higher intensities agree with those previously found. These experiments are being performed with a view to permitting an analysis of products other than CO CO, CH and C,H,. In general the results are similar to those obtained at lower temperatures but i t is evident that at this temperature one has reached a point that chain propagation is setting in.This fact may limit the temperature range over which i t will be possible to deter-mine the activation energy for HCO decomposition. The fact that the CO is less sensitive to oxygen pressure and is higher than would be ex-pected from the low temperature results may be due to hydrogen abstrac-tion by radicals other than methyl. Dr. Finkelstein has preliminary results on the decomposition of diethyl ketone in the presence of oxygen. At 25°C the yield of CO falls as the pressure of oxygen increases indicating that the COC,H radical reacts with oxygen to give something other than CO at this temperature. In this respect i t seems to resemble the acetyl radical. However the quantum yield of oxygen disappearance is still rising at the highest oxygen pressure used (0.18 mm.) and has reached a value of 6.Complete oxidation of I molecule of ketone to CO and H,O would require 7-5 molecules of oxygen. Since small amounts of C hydrocarbons are found either more than I molecule of ketone disappears per quantum (doubtful a t 25OC) or the oxidation is proceeding quite completely. Further conclusions are not warranted at this time. At 200' C the yield of CO + CO seems to reach a constant value of 3 to 3.2 with increase in oxygen pressure the CO being about 2.3 and the CO about 0.8. However the quantum yield of oxygen disappearance has reached 7 at 0.78 mm. oxygen pressure and shows no sign of reaching a constant value. C hydrocarbons are still formed to the extent of about 0.3 mole per quantum even under these conditions.It would be premature to give a mechanism based on these results. The total yield of CO + CO is about the same as for acetone thus in-dicating that either a CH radical or a C,H radical yields either CO or CO when i t reacts with oxygen. However since the CO yield falls with increase in oxygen pressure for acetone and rises under similar circum-Mr. Marcotte has also made experiments at 225OC. Thus CO + CO may have yields of 3-6 or more GENERAL DISCUSSION 3 09 stances for diethyl ketone the explanation for the two phenomena must have important differences. Dr. R. Spence (Harwell) said I would like to recall some experiments by Dr. Bates and myself on the oxidation of methyl radicals produced by the photodissociation of gaseous methyl iodide a t room temperatures.The products formaldehyde methylal water and iodine were identified and determined quantitatively and the reaction was shown to follow the equation qCH,I + 2 0 % -f (HCHO) + (CH,O),CH + H,O + P I . h9 Two primary mechanisms suggested themselves namely, CH,I-+ CH +- I CH + 0 -f CH,O + OH * * (1) and CH + 0 -+ CH,O,. - ( 2 ) The absence of any inert gas effect which might have been expected with mechanism (2) seemed to favour the hydroxyl mechanism (I) but Blaedel Ogg and Leighton failed to detect the OH radical by its ab-sorption spectrum and therefore proposed that a mechanism of type (2) is operative. Prof. Noyes’s results at high temperatures suggest that the fonnyl radical is produced from the interaction of methyl radicals and oxygen but the above results indicate fairly clearly that i t is not formed at room temperatures.This discussion has drawn attention to the fact that after twenty years we are still unable to give precise answers to such fundamental questions as how long such entities as 0 and CH can stay together on collision and what type of stabilizing collisions are subsequently required. Dr. N. Uri (Munchester) said The HCO radical was postulated by Marcotte and Noyes as an intermediate product in the photolysis of acetone in the presence of 0,. It appears to me very important to learn more about this radical which may also play an important role in the process of photo-synthesis. I consider that formaldehyde may be formed from formic acid by an electron transfer i.e.e + HCOOH +- HCOOH. + HCO + OH* HCO + R . H 3 HCOH + R, where the bond R H has a relatively small bond dissociation energy. On the other hand i t is postulated that HCO decomposes into H + CO. In this connection i t would be interesting to have more data on the formation of formaldehyde in the experiments discussed by Prof. Noyes, so as to ascertain under what conditions HCO + RH + HCOH + R occurs in preference to HCO -f H + CO. Prof. A. R. Ubbelohde (Belfast) said It may be useful to reconsider to what extent the recombination of two radicals as in H+O,+HO . (A) or CH + CH -f C,H . - (B) requires a ternary collision to stabilize the product. Why do the energy-rich products not fly apart again almost at once as they would in simple atomic collisions such as H + 0 -f HO ? In the theory put forward in 1935 for the formation of peroxide radicals R + 0 -f R02* * (1) R02* + RH + ROOH + R .. (2) I attributed the temporary stabilization of the RO complex to “ quantum smudging ” notable for R > C,H, but more rigorous conditions should apply when R is small as in reaction (B) and especially (A). Bates and Spence J . Amer. Chem. SOC. 1932 53 1689. Blaedel Ogg and Leighton J . Amer. Chem. SOC. 1942 64 2499 3 10 GENERAL DISCUSSION Prof. S . W. Benson (California) said Prof. Ubbelohde has raised a question with regard to the allowability of two-body recombinations such as have been employed by the present authors. I believe that this question can be answered with considerable more certainty than has heretofore been the case.A recent theory of Dr. N. B. Slater has shown that to a first approximation the rate of unimolecular decomposition of an activated complex with energy E in excess of a critical energy E* needed for decomposition is given by k(E) = (1 - f>"-l, where Y lies in the range of normal frequencies (1012-1014 sec.-l) and n is the number of normal co-ordinates required for the description of the decomposition. The average molecule with energy in excess of E* has energy - E* + kT so that for any n v and E* we can calculate the mean life t ( c ) = I / K ( E ) . If we use mean values such as E*/kT - 40 ; v = 1013 sec.-l we find (with Slater) that t(E) w 10-l3 sec. if n = I ; t ( ~ ) E I O - ~ sec. if n = 5 and t ( ~ ) E 10-4 sec.if n = 10. These mean life-times (except for n = I) are very much larger for low temperatures. Thus for ethane e*/kT - 140 at room temperature and if we set n = 5 the mean life of an excited molecule will be I O - ~ sec. These results imply that for the reverse reactions the recombinations of free radicals complexes may be formed having mean lives sufficiently long to assure even at very low pressures quite efficient stabilization by collisional deactivation. This will be true for the high-temperature pyrolysis reactions and will be more pronounced for photochemical re-actions where temperatures are considerably lower. A reaction to form the radical HO is one that is close to the permissible limits. Here n < 3 ; E* > 45 kcal.and the average complex formed at 130O C has a mean life t ( ~ ) < v(kT/E*)2 w 3 x I O - ~ sec. which is of the order of collision times a t S.T.P. At temperatures of 500' C to 600° C or higher this mean life is much shorter and i t is probable that most of the HO complexes decompose before they can be stabilized. In the more complex methyl radical the theory gives a reasonable accounting of the high collision efficiency which has been observed for their recombination. It will not take many degrees of freedom in a com-plex with such a high critical energy to give i t a very long mean life. Finally we may say that recombinations will manifest their three-body character when the complex has a low critical energy the temperature is high and the combining radicals are not complex.Dr. C. A. McDowell (Liverpool University) (communicated) I should like to mention that Dr. Thomas and I have found that in the presence of a large excess of oxygen the acetyl radical is quite stable up to about zoo0 c. Dr. A. D. Walsh (Leeds) said Were any analyses for formaldehyde carried out by Noyes and Marcotte ? A t the higher temperatures normally used for the study of methane oxidation it was difficult not to suppose the reaction of CH radicals and 0 to produce HCHO took place. Dr. Peter Gray (Cambridge) (communicated) I am very interested to read of these experiments on the peroxides and in particular of the role of the alkoxy radical RO. Peroxides are encountered as an important feature of some of the chemiluminescent (cool-flame) processes in hydro-carbon oxidation and the alkoxy radicals are presumably also present.Such RO radicals are also important in the thermal decompositions of the alkyl nitrates and nitrites. The initial step is the fission of the molecule to give the alkoxy radical and the appropriate oxide of nitrogen. Both these processes may be accompanied by a luminescence.41 Methyl McDonell and Thomas J . Chem. Soc. 1949 2208 2217. Gray and Y o f f e Proc. Roy. SOC. A 1949 200 114, fi Gray xg51 unpublished work GENERAL DISCUSSION nitrite gives nitric oxide which may not be expected to oxidize the CH ,O- . Thus this chemiluminescence observed with nitrates nitrites and in the oxidation by nitrogen dioxide of the simple alcohols may be a property of the alkoxy radical itself.Furthermore from experiments on the nitrates values may be ob-tained under favourable circumstances for the heats of formation of the alkoxy radicals. The heats of formation of the other species involved and the energy required to break the 0-N bond are needed the latter may be identified with the energy of activation derived from measure-ments of initial rates. Such values may be employed both as a check on other measurements and as a means of calculating provisional values of heats of formation of peroxides of which the activation energies of decomposition but not the heats of combustion are known. The heats of formation of the methoxy and ethoxy radicals may be obtained from the following data on nitrate and nitrite esters. METHOXY COMPOUNDS (i) CH,ONO,(g) = CH,O + NO ; Eact = 39-5 kcal.Qj = 29.4 - 8.1 Qf(CH30) = - 2 ,, (ii) CH,ONO(g) = CH,O + NO ; Eact = 36'4 8 Qf = 16-7 - 20.9 Qj(CH,O) = + 1.2 ,, CZH~ONO,(~) = C&',50 + NO ; Eact Qj = 33.4 - 8.1 Qf(C,H,O) = + z'o,, The formation of each radical from its elements is numerically small and errors in any of the primary quantities appear magnified. The values (- 0.4 f 1.2) kcal./mole and (2 f 2) kcal./mole might be assigned to the CH,O and C,H,O radicals respectively. Sir Alfred Egerton (ImperiaE College) said I can allay Dr. Stevenson's anxiety about the measurements of heat of formation of peroxide which are mentioned in the preliminary draft of the paper by myself Emte and Minkoff. The values found though the subject of careful measure-ments are unconvincing; for instance the heats of formation of the peroxides are in most cases greater than those of the corresponding more stable substances.It is not easy to determine accurately the heats of combustion of these explosive peroxides. We intend to redetermine them and to withdraw for the present the last section of our paper referring to the heats of formation of peroxides. Dr. A. G. Gaydon (Imperial College) said I should like to point out that the values for the dissociation energies of H, 0 and OH taken from my book are for oo K not 298OK. Fortunately the errors about 0.8 kcal./mole nearly cancel out for the equations as used. Dr. A. D. Jenkins (Courtaulds Ltd Maidenhead) (communicated) : Vaughan ei al. report that vapour-phase pyrolysis of methyl-tert.