D. ROWLEY AND H. STEINER 213 11. HYDROCARBON REACTIONS GENERAL DISCUSSION A. THERMAL REACTIONS Dr. P. Torkington (Brit. Rayon Res. Assoc.) (communicated) Map I point out rather an interesting correlation between results quoted in papers by Sir Cyril Hinshelwood and Prof. Glockler ? The latter shows that the C-H and C-C bond energies in n-paraffin chains increase in passing from the centre towards the ends ; the former states that in the thermal chain fission of n-paraffins the probability of C-C fission (and GENERAL DISCUSSION * 214 hence also of C-H fission accompanying H-transfer) also increases to- wards the ends. I suggested earlier in this Discussion that Prof. Glockler’s results might be interpreted as evidence for the existence of molecular orbitals (overall but not necessarily delocalized in the Sense of implying electron mobility) favouring the chain ends.The correlation here implies that the presumably higher terminal electron density is associated with greater activity i.e. electron mobility an interesting negative analogy with true bimolecular reactions. Regarding the concentration of energy via vibrational coupling I should like to draw attention to another previous remark of mine following Prof. Ubbelohde’s paper. The envisaged concentration of energy could surely never occur with “ perfect ” coupling ; it seems that if the occur- rence of energy concentration by vibrational coupling is accepted as proved then so also must the occurrence of two or more collisions pro- viding the energy sources.Transfer from two or more such sources by coupling might well lead to a concentration of energy favourable to re- action by’ normal interference. It seems quite probable that a long-chain molecule might be involved in several simultaneous collisions but against fundamental principles for a concentration of energy to arise from a single collision. Finally is it possible for a little more non-mathematical commonsense to make a little more headway against a little more formalism ? How does symmetry lead to electro delocalization 9 Why because one can write functions 4 = ,Zai$i because of symmetry does an electronic system become mobile ? Perhaps too much n-bond theory has preceded a study’ of normal molecules. In an unsymmetrically-substituted benzene de- rivative there is no symmetry but still the same delocalization as in the parent compound.The point is obvious and not worth developing. at any link. Dr. G. Gee (Welwyn Garden City) said I should like to outline a possible explanation of the observed thermal decomposition products of paraffins. Considering first radical mechanisms to a first approxima- tion the thermal dissociation of a long paraffin chain will occur randomly CH,(CHz)nCH + CH3(CHa)z + CH,(CHa)L - - ( 1 ) The resulting radicals will then tend to degrade by the facile process of losing a series of ethylene molecules from the end CH3(CH,),* + CH3(CH2)g2 + CH*CH2 and this process will continue until a methyl or ethyl radical remains. Thus primary dissociation of long chains will result mainly in the produc- tion of ethylene and ethyl and methyl radicals and the ratio of these primary products will not be markedly temperature dependent.+ CH,(CH,),CH=CH,+CH,(CH,),-, (2) ( 3 ) The most probable secondary reaction of a methyl or ethyl radical appears to be the dehydrogenation of a paraffin molecule a process which again will occur nearly randomly along the chain (3) and will be followed by dissociation of the resulting radical (4) CH3* + CH,(CH,)nCH + CH + CH,(CHa),*CH(CHa),-,,CH CH3(CH2).*CH(CH2),-,,CH3 (4) The situation now becomes more complex in that the olefin produced will be more reactive toward further methyl and ethyl radicals than is the original paraffin. Without seeking to follow o u t these processes in detail i t can be seen that a further series of degradative processes will follow from the secondary radicals in which the principal final products will again be ethylene methane and ethane.Thus i t is easy to see why these products predominate and also why their relative yields are largely independent of conditions. It also follows from these considerations that in presence of nitric oxide the main hydrocarbon product to be expected from radical dissoci- ation is ethylene since nitric oxide would compete more successfully with GENERAL DISCUSSION 215 (3) than with ( 2 ) . The fact that methane and ethane are still produced in presence of nitric oxide therefore argues in favour of Stubbs and Hinshelwood’s contention that a molecular reaction occurs.The mechanism of this reaction does not appear very clear but it may be plausibly sug- gested that ethylene could be eliminated from the chain at any point through the formation of a cyclobutane type of activated complex CH,R CH,-R -+ I I + CHeCH,+R-R CH,-.R CH,-R . - ( 5 ) where R R are alkyl radicals or hydrogen. It is not a t first sight clear why this reaction should lead to methane and ethane rather than to larger paraffins but it appears likely that increase in the size of the