-amyl peroxide (I) gives an 8 % yield of methyl ethyl ether which they ascribe to vapour phase combination of a methoxy radical with an ethyl radical.The scheme of decomposition of this peroxide will be substantially ETHOXY COMPOUNDS = 39-5 kcal. CH, / CH, / CH,-O-O-G-CH -+ CHSO + O-C-CH, Gray and Yoffe J. Chem. SOC. 1950 3180 312 GENERAL DISCUSSION Instead of the combination reaction suggested above i t is quite possible that the ether could be formed by reaction between the ethyl radical and the substrate by attack at the -0-O- bond : CH3 / \ CH3 / \ CZHS + CH300C-CH3 -+ CH30CzHs + -0-C-CH3 C2HS CZHS Whilst no data are available for estimating the activation energy of this process from experimental results one can make a comparison with E for the alternative hydrogen abstraction reaction, CH, / \ C,H + CH,OOC-CH -+ C,H + radical, C2HS in the following way.In the former reaction an -0-0- bond is broken to form an 0-C bond so that the exothermicity will be of the order (go-55)=35 kcal. and we can represent this process by the diagram : FIG. I. In the second case the reaction is nearly thermoneutral and the bond broken is a C-H bond. The crucial factors are the shapes and positions of the two repulsion curves but if they are not very different in the two cases the activation energy of the former process will almost certainly be lower than t h a t of the latter and therefore less than approx. 13 kcal. If the decomposition of this peroxide followed a chain type of reaction the chains would be very short if 8 % of the products resulted from the termination process so that the methyl ethyl ether may be at least partly formed by the reaction proposed here.Thus the corresponding diagram will be Fig 2. Thus the general reaction R + R’OOR” -+ R’OR + R”0 or R”0R + R’O, R’O -f R + CH,O or R”Q 3 R + CH,O. may occur in diaIkyl peroxide decompositions where R results fro GENERAL DISCUSSION 313 It may be noted that nitrates or nitrites might undergo a similar reaction even more readily since the 0-N bond has a dissociation energy of the order 38 kcal. and the reaction will be even more exothermic than the corresponding peroxide reaction. Dr. L. Bateman and Mrs. Hilda Hughes (Welwyn Garden City) (communicated) The decomposition of cyclohexene hydroperoxide in solution in the temperature range 60-100’ C exhibits rather different features to those reported by Bell ei! al.and also by other workers for other hydroperoxides.718 Our investigations in this field are still pro-ceeding but certain results appear to warrant consideration here. The non-formation of diphenyl and phenol indicates that the solvent is inert under our conditions. Cyclohexenone and R0,-double bond addition compounds appear to be the main organic products. Each mole of peroxide decomposing liberates nearly 0.5 mole of water and much less oxygen (- 0.1 mole). The Arrhenius activation energy is 26 kcal./mole. We conclude that the measured rate refers to the one-step bimolecular decomposition, In benzene solution the reaction is almost exactly second order.RO2-H + HO-OR + RO2- + H2O + RO-. - (1) Radical t C H5 + H - - - -Radical /C H C H~ + CN 00ccc>15 I \ ’C H i FIG. 2. In cyclohexene solution the decomposition proceeds faster (roughly five-fold) and the order is reduced to 1-7-1.8. These characteristics the isolation of cyclohexenol as the major product,O and a linear relationship between [R02H] /rate and I /[RO,H] are consistent with a chain reaction comprising (I) as initiation step and RO- + R’H + ROH + R’- - (4 R’- + ROJ3 + R’OH + RO- . * (3) zRO-1 . * (4) R’- + RO- 3 Inactive products (5) 2R’- i ’ . * (6) wheze in this case R and R’ are identical and the reasonable assumption 10 is made that kz = A&,. In the presence of stearic acid the reaction is greatly accelerated and becomeF of the first-order with respect to the peroxide.These facts correlate strikingly with the catalytic action of the hydro-peroxide on the oxidation of the parent olefin with molecular oxygen, Farkas and Passaglia J . Amer. Chem. SOC. 1950 72 3333. Kharasch Fono and Nudenberg. J . Org. Chem. 1951 16 113. Farmer and Sundralingam J . Chem. SOC. 1942 121. lo Cf. Bateman Gee Morris and Watson this Discussion 3 I4 GENERAL DISCUSSION Normally the relationship is obeyed ; in the presence of stearic acid this changes to roxidation = const. d([RO2HI2) . * (7) Reaction (3) has not been considered by Bell et al. and an indication of its existence in their work viz. the formation of methanol4 from CMe,O,D and Me radicals has been otherwise explained.However we believe the following experiment provides strong evidence for its occur-rence. Cyclohexene hydroperoxide was decomposed in the substituted I 4-diene ethyl linoleate (R’H). R’ is thus a mesomeric pentadienyl radical which if formed will react preferentially to give a conjugated I 3-diene derivative. In fact conjugated diene units were produced, they were located only in an ethyl hydroxy-linoleate and the formation of this substance was greatly in excess of that expected to result either directly or indirectly via radical-radical interaction. The alternative reaction R’- + ROZH -+ RO2- + R’H . . (8) has possibly been shown to occur by Bell et al. at 195’ in the gas phase reaction of Me and Et radicals with CMe,O,D. It should be noted how-ever that even in these systems the reaction is roughly 10 kcal.endo-thermic and this figure is increased to about 30 kcal. when R’H is an olefin and allylic resonance energy has to be supplied. At lower temper-atures and especially in unsaturated solvents (8) must be quite unim-portant compared with ( 3 ) which is about 30 kcal. exothermic (if Do-o = 40 kcal.). Reaction (3) is envisaged as a radical substitution reaction (SRz-to extend the Hughes-Ingold terminology) formally analogous to (I) cf. H j ( 7 . (-y H !n t-)! ui R’- + O-:-OR with RO,-I-H + 0-\--OR, where the arrows denote one-electron displacements. The corresponding ionic processes involving heterolytic bond scission can obviously be re-garded likewise. We do not suggest that the decomposition mechanism now advanced is necessarily applicable to other hydroperoxides under different con-ditions-it is quite evident that analogy can be especially misleading in this field.Thus the effect of increasing temperature may be particularly critical in facilitating a unimolecular S,I relative to an S,Z decomposi-t ‘on because the increasing tendency of the comparatively weak 0-0 bond to rupture will be roughly paralleled by a decrease in the molecular association in the hydroperoxide (evidenced by hydrogen bonding) which is undoubtedly precursory to reaction (I). Nevertheless we do suggest that the relevant conclusions of Bell et al. require careful reconsideration, in particular the postulation of unimolecular dissociation of tert.-butyl hydroperoxide in solution at temperatures not greatly above 100’ C.Not only are the rate data at 150’ C given in Table I1 of these authors’ paper consistent with our scheme (they correspond to a reaction order of 1-55), but Bolland and Morris 11 have made the very significant observation that this hydroperoxide in common with numerous primary secondary and tertiary allylic hydroperoxides catalyzes olefin oxidations in accordance with eqn. (7). Dr. W. E. Vaughan (EmeryvilZe California) (communicaded) The questions raised by Dr. Bateman and Mrs. Hughes concerning our paper seem to u s to be well answered by the experimental evidence presented, namely the isolation of methane-d and ethane-d and of ethyl tert.-butyl l1 See Quart. Rev. 1949 3 I GENERAL DISCUSSION 31 5 peroxide from the appropriate reactions.in substantial yields provide definite proof for the reactions These actual product isolations R' + (CH,),COOD -+ RD + (CH,),COO* . - (1) and R. + (CH,),COO' -f (CH,),COOR . ' (4 and they are not otherwise explained by the mechanism proposed by Dr. Bateman and Mrs. Hughes. It is our opinion that reaction (I) is not endothermic as claimed by Bateman and Hughes but rather is exothermic to the extent of about 20 kcal./mole when R is methyl. No quantitative evidence is available, but qualj tative comparison of ted-butyl hydroperoxide as a hydrogen atom donor with cyclohexene and formaldehyde leads us to the conclusion that Do-,(ROOH) is of the order of 80 kcal./mole rather than 105 or 118 kcal./mole as observed in certain alcohols and in water.Further proof for the chain decomposition we have proposed may be obtained from a comparison of the rates of decomposition a t 140~ of tert.-butyl hydroperoxide in a reactive solvent n-octane and in an inert medium chlorobenzene. In the latter solvent reaction (3) cannot intervene because decomposition leads to quantitative yields of tert.-butyl alcohol and oxygen. Moreover in this solvent the rate is some 40 times that in n-octane. Bateman and Hughes have acknowledged the difficulties in drawing analogies between decompositions of various peroxides under differing conditions. However the decomposition of cyclohexene hydroperoxide in an unsaturated solvent can also be interpreted by our mechanism which has the evidence of actual isolation of a stable peroxide intermediate.For ethyl linoleate this intermediate would correspond to (B) ; R'* + ROOH -f R'OH + RO' . * (3) R* + - CHdH-CH2-CHSH- -+ RH + -CH=CHL:CHLCHCCH-(A) (4) OOR (5) I ROO' + (A) -+ -CHdH-CHdH-CH-0. (B) (6) I (B) -+ KO' + -CH=CH-CHeCH-CH-OH (C) I (C) + ROOH (or RH) -+ ROO' (or R.) + -CH=CH-CH=CH-CH- (7) The allylic stabilization of (A) would increase the possibility for the postulated association reaction (2) at the expense of metathetical reactions such as (I) or ( 3 ) . The decomposition of cyclohexene hydroperoxide in cyclohexene can also be interpreted by a strictly analogous mechanism. However we think that in an inert solvent the reactions of the hydroperoxide group are greatly complicated by the presence of reactive methylene groups and the double bond in cyclohexene hydroperoxide.Although our in-formation offers nothing either pro or con concerning reaction (3) we see no objection to it on purely theoretical grounds and have in fact, indicated that the processes proposed in our paper do not exclude other possible competitive transformations which may be proceeding simul-taneously and undetected. Dr. W. A. Waters (Oxford) said Whilst we all realize the great value of the kinetic studies of olefin autoxidation carried out by members of the British Rubber Producers' Research Laboratory I would suggest that there will be little value in increasing the precision and detail o GENERAL DISCUSSION their present types of kinetic work beyond the stage which it has now reached.Kinetic studies of chain reactions give velocity equations which are dependent very much on one particular chain-ending process e.g. 2R' 3 products or R' + ROa' +- products (dimers often unspecified). When however organic chemists start looking for these diagnostic compounds in autoxidized mixtures they seldom find them and sometimes can show that the postulated dimers are far too reactive to persist in the final oxidation product. For example, in our current study of the effects of phenols on benzaldehyde autoxida-tion 12 Mr. C. Wickham-Jones and I deduced kinetically that the chain-termination product in mixtures containing +-cresol was a dimer of a mesomeric phenoxy radical (ArO*)2. We have failed as yet to isolate any of the known dimers of the tolyloxy radical from our reaction product and have found that the material which is eventually formed must be more complex.Moreover by adding the known dimers separately to our reacting system we have shown that they have a comparable reactivity to the p-cresol and so could not persist throughout the whole autoxidation. With 2 6-xylen-1-01 which gives similar kinetics the obtainable reaction product is not the phenol dimer but its further oxidation product. the corresponding diphenoquinone ; even this is an autoxidation retarder which therefore must be converted to still another product. One important point that kinetic investigators should remember is that the chain-terminating reaction which will be revealed by their reaction velocity measurements is the one which most rapidly removes active free radicals from the system.None of a whole series of slower radical-removing processes or of secondary reactions involving products which initially are present in very low concentration may be able to influence the reaction velocity to a discernible extent unless the experi-mental conditions are especially designed so as to detect one of them. " Initial velocity I' measurements in particular are of little help in the diagnosis of the chemical identities of products of chain termination. In reactions involving hydroperoxides it is important to remember that if the system becomes appreciably acidic or basic then an ionic decomposition of the hydroperoxide may set in to give products which are often quite different to that of its homolytic fission.The homolytic fission seems to proceed RlR2R3G-O-OH -+ RlR,R3C-O' + 'OH L RlR2C==0 + 'R3, where the fission rules for the breakdown of the ROO radical are in general those given by Walsh.lS The acid-base catalyzed reaction however, proceeds RlR2R3C-0-0-H + H+ -* where R is the group that most easily carries with it an electron pair. ways. These different products influence the subsequent reactions in different la J . Chem. SOC. 1951 812. l3 Trans. Faraday SOC. 1946 42 269 GENERAL DISCUSSION 317 Dr. A. J. Harding (Cambridge) said A good approximation to the kinetics of hydrocarbon oxidation may be derived without postulating intermediates of definite types such as peroxide or aldehyde. It is only necessary to assume that the chains consist of links which are alternately reacting with oxygen and with the hydrocarbon that the chains are terminated by destruction of the radicals which would otherwise react with hydrocarbon and that the chains are initiated by the breakdown of an intermediate which they themselves produce.The use of special features such as the ratio pmax/Apmax becomes essential if the finer points of the kinetics are to be evaluated. One of these points is the type of degenerate branching. Dr. Mulcahy has described the type in which the reaction fails to reach an infinite rate (i.e. ignition) because the re-actants are consumed before the intermediate concentration becomes sufficiently great. Another type of degenerate branching has been suggested for the oxidation of hydrocarbons at higher temperatures.In this second type the intermediate (e.g. aldehyde or peroxide) which generates chains is itself oxidized by the radicals i t produces. Since the production of intermediate is proportional to the first power of its concentration and the destruction is proportional to the second power a limiting concentration of intermediate is reached. This means that even if no reactants are consumed in the process the reaction attains a limiting finite velocity. A distinction between the two types of degenerate branching is readily made. In the first case addition of intermediate at the beginning of the reaction will give an increase of maximum rate as well as a decrease of induction period. In the second addition of intermediate will reduce the induction period but will produce no change in the reaction velocity until so much intermediate has been added that the induction period has been completely eliminated.The oxidation of ethylene at temperatures in the region of 400" C with formaldehyde as intermediate shows the behaviour expected of the second t41~e.l~ A derivation of the variation of pmax/Apmax with hydrocarbon con-centration for the second type of degenerate branching leads to a relation-ship approximately the same as that found by Dr. Mulcahy for the first type. This relationship does not therefore provide a clue to the type of degenerate branching although i t does give valuable support to the thesis that degenerate branching is taking place. l4 Dr. Peter Gray (Cambridge) (communicated) May the fact that the propylene oxidation studied by Mulcahy appears to occur through a peroxide which may react in such a way as not to produce active radicals be associated with the characteristics of propylene as a chain-breaking agent ? (In this case the constant C might include a term corresponding to the propylene concentration.) Dr.M. F. R. Mulcahy (Melbourne) (communicated) Dr. Gray's sug-gestion would require the chain-terminating effect of propylene to be specific towards the radicals produced immediately on decomposition of the peroxide. The participation of the hydrocarbon in a chain-terminating reaction in the main cycle would affect the second rather than the third term in eqn. (4) (tending to make B independent of [RH] under conditions of excess oxygen).However i t now appears that the significance of the value of C derived from the experimental results is somewhat complex. Some very recent experiments carried out in this laboratory by Mr. Ridge with propylene at 288" C have shown that the intercept of the pmax/Apmax against [RH] curve on the pmax/Apmax axis is dependent on the surface conditions and may become positive. It seems therefore that the simple interpretation of this intercept (C) given in my paper is in need of some amplifications. 1 4 Harding Thesis (Cambridge 1948). Norrish X V I Int. Coll. C.N.R.S. (Paris 1948) p. 16. Harding and Norrish (in preparation) 318 GENERAL DISCUSSION Dr. E. J. Harris (London) (communicated) In 1935 l5 i t was suggested that compounds known to promote the branched chain reaction in the slow oxidation of hydrocarbons such as ethyl nitrite and ethyl peroxide, did so by providing -OR radicals and the converse process a condensa-tion of -OR to peroxide seemed not unlikely.In 1938 Neumann and Tutakin l6 showed that small additions of diethyl peroxide to butane + oxygen mixtures would cause the appearance of a flame similar to the cool flame seen during the oxidation reaction. To put the hypothesis of the intervention of peroxide on a frrmer basis two approaches are possible namely the physical and chemical investiga-tion of the slow oxidation and the study of the peroxides themselves under appropriate conditions. In 1936 Sir Alfred Egerton suggested the latter method as a profitable one.Before mentioning a few relevant properties of peroxides i t will be useful to state the analytical evidence concerning peroxides isolated from oxidation reactions. At 320-270' propane and butane can be made to yleld considerable quantities of hydrogen peroxide making up 10 yo of the total condensate. The H,O combines with two molecules of form-aldehyde one of the other main products and in one experiment l7 0.5 g. of the compound was isolated. If however the reaction is carried out when the walls of the vessel are slightly contaminated with alkaline material the reaction no longer leads to the formation of the peroxide, though all the other products are nearly the same. This and other evidence suggested that the H,O,-formaldehyde compound was only formed after condensation.The fact that H,OB can be isolated indicates that radicals -OH or -0,H intervene in the slow oxidation the sensi-tivity to the state of the wall may mean that hydrogen peroxide is only formed when collisions involving the wall permit it. The analytical problem of showing the presence of an alkyl peroxide is of course com-plicated by the presence of a large amount of H202. Results of the study of the alkyl peroxides 18 19 2o and their influence on slow oxidation reactions are consistent with the following : (a) They may be formed in traces as by-products but chemical evidence is lacking. (b) The alkyl hydrogen peroxides like H,O, are very sensitive to the state of the surface being decomposed by traces of alkali. They will ignite giving a diffuse blue luminescence when introduced with oxygen into a vessel at 250-3ooo.The luminosity resembles that sometimes seen in the hydrocarbon oxidation and is general rather than flame-like. The dialkyl peroxides will decompose with a blue flame in certain conditions of temperature pressure and gas mixture. Below the critical limits they decompose homogeneously. The artificial blue flame does not inevitably promote hydrocarbon oxidation even when induced in hydrocarbon + oxygen mixtures. The result depends upon the surface and the temperature. ( d ) Traces of the peroxides reduce the induction period of the slow oxidation of hydrocarbons without altering the rate of the subse-quent reaction. This is explicable on the basis of radical formation. The small quantities of peroxide which are needed to produce visible effects make i t not surprising that they have not been isolated.Only 0.007 cm. of dipropyl peroxide will give a flash with air at 270° and 0-01 cm. 16 Egerton Smith and Ubbelohde PhiE. Trans. Roy. SOC. 1935 243 433. 16 Neumann and Tutakin Acta physicochim. 1938 9 861. l7 Harris and Egerton Chem. Rev. 1937 21 287. 1s Harris and Egerton Proc. Roy. SOC. A 1938 168 I. l9 Harris Proc. Roy. SOC. A 1939 173 126. 20 H a r k Proc. Roy. SOC. A 1940 175 254 GENERAL DISCUSSION 3 I 9 of ethyl hydrogen peroxide can be seen to luminesce when oxygen is present. In spite of the explosive properties of the peroxides the alkyl hydrogen peroxides like H,O,,21 can survive quite high temperatures if the vessel has a low surface/volume ratio and particularly if a carrier gas is present.Thus at 180' only 5 yo ethyl hydrogen peroxide was decomposed when passed with N through a tube at 180' (contact time 4 sec.). A t 320°, z yo survived. Under somewhat similar conditions ( I sec. contact time), 85 yo H,O sometimes survived. Post-war work on hydrocarbon oxidation (e.g. Hinshelwood 22 Mulcahy 23) appears to be in accord with the scheme outlined by Egerton and Harris 24 in 1938 in which i t was proposed that the initial step was peroxide forma-tion followed by splitting to give radicals -OR -OH and -O,H, which carry on the main reaction. Analogous schemes were outlined by Ubbel~hde,,~ Pease 28 and Lewis and v Elbe.,' Egerton and Harris, however did expressly suggest (cf. their Table 11) the reaction z -OH -f H,O as a chain-terminating one and this together with a reaction of -0,H with a hydrogen compound would explain hydrogen peroxide formation.Organic peroxides as such were only invoked in the initiation of the reaction and this agrees with the fact that surfaces active in decomposing the alkyl hydrogen peroxides (e.g. salt-coated ones, which under the experimental conditions become alkaline) are those which give rise to long induction periods for the slow oxidation reaction. Dr. G. J. Minkoff (ImperiaE College) said I should like to raise two points in connection with Dr. Mulcahy's paper. First with regard to the evidence of Badin which is quoted in support of the formation of hydroperoxides in oxidation Badin observed a line at 11.4 p in the infra-red spectrum of tert.-BuOOH; he concluded that since H,O also has a line at 11*4p this line is characteristic of the -0OH grouping.This is unsound for several reasons; in many observations of the infra-red spectrum of freshly prepared ten!.-BuOOH I have only once found a line at 11-4p and that was in an impure sample ; the position of the line was otherwise found to be at 11.6~. Di-tert.-butyl peroxide does have a line at 11*4p so that clearly the line cannot be characteristic of -0OH ; in fact as was pointed out before i t may be connected with both the -04- group and with the tevt.-butyl group. The line which Badin found may well have belonged to H,O,. The other point is connected with the attempt t o draw up a mathe-matical expression for the rate of pressure change.This is based on the proportionality observed between peroxide concentration and the rate of pressure change. In the derivation in terms of initial concentrations the tacit assumption is made that P is the hydroperoxide of the original hydrocarbon. However this assumption may not be correct because in the degradation of a long cha'n paraffin several steps must occur all prob-ably involving peroxides. In the calculations made by Bolland this ob-jection does not arise since only one peroxide is involved. Another complicating factor is that the different peroxides formed (i.e. primary, secondary peracids etc.) will react to different extents with the potassium iodide reagent; thus the maximum peroxide concentration may be masked by the lack of reaction with KI of some of the peroxide present.Dr. W. A. Waters (Oxford) (communicated) Sir Cyril Hinshelwood's review of the influence of substituents on the rate of hydrocarbon oxida-tion can be given an alternative interpretation which is equally consistent ar Harris Trans. Faraday Soc. 1.948 9 764. 2 Hinshelwood Faraday SOC. Dascussaons 1947. z8 Mulcahy Trans. Furuduy Soc. 1949. 45 575. ,4 Egerton and Harris Proc. I8me Congr. Chim. Ind. (Nancy 1938). 2s Ubbelohde and Egerton Proc. Boy. SOC. A 1935 152 354. 26Pease J . Amer. Chem. SOL 1929 51 1839 et seq. 27 Lewis and v. Elbe J . Amer. Chem. SOC. 1937 59 976 320 GENERAL DISCUSSION with the theories of general polarity. The substituents which he finds to increase oxidation rates are also those which promote the attack of methyl radicals on C-H groups.Kharasch and his colleagues for in-stance have reported that methyl radicals from the decomposition of diacetyl peroxide preferentially attack C-H bonds vicinal to C-C1, and also CH-CO- CH-0- CH-CO-OMe but do not attack CH, groups. From studies of tert.-butyl peroxide it now appears that R-0' radicals (e.g. Me,C-O.) have not quite the reactivity of alkyl radicals such as methyl and are even more selective in the same sense in their reactivities towards C-H groups.28 I would therefore suggest that the reaction which immediately follows Sir Cyril's chain initiation process R-0-0-X + R-0' + '0-X / / \ \ i.e. R-0' + H-C- -+ R-0-H + 'G-may be much slower and much more selective than the subsequent stages of the gas-phase oxidation involving R' and R-0-0.radicals. It may well be rate-determining for the whole breakdown of the oxidized compound. A similar state of affairs is well known in polymerization chemistry where the sequence, (i) Catalyst (eg. benzoyl peroxide) -+ Radicals R' ; (ii) R' + Monomer -+ Hydrocarbon type radical R-M' ; (iii) R-M + M 3 R-M-M' etc. (fast chain growth), results in a polymerization kinetically dependent upon the rates of both processes (i) and (ii) as for instance in the benzoyl peroxide catalyzed polymerizations of both styrene and vinyl acetate. In polymer chemistry again the stability of CH in comparison with A H z - is strongly marked as for instance in the ready chain transfer to -CHz-CH=CH- and the converse tendency of compounds with CH,-C=C groups to build up high polymers.In connection with later states of hydrocarbon oxidation i t may be noted that in the Dyson Perrins Laboratory we have recently shown that in the liquid phase the breakdown occurs quite readily a t temperatures as low as SO-IOO~ C. This is con-firmatory evidence for the view that in hydrocarbon oxidation at elevated temperatures only the initial stages of the reaction sequence are rate-determining. Dr. A. D. Walsh (Leeds) said It is helpful in considering the kinetics of oxidation reactions of hydrocarbons in the gas phase to write the step that produces the alkyl peroxide radical in the form Commonly the * is omitted but its inclusion serves to remind us that in the first instance the RO radical is energy-rich.This is important for at least the following reasons. (i) It makes it less likely that the reverse reaction to (I) will be neglected in considering the full kinetic scheme. (ii) The excess energy is presumably distributed over various vibrational degrees of freedom. The radical is therefore to be thought of as vigorously twisting turning and generally distorting itself far more than does a " normal " radical. This means that subsequent reactions which axe rather more complicated than are usually found in gas-phase oxidation processes are rendered more pIausible than they would otherwise be. Cf. also Farmer and Moore J . Ckem. Soc. 1951 131. Alkyl-CO' + AllrJtl' + CO R + 0 = RO,*. (1 GENERAL DISCUSSION 321 Those gas phase chain reaction steps which are best substantiated ( a ) those in which one linkage is broken e.g.( b ) those in which one linkage is formed e.g. (I) above the reverse commonly' fall into one of 4 classes 29 : CH,CH3 = CH + CH - (2) CH + NO = CH,NO* . * (3) of (2) and (G) those in which one linkage is formed and one is broken e.g., abstraction of an atom (particularly H) from a molecule by a radical as in ( d ) energy transfer reactions e.g. reactions involved in the quenching OH + RH = H,O + R; * (4) A * + B = A + B * . . - ( 5 ) of fluorescence : As illustrations of these four types it is instructive to look at the paper by Bell Raley Rust Seubold and Vaughan. The great merit of that paper is that some experimental evidence for each step postulated has been obtained.Consideration of these steps shows that they all con-form (or in one case can readily be slightly altered to conform) to the above types. In other words the steps occurring most commonly appear to be the simplest. This should not be surprising for in postulating a reaction step, i t is vital to consider just how the reactants could be converted into the supposed products; and such consideration makes it difficult not to conclude that more complicated steps than (a) t o (a) are likely to have unfavourable steric factors or activation energies.30 If a postulated gas-phase reaction step does not conform to one of the 4 types one cannot say it is impossible-but one can and should demand that the evidence for it be unusually strong unless it is to be dismissed as mere speculation.This is incidentally a serious criticism of many of the reaction steps postulated by certain authors. To take but one example the supposed reaction RCH,CHO + 0 = RCHO + HCHO involves the simultaneous breaking of two linkages and the forming of two linkages. Unless particularly strong evidence in support of this was forthcoming-and such has certainly not yet been produced-this postulated step can be taken as very unplausible. In the special case of a reaction such as (I) however i t is not unplaus-ible to suppose a subsequent reaction for RO,* more complicated than types (a) to (d). If this is re-written as it is split into simpler steps and the fact that the second involves the breaking of two linkages and the forming of one does not appear un-plausible.Another example might be a somewhat complicated isomer-ization of RO,*. (iii) Even if the reaction supposed subsequent to (I) conforms to (a) to (d) inserting the * in (I) helps one to remember that the activation energy' for the following reaction need not be as great as would be the case for " normal " RO,. Some or all of the activation energy may be provided by the energy produced by (I). An example is the reaction CH + 0 = HCHO + OH. CH3 + 0 = [C,HO,"] = HCHO + OH. The step RO + RH = ROOH + R - (6) 29 Cf. Chamberlain and Walsh Rev. Inst. Frangais du Pe'trole 1949 4 307. 3 0 Cf. Ubbelohde Rev. Inst. Frangais du Pe'trole 1949 4 315. 31 Norrish this Discussion. 322 GENERAL DISCUSSION although probably exothermic when RH is an olefin is probably endo-thermic when RH is a paraffin.32 The activation energy required for (6) in gas phase reactions however need not be as great as this endothermicity.The energy of formation of RO, carried over into a subsequent reaction reduces any improbability of that subsequent step. To realize this is to recognize that the process of gas-phase peroxide formation may’ be subject to inhibition by inert gases. The greater the dilution of the reactants the greater the chance that collisions (and reactions of type (a)) will take so much energy from RO,* that the probability of (6) is seriously reduced. Peroxide formation is involved in cool flame phenomena. To test whether inhibition by inert gases of cool-flame formation can occur however one needs to choose conditions carefully’.At temperatures near the lower temperature limit of spontaneous cool-flame formation the dominant chain-ending reaction is usually a surface destruction of radicals and the inhibition may be swamped by a greater promotion viz. the hindering of radical diffusion to the walls. Along the upper temperature limit of cool flame formation, however the dominant chain-ending reaction is a gas phase one (in most cases the decomposition of a bulky alkyl radical to an olefin and a small alkyl radical 33 3ii). At such temperatures for diethyl ether (Fig. 3 of ref. (33)) and for propane (Fig. 11 of ref. (35)) a small inhibition by’ inert gases is observed. It seems probable that this inhibition represents the expected effect of reduction of the energy content of RO,* though of course i t has to be remembered that cool-flame formation3(j and pro-pagationa7 involve a thermal factor and inert gases may also inhibit purely by virtue of their thermal properties.Finally i t is important to be clear about the nature of the excess energy carried by RO,*. For the analogous HO,* radical however the following reaction has been supposed to occur 38 The idea is closely similar to that of (ii). This is surely vibrational in nature. HO,* + NO = EIO + NO -/- 0, the HO,* radical transferring the whole of its energy to the NO,. Such a complete transfer of energy seems only likely for electronic energy. Yet the effects of inert gases (M) on the second limit of the €-I + 0 ex-plosion are usually supposed to require the transfer of energy from HO,* (i.e.H + 0,) to M ; and the different gases that are effective make i t difficult to suppose other than vibrational energy is concerned. Some confusion seems therefore to exist. Dr. C. A. McDowell (Liverpool) (communicated) I was interested to read Dr. Walsh’s remarks on the existence of RO,* radicals i.e. RO, radicals endowed with excess energy. Similar views were expressed by Dr. Thomas and myself in our paper on the inhibitory effect of nitrogen peroxide on the gas-phase oxidation of acetaldehyde. 39 In considering possible subsequent reactions of RO,* radicals one has to be extremely careful for the excess vibrational energy with which these radicals are endowed may or may not persist throughout the numerous collisions which these molecules may make before they undergo reaction.If the excess of vibrational energy does not persist then ob-viously little is to be gained by ma’ntaining the above notation. Numerous 32 Walsh J . Chem. SOC. 1948 339. 33 Chamberlain and Walsh 3rd S y m p . Combustion Flame and Explosion Phe-34 Walsh in course of publication. 35 Malherbe and Walsh Trans. Favaday SOC. 1950 46 835. 36 Malherbe and Walsh Trans. Faraday SOC. 1950 46 824. 37 Spence and Townend 3rd S y m p . Combustion Flame and Explosion Phe-38 Dainton and Norrish Proc. Roy. Soc. A 1941. 177 395. 3% McDowell and Thomas J . Chem. Soc. 1950 1462. nomena (Williams and Wilkins Baltimore 1949). nomena (Williams and Wilkins Baltimore 1949) GENERAL DISCUSSION 323 cases are however known where vibrational energy does persist through-out a time interval during which a molecule may make as many as 5 x 104 collisions before one quantum of vibrational energy is dissipated.Theoretically there is no reason why vibrational energy should not be transferred from one molecule to another provided their vibrational levels are sufficiently close together to permit resonance to occur. Such cases are well known and it should perhaps be pointed out that it is not necessary that all the excess vibrational energy should be transferred from one molecule to another ; in fact it seems probable from elementary theoretical considerations that the excess energy is more likely to be transferred in small amounts in successive quanta. One other type of transfer which must be borne in mind is the transfer of vibrational energy from one molecule into translational energy of another.This is most easily under-stood in the case of the transfer of vibrational energy from vibrationally excited molecules to rare gas atoms. This latter type of transfer probably also occurs with light molecules. Prof. R. G. W. Norrish. (Cambridge) (communicated) The reaction RCH,CHO + O=RCHO + H,CO was originally postulated as a source of excited formaldehyde to account for the luminosity of the cool flame. In view of the very low luminosity observed (ca. I quantum per 10 mole-cules of hydrocarbon reacting) its probability in comparison with the other radical reactions would be very low in accordance with Dr. Walsh’s expressed view. However I have as will be observed in my paper, dissociated myself from this reaction except as a possible component in the dark blue flame and even then the suggestion must not be taken too literally for the reaction may occur in two stages : RCH,CHO + 0 = RCH CHO + OH RCH .CHO + OH = RCHO + H,CO. In any case I am not fully able to agree with limitations which Dr. Walsh would lay down for chain processes. Not enough is yet known about the possible configurations in the transition state. They conform to the views we have already expressed with reference to HO ; see for example the footnote in my paper or Axford and Norrish on the oxidation of formal-deh~de.4~ Prof. A. R. Ubbelohde and Mr. Small (Belfast) (communicated) : The suggestion that RO,* is in the first instance “ energy-rich ” and is more effective in reaction ( 2 ) above if i t can use this energy before i t is randomized by collisions etc.has an interesting experimental corollary. It has recently been observed 41 that molecular hydrogen inhibits certain reaction chains in both aldehyde and hydrocarbon oxidations much more markedly than molecular nitrogen. Tentatively this may be attributed to the efficiency with which hydrogen can effect the transfer between internal and translational molecular energy in collisions. I agree with his remarks about RO,*. RO,* + H + H,* + RO,. In (pseudo) unimolecular reactions the special efficiency of hydrogen collisions in the reverse process translational + internal is well known, and from the principle of microscopic reversibility collision process (3) would also be expected.Dr. A. D. Walsh (Leeds) (communicated) Prof. Sir Cyril Hinshelwood refers to my suggestion that bonds between strongly electronegative elements should be strengthened by electron-repelling groups. As originally put forward this was based upon the possibility of increas-ing the overlap of the two atomic orbitals concerned in a single bond 40 Axford and Norrish Proc. Roy. SOC. A 1948 192 518. 41 Small and Ubbelohde J . Chem. Soc. 1950 723 and unpublished results. L 324 GENERAL DISCUSSION between the elements. It is perhaps worth pointing out however that the suggestion could also be based upon the possibility of reducing lone pair-lone pair repulsion between the 0 atoms by attaching electron-repelling groups.The interaction between lone pair electrons on a halogen atom and electrons of a neighbouring system is greatest for the most electro-negative halogen (F) and least for I.43 A similar statement applies to interaction involving lone pair electrons on a Group VI eIeme11t.4~ The interaction between lone pairs on the adjacent 0 atoms of an -0-O-group is expected to be repulsive (witness e.g. the skew nature of the H,O, molecule). Attaching electron-repelling groups to an 0-0 group will reduce the effective electronegativity’ of each 0 atom and so be expected to reduce the lone pair-lone pair repulsion i.e. be expected to increase the 0-0 bond strength. Dr. N. Uri (Manchester) said In the primary step relating to the oxidation of hydrocarbons as postulated by’ Prof.Norrish viz., RH + 0 -+ R + HO, the endothermicity and activation energy are likely to be of the order of 40-70 kcal. if the dissociation energy of the HO radical into H + O Z ( D ~ o 2 ) is as low as 36 kcal. This latter value is obtained 4 4 9 45 from the electron affinity (in solution) of the HO radical which is in turn evaluated by inter-polation from the energies corresponding to the absorption maxima of various ferric ion pairs. While some unrecognized error may cause our quoted value of D H O ~ to be a few kcal. low it would be difficult to account from hydrocarbon chemistry for a value of D H O ~ as high as 65 k~a1.,4~ unless i t also assumed that the activation energy of the first step of the oxidation of hydrocarbons is much larger than its endothermicity.Dr. A. J. B. Robertson (King’s College London) said The heat evolved in the formation of HO may be determined by the method outlined by’ Stevenson in this Discussion. I find that hydrogen peroxide is dissoci-ated by electron impact to give the HO,+ ion with a small probability. The appearance potential is 16.1 eV. The ionization potential of H,O, is found to be 12.1 elr. The ionization of O, HO and H,O very prob-ably involves the removal of a weakly anti-bonding or non-bonding electron located approximately on an oxygen atom. In these circum-stances we may provisionally estimate the ionization potential of the HO, free radical as 12.2 eV intermediate between that of 0 and H,O, and i t may be less. This assumption gives 46 kcal. for the heat evolved in the formation of HO from H and O, subject to an experimental uncer-tainty a t present of about g kcal.If the ionization potential of HO is lower than 12-2 eV the formation of HO is even less exothermic. Dr. R. Spence (Harwell) said Prof. Norrish has raised many inter-esting points. I have always thought that the phenomena of com-bustion are so numerous and varied that general mechanisms should not be pressed too far unless all the well-established facts have been taken into account. For instance I do not think that the isolation of aldehydes as products of the reaction is necessarily evidence that they occur as essential intermediates. This was one of the important lessons to be learned from the early work of Bone who tried to establish the hydroxylation theory of combustion by the identification and separation of a set of products which were regarded as intermediates.Formalde-hyde for instance is a product of the oxidation of acetylene at 3 o 0 - 3 ~ 0 ~ C.4’ It is known that its rate of reaction with oxygen at these temperatures la See Baker and Hopkins J . Chem. SOC. 1949 1089. 43 Unpublished work of Baker and Barrett. 44 Evans and Uri Trans. Faraday SOC. 1949 45 224. 45 Evans Hush and Uri (in the course of publication). 46 Walsh J . Chem. SOC. 1948 331. 47 Spence and Kjstiakowsky J . Amer. Chem. SOC. 1930 52 4846. 48 Spence J. Chem. SOC. 1936 652 GENERAL DISCUSSION 32 5 is quite slow even at moderate partial pressures. Thus ordinary formaldehyde cannot be an intermediate in the combustion of acetylene ; i t can only be an intermediate if i t reacts whilst still in the excited state and i t is important as Dr.Walsh has mentioned earlier to recognize this in the mechanism. Then as regards the OH radical I have referred earlier in this dis-cussion to the work of Dr. Bates and myself 49 on the reaction between methyl radicals and oxygen and to the possibility of the formation of OH radicals or of CH,O in the primary step. More recently Blaedel Ogg and Leighton 5 O failed to detect the OH radical in this reaction by optical methods so that if present its concentration must be very small. Another piece of evidence which is not easy to fit into a simple radical chain theory was obtained by my wife 51 when working in Prof. Townend's laboratory. The propagation of a cool flame through a vertical tube at room temperature containing acetaldehyde ether hexane or heptane and oxygen or air is determined purely by thermal considerations.Free radical chain effects usually associated with the vessel wall and with the addition of inert gases were not observed. The addition of methyl radicals or of iodine produced no measurable effect ; the only indication that a chain mechanism might be operating was the inhibitory effect of additions of NO,. Another experimental result which is not easy to reconcile with a mechanism involving oxygen atoms is the homogeneous slow combustion of formaldehyde in reaction vessels of only I mm. diam.48 In this case, the chain carriers must be relatively unaffected by the wall. The surface oxidation only predominates when the vessel is packed with powdered glass.The " dark blue " flame mentioned by Prof. Norrish is to be identified with the " blue" flame of Townend and co-workers 51 which follows in the path of the " cool " flame. Prof. R. G. W. Norrish (Cambridge) (communicated) There can be no question that aldehydes are formed as intermediates in hydrocarbon oxidation because they are readily observed the question a t issue is whether they are to be identified as the moderately stable intermediate Iesponsible for the delayed branching. I have given the reasons in my paper for supposing that they are particularly in the second paragraph : I do not think that more weighty evidence has yet been produced for any other intermediate product playing this role.It must be remembered that the process of degenerate branching also takes place at temperatures where peroxides are completely unstable and unobservable and that Semenov's hypothesis demands a finite measureable concentration. The possible origin of aldehydes via a transitional peroxide is not ruled out on this account e.g., RCH + 0 -+ [RCH,O] + RCHO + OH as I have indicated in my paper. I do not feel that the remarks about the oxidation of acetylene are relevant. My paper did not concern this reaction but i t may be men-tioned that the rate of oxidation of formaldehyde at 286" C is appreciable (for partial pressures of formaldehyde and oxygen of IOO mm. in a vessel of 23-6 mm. diam. and volume 80 cm. the rate of reaction as measured by pressure change is 1-9 mm./rnir~.).~ Further i t is highly probable that the formaldehyde product in the acetylene oxidation is excited and thus more highly reactive.However I do not wish to commit myself to any view concerning acetylene oxidation in the present remarks. 49 Bates and Spence J . Amer. Chem. SOC. 1932 53 1689. 50 Blaedel Ogg and Leighton J . Amer. Chem. SOC. 1942 64 2499. 51 Kate Spence Thesis (Leeds 1945). 52 Axford and Norrish Proc. Roy. SOC. A 1948 192 518 326 GENERAL DISCUSSION As shown by Dr. Reagh and myself,5s the surface effect in the slow oxidation of hydrocarbons only becomes apparent at a certain limiting diameter of about 5-10 mm. The theory of slow branching requires that the reaction shall be entirely inhibited a t a finite diameter when the net branching factor becomes less than zero by the increase of the surface deactivation.This was found to be the case for methane ethane propane, acetylene ethylene and propylene. We should not expect the cool flame to be affected until a limiting diameter of tube was reached. Finally, with regard to the oxidation of formaldehyde this shows none of the criteria of a degenerate branched reaction and the results have been explained by us in terns of a straight chain mechanism. We should not expect the dramatic results observed with hydrocarbons on decreasing the diameter. There is however a competition in chain ending between the volume and surface reactions in which the surface effect only becomes predominant a t very small diameters ( < I mm.) corresponding to packed vessels.Nothing here appears to me to be inconsistent with our inter-pretation of the oxidation of formaldehyde as a straight chain reaction. With regard to the “ dark blue ” flame i t will be seen that I have drawn attention to the similar flames in acetaldehyde observed by Townend and his co-workers in my paper. Dr. G . Porter (Cambridge) (communicated) As far as I am aware, no short-lived free radical has ever been detected by its absorption spectrum in a chemical reaction at normal temperatures nor in an ordinary photo-chemical reaction despite numerous attempts with systems where radical mechanisms are known to be operative. I think it important to bear this in mind when assessing negative evidence of the kind mentioned by Dr. Spence though the workers quoted used a particularly sensitive method and their results certainly suggest a short lifetime of I O - ~ sec.or less of the OH radical if it was present. By using the method of flash photolpsis54 very high instantaneous concentrations of intermediates can be obtained and I have been able to observe the OH radical in absorption at high intensity in two systems which almost certainly involve the reaction between methyl and oxygen, viz. the photochemical oxidation of acetone and the chlorine photo-sensitized oxidation of methane. Unfortunately even this positive result does not enable one to conclude that OH is formed in the primary reaction of methyl with oxygen rather than in the subsequent steps though i t is hoped that studies of the kinetics of OH appearance will make this dis-tinction possible.Dr. L. Bateman (Welzuyn Garden C i t y ) (communicated) Norrish’s contention that peroxides do not absorb light and are therefore photo-chemically inert in the wavelength range 3000-4000 A is misleading and his arguments as presented have not the force claimed. Admittedly such absorption is weak E of typical saturated and allylic hydro- and di-peroxides ranging from about 0.5-10 at 3000 to 0-01-1 at 3650 A but Gee and I have presented quantitative Fvidence 55 that the very strong catalytic action of irradiation a t 3650 A on the autoxidation of liquid olefins originates in the photolysis of the hydroperoxide. This catalysis certainly extends to wavelengths as high as 4000 A and possibly higher.56 The quantum efficiency of the photolysis at 3650 is of the order of 0.1.However simple carbonyl compounds are even less absorbing in this spectral region (for acetone E~~~~ N o - o o ~ ) while photolysis is of com-parable efficiency to the peroxide (Norrish Crane and Saltmarsh 5 7 report + = 0-2 at 3150 A). 53 Norrish and Reagh Proc. Roy. SOC. A 1940 176 429. 54 Porter Proc. Roy. SOC. A 1950 200 284. 55 Bateman and Gee Proc. Boy. SOC. A 1948 195 376. 56 Bateman Trans. Faraday Soc. 1946 42 266. 57 Norrish Crane and Saltmarsh J . CAem. SUG 1934 1456 GENERAL DISCUSSION 327 Prof. R. G. W. Norrish (Cambridge) (communicated) I cannot sub-scribe to the cogency of Dr. Bateman’s criticism with acetaldehyde for example at 3400 the extinction coefficient is 1.0 whereas for tert.-butyl hydroperoxide and also for hydrogen peroxide the extinction coefficient is 0.1.In addition formaldehyde has an absorption coefficient of 0.07 a t 3560 while the peroxides have fallen to negligible values. I would point out that the peroxide with which he was concerned namely cyclo-hexene hydr~peroxide,~~ is one that would be expected to have consider-able absorption owing to the unsaturated nature of the compound. The combined effect of the double bond and the peroxide group will un-doubtedly cause light to be absorbed more strongly at longer wavelengths than in the compound where the peroxide group is the only chromophore. Add to this the fact that the intermediate concentration of aldehydes in the oxidation of methane and ethylene and in higher hydrocarbons at the higher oxidation temperatures is of the order of millimetres while the concentration of peroxides is undetectable analytically and I think i t must be agreed that the photochemical effect is to be ascribed to the aldehyde.In addition Mr. Booth in this laboratory has recently been studying the photolysis of tert.-butyl peroxide using a solution of 2 yo by volume in n-hexane. He finds no photolysis whatsoever when all wave-lengths below 3350 are filtered out. but considerable decomposition when only those wavelengths above 3000 Dr. C. F. Cullis (Imperial College) (communicated) Prof. Norrish has suggested that the large structural effect encountered in the oxida-tion of the normal paraffins is simply attributable to the greater number of points of attack in the longer molecules Thus he argues that n-octane is more readily oxidized than n-pentane since there are more CH groups in the former molecule and the removal of hydrogen atoms from such groups is considerably easier than from terminal methyl groups.On this view the increase in oxidation rate with chain length should be roughly proportional to the number of CH2 groups whereas in fact the variation is much more marked.59 In any case an extension of this argu-ment fails to explain qualitatively the facts relating to the oxidation of branched chain paraffins. It is well known in hydrocarbon chemistry that the order of reactivity of C-H bonds towards radical attack decreases in the order 3” > zo > IO, and Rice has shown that at 3ooOC the chances of attack at primary, secondary and tertiary carbon atoms are in the ratio I 3 33.60 For the isomeric hexanes for example i t is possible to calculate the relative probabilities of hydrogen atom removal from the molecules concerned by multiplying the numbers of such atoms attached to primary secondary and tertiary carbon atoms by Rice’s factors and summing over the whole molecule.If the ease of removal of a hydrogen atom from a hydrocarbon molecule is the main factor controlling the ease of oxidation as Prof. Norrish’s argument implies the figures in the final column but one of the following Table should be measures of the oxidizability of the compounds concerned. On this basis therefore the order of ease of combustion of the isomeric hexanes would be V > I1 = I11 > I > IV whereas the experimentally determined relative oxidation rates (given in the final column of the Table) show that the order is I > I1 > I11 > IV > V.59 Prof.R. G. W. Norrish (Cawbridge) (communicated) My remarks about the relative ease of oxidation if hydrocarbons are intended to indicate that other things being equal the ease of oxidation per carbon atom will increase from methane and ethane which will have values are removed. 58 Bateman and Gee PYOC. Roy. SOC. A 1948 195 376. Cullis and Hinshelwood Faraday Soc. Discussions 1947 2 I 17 ; Cullis and Mulcahy Revue l’lnst. Franpais Pe‘trole 1949 4 283. 60 Rice and Rice The Aliphatic Free Radicals (Johns Hopkins I935) p. 100 328 GENERAL DISCUSSION characteristic of the CH,.group to a value characteristic of the CH group for an infinitely long chain.Such comparisons as are made by Dr. Cullis in his criticism of my statement appear to me to have little relevance because the bases of comparison are so doubtful. Firstly his comparison of the rates of oxidation of hydrocarbons is based on the rates of pressure changes the reactants being at comparable pressures in all cases. This does not take account of the great increase in " carbon concentration " as we pass to higher hydrocarbons which would have the effect of weighting his results strongly in favour of the higher hydrocarbons. Secondly the comparisons could not all be made at the same temperature and the accumulation of intermediate products greatly' depends on the tem-perature as well as the nature of the hydrocarbon.For example in TABLE I1 Compound CH . CH . CH . CH . CH . CH (I) CH, I (11) (111) (IV) CH,. C H . CH . CH . CH, CH I CH,. CH,. CH . CH,. CH, CH, I I CH, I I CH, C . CH,. CH, CH,CH, CH,. CH . C H . CH (V) Relative Probabilities of H Atom Removal From IO C Atoms 4 x 1 9 x 1 9 x 1 I 2 x I I 2 x I From 2' C Atoms 8 x 3 4 x 3 4 x 3 2 x 3 -From 3" Z A toms Total telative Proba-ilitr of I Atom :emoval 30 54 54 18 78 telative Oxida-tion Rate 1580 560 60 I 2 I one experiment of Dr. Patnaik the oxidation of butane at 270' C using a butane oxygen ratio of 1-5 and a total pressure of butane of 76.1 mm., intermediate products such as aldehydes and alcohols accounted for over 25 mm.in a total pressure change of 60 mm. It is not legitimate to com-pare the rates of oxidation of hydrocarbons by pressure changes in view of this great variability of end-products and the only certain way is to measure the rate of disappearance of hydrocarbon by analytical technique. This has not yet been done. Finally Rice's figures refer to the relative reactivity of CH, CH and CH respectively with alkyl groups whereas in comparing the rates of oxidation one is concerned with their reactivity towards oxygen or OH. There is no reason why there should be any close quantitative parallelism in the two cases which will undoubtedly be differently conditioned by temperature and steric hindrance. Such calculations as have been made on this basis can have little quantitative value until much more data have been accumulated and until some more uniform basis of comparison of reaction rates has been devised.Dr. D. W. G. Style (London) said When illuminated with Schumann ultra-violet light diethyl peroxide and ethyl nitrate emits as a fluorescence the same band system which extends from approximately 3300 %L to 5000 A. The emiss:ons from methyl chloroformate and methyl nitrite are also identical but different from that from the first two substances GENERAL DISCUSSION 3 29 I t is probable that the emitters of the two band systems are respectively C,H,O. and CH,O-. Have these spectra been observed with other sources, particularly cool flames ? Dr. Peter Gray (Cambridge) (communicated) It is very interesting to see the spectra of molecules containing the alkoxyl group shown by Dr.Style and to learn that the emitter is likely to be the alkoxy radical itself. It is hoped soon to examine a system in which the methoxy (CH,O-) group thermally produced may be an emitter viz. the lumin-escent thermal decomposition of methyl nitrate. The initial decom-position of methyl nitrate gives nitrogen dioxide and methoxy groups and is accompanied by a blue luminescence. The same behaviour is displayed by methyl nitrite. The light may be due either to CH,O-or to excited formaldehyde CH,O* and now both spectra are known i t may be possible to identify the source of this luminescence. Sir Alfred Egerton (Imperial College) (communicated) The emission bands which Dr.Styles has found and ascribed to the C,H,O radical in the spectrum of glow in diethylperoxide are interesting. When Dr. Harris was working on the decomposition of diethylperoxide in my laboratory' we looked for bands in the glow but a t that time had not the means for photographs of long exposure. It is possible these bands have some relation to the pronounced absorption bands obtained in the com-bustion of hydrocarbons which Pidgeon and I reported in 1933. Dr. C. A. McDowell (Liverpool Univevsity) (communicated) With re-gard to the oxidation of acetaldehyde Dr. Thomas Mr. Farmer and I have shown fairly conclusively by inhibition studies G 1 ~ 6 2 that the thermal reaction u p to about 140'C is a chain reaction in which two radicals, the acetyl and peracetyl radicals play the predominant part.Dr. G. J. Minkoff (Imperial College) said The heats of combustion presented in our paper clearly have some peculiarities since the values of (-0-0-) which can be derived are of the order of 100-140 kcal. We intend to repeat thf measurements a t some future time but have mentioned the results already obtained as the work was done most care-fully and as the values are self-consistent within themselves (cf. Qc for members of series and for isomers). We have also recorded the refractive indices and melting points which we obtained so as to have as many published data on this subject as possible. These should be of use in determining the purities of future samples though we would stress the great advantages of determining the infra-red spectra of the peroxides during the preparations ; as the standard methods of purification (e.g.fractional distillation) are often inapplicable i t is a great help to know what particular impurities have to be removed. A brief point is that the latent heats of evaporation show high Trouton coefficients for all the hydroperoxides; this agrees with infra-red studies of the OH fre-quencies in suggesting that the compounds are strongly associated. DeteIminations of the heat of formation and of D(-0-0-) should thus if possible make allowance for the heat of dimerization etc., involved. Finally I should like to explain the use of the terms " acetyl, propionyl and butyryl hydroperoxides " to describe the corresponding peracids. Mr. Everett and I have measured the (classical) dissociation constants in aqueous solutions of these compounds and of H,O, Me0,H and EtO,H and have found that the peracids are intermediate in acid strength between the corresponding hydroperoxides and carboxylic acids.In fact the increase over the hydroperoxide can be shown to be approx-imately accounted for by inductive and polarizability effects. We there-fore suggest naming the peracids " acyl hydroperoxides " to avoid the implication of a great acid strength sometimes inferred from the term McDowell and Thomas J . Chem. SOC. 1949,2208 2217 ; 1950 1462 ; Trans. Faraday SOC. 1950 46 1030. G 2 McDowell and Farmer unpublished work 3 30 GENERAL DISCUSSION peracid (cf. perchloric and permanganic acids). Further if the diacyl peroxides are referred to as such there will be less chance of confusion with the corresponding acyl hydroperoxides (cf.dibenzyl peroxide benzoyl peroxide and perbenzoic acid). Dr. W. E. Vaughan (Emeryville California) (communicated) In the course of our own studies on organic peroxides we have obtained certain physical data on tert.-butyl hydroperoxide and di-tert.-butyl peroxide which differ from those just presented. Although we believe our com-pounds were better than gg yo pure we do not wish to claim superior accuracy for our data but wish rather to present them for the record in the hope that the differences may be resolved in the future. +alp . B.p. "C . Vapour pressure . K p . "C . Heat of vaporization . . Heat of combustion . tert.-Butyl hydroperoxide 1.4010 5'5 I 15-0 log10 P (mm.1 I 1-5 kcal./mole 65 4.2 kcal./mole Di-lert.-butyl Penoxide 1.3890 - 40.0 I 11'0 9.6 kcal. /mole 12 75.0 kcal. /mole Dr. M. Magat (Paris) said Leadbeater 63 has recently prepared in our laboratory some carefully purified peroxides and measured their refractive indices and Raman spectra. The following results were ob-tained : Diethyl peroxide nb5 = 1.3720 f 0-0005 Di-a-hydroxydiethyl peroxide nh6 = 1.4265 Ethyl-a-hydroxyhydroperoxide n:4 = 1.4150 f 0-0010. For the diethyl peroxide the agreement with the refractive index given by Egerton et al. is excellent. The characteristic Raman frequency for the 0-0 bond found in all peroxides investigated by Leadbeater as well as in dibenzylperoxide and in H,O is located at 880 -& 3 cm.-l.I would like also to call attention to the danger of working with very pure crystallized peroxides even with those classified as relatively stable : a spontaneous or induced rupture of the crystal has led to an extremely serious explosion when monohydroxydiethyl peroxide was purified by' crystallization. Dr. N. S . Wooding (Courtaulds Ltd. Coventry) (communicated) : Cobalt acetate in concentrations of IO- to 10-3 M has been shown to be an effective catalyst for the oxidation of trimethylethylene in solution by gaseous oxygen. Recently i t has been found that the autoxidation of cellulose under alkaline conditions was also catalyzed by cobalt acetate. The effect of other metal ions was investigated and some were found to catalyze while others retarded the autoxidation.Such behaviour has been reported elsewhere.65 However manganese salts were found to behave both as catalysts and as retarders depending upon the concentra-tion of salt used. A possible explanation of this effect has been suggested in terms of the mechanism of metal ion catalysis postulated in this paper and else~here.~42 O 5 I would like to ask Prof. Eawn if he has found any evidence for negative catalysis by metallic cations in the autoxidation 0.0005 63 Leadbeater Cow@. rend. 1950 230 829. 64 Entwistle Cole and ?Tooding Textile Res. J. 1948 19 527 609. 65 George and Robertson Tram. Faraday SOC. 1946 42 217 GENERAL DISCUSSION First Order Constant @in.-1) 33 1 Time of Half Cnange (-4 of trimethylethylene or any of the other systems he has investigated, since such phenomena if the suggested explanation is correct should be observable in other systems under the appropriate conditions.Prof. C. E. H. Bawn (Liverpool) (communicated) Under the experi-mental conditions so far studied the condition in which the rate of oxida-tion is independent of the catalyst concentration as observed by George and Robertson with saturated hydrocarbons has not been attained. There is no evidence at present that the catalyst terminates chains by a process such as RO + Co++ -+ R0,- + Co+++. Dr. C. F. H. Tipper (Edinburgh) said It is nearly always assumed when considering the oxidation of hydrocarbons in solution that the hydroperoxides initially formed decompose to give free radicals.However, under certain conditions for example in solvents of high dielectric constant or especially in the presence of acids as Dr. Waters has stressed, heterolytic fission of the 0-0 bond may occur to give ions. The evidence for this has so far been mainly' organic,66 and so I would like to report the results of some kinetic measurements on the decomposition of decalin hydroperoxide in various solvents. The overall decomposition was found to be first order in all cases, at any rate at low concentrations ( < I O - ~ mole/l.). The rates at 130' C and overall activation energies of decomposition in different solvents are shown below. TABLE I11 Chlorobenzene . . o-Dichlorobenzene . Acetic acid . Acetic acidlwater i 40% H,O by volume J Ethylene glycol .. Solvent Very slow <IO % decrease'in conc. in 3 hr. 0.002 68 258 30,250 0*0105 66 29,800 0.0332 21 16,700 0.359 2 22,140 I I I The addition of water to the ethylene glycol or the acetic acid thus presumably greatly increasing the dielectric constant of the solvent, increased the rate of decomposition considerably. In the case of the acetic acid addition of 2 yo by volume of water had little effect but further addition up to about 30 yo by volume caused a large increase in the rate. Above 30 yo increase in the water content had no effect. Also in chlorobenzene solution a t I 15.5' C no decomposition was detectable over a period of hours but if 1-5 mole yo of acetic acid was added the time of half decomposition fell to 64 min.These results would seem to show that in the first three solvents de-composition of the hydroperoxide molecules into radicals is taking place, possibly followed by a chain decomposition but that with a change in conditions an ionic decomposition can occur very readily'. Water and organic acids are very often stable products of oxidation of hydrocarbons in solution and thus as the reaction proceeds it is possible that ionic decomposition of the hydroperoxide formed might become important. Prof. Bawn (Liverpool) said IXr. S. F. Mellish and I have observed that the stable free radical aa-diphenyl p-picryl hy'drazyl reacts rapidly with the radicals of the type RO. and this provides a simple and convenient method for studying the rates of dissociation of peroxides into radicals Bartlett and Cotman J .Amer. Chem. SOC. 1950 72 3095. Kharasch Fono and Nudenberg J . Org. Chem. Igjo 15 748; 1951 16 113 128. 66 For example Robertson and V'aters J . Chem. Soc. 1948 1577 3 32 GENERAL DISCUSSION (ROOR -f RO. + RO.). The vividly coloured radical which gives stable solutions in a wide range of organic solvents undergoes a sharp colour change on reaction with radicals which may be measured in a simple colorimeter. This method has been used to measure the rate of dissocj-ation of polymerization in;t'ators such as peroxide and azonitriles. Dr. W. A. Waters (Oxford) said A similar instance to the decomposi-tion of decalin hydroperoxide quoted by Dr. Tipper is that of tetralin hydroperoxide which was reported some time ago by Robertson and This afforded a good example of a reaction which appeared from kinetic study to be much more simple than i t really was.The results reported by Dr. Henderson show the extent to which very minor products or impurities can influence the course of autoxida-t:on. In the autoxidation of a related compound dibenzyl ether (Ph . CH,),O which has recently' been studied by Mr. Wickham-Jones and myself we have been able to show that irregularities in the oxidation are due to the formation of a trace of a phenolic by-product. Dibenzyl ether gives a stable peroxide and in the main the uncatalyzed rate of oxygen absorption is independent of the concentration of the peroxide formed. Homolytic dissociation of the peroxide to give more chain-starting free radicals does not therefore play a major role in determining the oxygen uptake rate.However the autoxidation is self-retarding, and we have been able to show that as the peroxide of dibenzyl ether decomposes there is gradually formed just enough phenolic material to give a positive indophenol reaction. We ascribe the gradual retardation of the autoxidation to the formation of this phenol and have noted that when our reaction vessel was packed with chopped glass wool there was less peroxide decomposition less formation of phenol and less retardation of the autoxidation. Here the decisive factor seems to be surface catalysis of the mode of the secondary reaction-the peroxide decomposition-yet i t significantly influences the whole autoxidation process. Sir Alfred Egerton (Imperial College) (communicated) With refer-ence to Dr.Henderson's paper i t is well known that reproducible results are not obtained in hydrocarbon oxidation until the surface of the vessel has been conditioned by previous experiments ; the wall catalyst obtained in the liquid-phase oxidaton of ethylbenzene seems to be in line with this effect. Prof. J. P. Wibaut (Amsterdam) said In collaboration with Dr. A. Strang 6 8 we have carried out an investigation into the oxidation in the liquid phase by molecular oxygen of a number of normal alkanes with 8 to zz carbon atoms and of some branched octanes. When cobalt stearate is used as a catalyst the oxidation proceeds at a measurable rate. It has been found that the first stage of the oxidation reaction consists in the formation of a hydroperoxide.z 5-Dimethylhexane is slowly oxidized by molecular oxygen a t zoo C and a crystalline dihydro-peroxide melting a t 106.5~ C is formed : CH CH3 CH CH3 I I 1 I I I I H Ir H3C-C-CHz-CH2-C-CH3 + 2 0 2 4 H,C-C-CH2-CH2-C-CH3 0 0 0 0 I H H The presence of small quantities of peroxides can also be detected in samples of other saturated hydrocarbons which have been kept for a long time for instance in 3 4-dimethylhexane z 5-dimethylhexane, 3-methylheptane n-nonane n-hexadecane methylcyclohexane. The primarily formed peroxide is decomposed under the influence of the cobalt 67 J . Chem. Soc. 1948 1578. 68 W'ibaut and Strang Proc. Kon. Neder. Akad. Wet. B 1951 54 ( z ) 101 GENERAL DISCUSSION 333 ions ; the radicals thus formed start a chain mechanism so that the oxida-tion reaction proceeds : ROOH + Co++ -+ RO' + OH- + Co+++ ROOH ~2 ROO- + H+ Co+++ + ROO- -+ Co++ + ROO.We ascertained what products are formed by the oxidation of 2 5-dimethylhexane ; with reference to these reaction products a reaction scheme can be drawn up. The characteristic feature of the reaction scheme for the catalytic oxidation in the liquid phase of z 5-dimethyl-hexane and of other branched hydrocarbons is that the chain mechanism is initiated by an alkoxy radical. We ascertained the maximum rate of oxidation of normal alkanes with 8 g 10 12 14 16 18 20 and 22 carbon atoms. The experiments were carried out at 110.4" C and with 0.112 mmoles cobalt stearate per 61.7 mmoles hydrocarbon.There is a linear relation between the number of carbon atoms and the maximum rate of oxidation from C,,H, to C Z z H 4 6 . This can be explained by assuming that all the secondary carbon atoms have an equal chance of reaction. Some branched alkanes oxidized much more easily than the isomers with normal chains ; we ascertained the maximum rate of oxidation at 78.1" C measured in ml. 0 per 61.4 mmole hydrocarbon per hour (catalyst 70 mg. cobalt stearate) to be : a-methylheptane . . 5.0 3-methylheptane . . 3'0 3 4-dimethy'lhexane . . 11.0 z 5-dimethylhexane - 35'0 3-methyl-3-ethylpentane . 0-0 z 2 4-trimethy'lpentane . 0-0 The oxidation begins at a tertiary C-H bond from which a hydro-peroxide group is formed. A considerable quantity' of acetone is formed during the oxidation of 2-methylheptane methylethylketone being formed during the oxidation of 3-methylheptane.The fact that z z 4-trimethylpentane (iso-octane) is not oxidizable under the conditions chosen can be explained by steric hindrance. The quaternary carbon atom which has three methyl groups screens the tertiary carbon atom to such a degree that an oxygen atom cannot approach the tertiary hydrogen atom to within the distance pertaining in the transition state. Dr. M. F. R. Mulcahy (Melbourne) (counrtzunicated) Several con-tributors 89-7 have discussed the liquid-phase oxidation of hydrocarbons catalyzed by decomposing peroxide or metallic catalysts in terms of the propagation mechanism : R- + 0 + ROZ-ROZ- + RH + ROOH + R-. It may perhaps be of interest to recall that there is evidence that in the absence of catalysts (and of light) formation of hydroperoxide may occur by' a different mechanism.This was shown by the work of George and Robertson 73 on the " thermal " oxidation of tetralin. A similar result has recently been found by Mr. Watt in this laboratory with the uncatalyzed oxidation of benzaldehyde,' the kinetic behaviour being analogous to that of the tetra'in reaction. In the presence of benzoy'l peroxide however a reaction of the type shown above is initiated and is (additionally) superimposed on the uncatalyzed reaction. 69 Bateman Gee Morris and Watson this Discussion. 7O Bau-n Pennington and Tipper this Discussion. 71 Brook and Matthews this Discussion. 72 Mulcahy this Discussion. 73 George and Robertson Proc.Roy. SOC. A 1946 183 309 ; George Proc. 74 Mulcahy and Watt Nature (in press). Roy. SOC. A 1946 183 337 334 GENERAL DISCUSSION Dr. N. Uri (Munchester) said The liquid-phase reactions in which a wall effect is observed are few and therefore remarkable. There are ody four such cases known to me : (a) the experiments relating to the oxidation of liquid ethyl benzene described by G. M. Henderson in this Discussion; (b) my own findings in the catalytic decomposition of hydrogen peroxide under certain experimental conditions ; 75 (G) the observations made by Dain and Kachan 7 G in the photochemical oxidation of water by ceric ions. It is not unlikely that the wall effect in this case leads to a recombination of OH radicals and a subsequent instantaneous decomposition of hydrogen peroxide by’ ceric ion.(d) In our work on the photo-initiated free radical polymerization of vinyl compounds in aqueous solution77 we made the observation that under conditions when practically all the free radicals are formed a t a distance < rmm. from the wall the polymer is produced (by’ recombina-tion of active endings) exclusively on the wall; none appears to be SUS-pended in solution. No doubt these experiments require some co-ordination and in this connection i t is interesting to note that all these reactions involve free radicals as intermediates and the process effected by the wall is con-sidered to be a termination process which may in some cases lead to a new type of chain reaction as reported by Henderson.In the oxidation of lubricating oils the effect of iron or copper as single catalysts was studied by Brook and Matthews. It is well known that in the catalytic decomposition of hydrogen peroxide the joint action of iron and copper as co-catalysts is much more than additive. It would therefore be interesting to study those effects in the oxidation of hydrocarbons particularly if peroxides are postulated as intermediate products. Dr. G . M. Henderson (Blackley) (communicated) The effects of walls on liquid-phase chain reactions are fairly’ well indicated in past work and to restrict it to two examples on oxidations excluding all references to solid catalysts the most relevant papers are those of Stephens 78 working with cyclohexene and Medvedev and Podyapol~kaya,~~ working with tetra-lin where many effects akin to the present work were noted and the effects of walls mentioned only with very different explanations. The novel feature of ethyl benzene to which we wished to draw attention as a convenient expet imental medium is that the two hydroperoxide forming mechanisms seem more clearly distinguishable and mutually exclusive. With very fresh and pure ethyl benzene we presume that the original wall termination process is without further visible effect but that in an older sample or after sufficient of the initial peroxide has further decom-posed some of the breakdown products possibly acids act as haptens (to borrow a term from immunology) and a new wall termination process arises which gives rise to visible effects on the rate of reaction. One can generalize that in other similar reactions that the two possible re-actions might then run concurrently or that the second stage might never even arise. Sir Alfred Egerton (Imperial College) (communicated) This paper refers particularly to the inhibition of the oxidation of lubricating oils, I would like to draw the authors’ attention to a paper by Hanson and 75 Uri J . Physic. Chem. 1949 53 1070. 76 Dain and Kachan A C.S. Abstr. 1949 43 7349. 77 Unpublished observations. 78 Stephens J . Amer. Chew. Soc. 1936 58 219. 7g Medvedev and Podyapolskaya J . Physic. Chew. U.S.S.R. 1939 12 79 tiENERAL DISCUSSION 335 myself 8o on " Nitrogen oxides in internal combustion engine gases " in which the promoting action of the nitrogen oxides on the oxidation of lubricating oils was investigated and a subsequent paper entitled " In-fluence of catalysis on oil oxidation " 81 iron oxide was found to inhibit the nitrogen oxide catalyzed reaction. Dr. J. B. Matthews and Mr. J. H. T. Brook (Thornton) (communi-cated) In connection with Dr. Uri's suggestion that the study of mixed iron and copper catalysts would be of interest in the oxidation of hydro-carbons] results have been given in the paper showing that in the inhibited reaction the two catalysts are additive in their effect on the initial rate of reaction. In the uninhibited reaction] however the measured values of t$ are greater than the added effects of the individual catalysts thus lending support to the inclusion of peroxides as intermediate products. Values of t$ for iron and mixed iron-copper catalysts are given in Table 11 c s I 0 WEiCHT FRACTION OF HYDROCARBON FIG. I. of the paper and values of 4 for copper can be calculated from the data in Table I together with values of the intercepts with the oxygen absorbed axis of the asymptotes to the curves in Fig. 3 using the equation to describe the asymptote. Mr. J. H. T. Brook ( T ~ C Y M ~ O O ~ Z ) said I would like to add some further information on the kinetics of the iron-catalyzed inhibited oxidation. Using 50 p.p.m. of iron added as ferric stearate at 1 5 0 O C and using diphenyl as the inert solvent the dependency of the reaction upon the oil concentration was found to be of the form (as Fig. I above) : = ( A + B)t - (B/4) [RHI cc 220 + 330 [RH]' No immediate explanation of the difference in kinetics between the iron and copper-catalyzed reactions is apparent. ' 0 PYOC. ROY. SOC. A 1937 153 90. 81 Symposium on Engine Wear Inst. Mech. Eng. 1937