radicals R and R would lead to steric difficulties. If this is indeed the case each successive step in the degradation of a long chain (by removal of ethylene) will lead to a more readily dissociable paraffin and-as in the radical decomposition-the principal final products will be ethylene methane and ethane.Dr. F. J. Stubbs and Prof. C. N. Hinshelwood (Oxford) (co.mununi- cated) In connection with Dr. Gee’s interesting suggestions we may state for comparison the views which we have formed about the reason for the similarity of products from a chain reaction and a molecular process. For the typical case of n-hexane the molecular process would be CH,CH,CH,CH,CH,CH -+ CH + CH CHCH,CH,CH CH,CH,CH,CH,CH,CH -z C,H + CH CHCH,CH, where El and El’ are the activation energies of the two reactions. The various steps in the chain reaction would be Initia- (1) (1’) in the molecule CH,CH,CH,CH,CH,CH,. El El’ CH,CH,CH,CH,CH,CH 3 X + Y tion { X+CH,CH,CH,CH,CH,CH + XH+’CH,CH,CH,CH,CH,CH DCH,CH2CH2CH,CH,CH -+ CH;+CH CHCH,CH,CH *CH,CH,CH,CH,CH,CH -+ C,Hi+CH CHCH,CH (2) CH;+CH,CH,CH,CH,CH,CH +- CH4+’CH,CH2CH,CH,CH,CH (2’) C,H; +CH,CH,CH,CH,CH,CH -f C,H,+’CH,CH,CH,CH,CH,CH where e, e, etc.are the respective activation energies. With ling chains the ielative proportions of methane and ethane formed will depend only on the repeating steps ( I ) and (1’) in the above scheme and their sequels ( 2 ) and (2’). The nature of the initiating re- action is unimportant for this purpose. For the products from the molec- ular process and chain reaction to be in the same proportions all that is necessary is that (&-El’) should equal (el-el’). This seems quite likely since the alternative steps in the molecular process are closely parallel with the steps (I) and (1’) respectively the two positions of rupture being in the same relative position in the radical CH,CH,CH,CH,CH,CH; as Dr.M. Szwarc (Manchester l7niversit-y) said Stubbs and Hinshelwood proposed in their paper an interesting mechanism for a unimolecular decomposition. I would like to examine some of the conclusions which it seems to me follow from their assumptions. Let us consider a system composed of two components viz. an inert gas X and a reacting gas M which may decompose unimolecularly. Furthermore let u s choose the experimental conditions in such a way that the energy supplied to molecules M arises from collisions between M and X. According to Hinshelwood and Stubhs we can observe two kinds of collisions. (if One type of collisions between M and X lead to the formation of energized molecules M* i.e.molecular species which contain an amount of energy at least sufficient for their decomposition GENERAL DISCUSSION 2 16 but distributed unfavourably amongst various internal degrees of freedom. This process we represent by the equation X + M + M* + X (collision of the 1st type). The decomposition of energized molecules requires a reshuffling of their energy which has to be accumulated eventually in the reacting centre i.e. M* +- MS + Products MS denoting an " activated " molecule i.e. a species which contains at least the necessary amount of energy localized in the reacting centre. Although collisions of the first type are frequent the probability for the energy redistribution is low and consequently the reaction caused by these collisions attains a first-order character at comparatively low pressure of the gas X.Plotting the rate of this decomposition as a function of partial pressure of gas X we obtain the curve I shown in Fig. I . (ii) Collisions of the second type lead directly to the formation of I ' activated " species MS X + M -t MS + X (collisions of the 2nd type). Since the chance of the occurrence of such a collision is very low and the probability for the subsequent decomposition of Mt is very high the reaction caused by collisions of the second type retains the second- order character up to comparatively high pressures of the gas X. Plotting the rate of the latter reactions as a function of pressure of the gas X we obtain the curve 2 shown in Fig.I. FIG. I. Fig. I illustrates the features of the decomposition claimed by Hinshelwood and Stubbs. There is a comparatively low pressure region (C-D) in which the first type of collision predominates and in which the reaction obeys first-order kinetics. There is a comparatively high pressure region (E-F) in which collisions of the second type are predominant and conseqfiently in this region the reaction obeys second- order kinetics. However in addition to these two regions there is a very low pressue region (A-B) in which both types of collision call for a decomposition of the second order and finally there is a very high pressure region (G-H) in which both types of collision demand first-order kinetics of decomposition.If we plot therefore the order of reaction as a function of the partial pressure of gas X we obtain a curve represented in Fig. 2. I would like to ask Prof. Hinshelwood if he agrees that this behaviour of decomposition is a consequence of his assumption. There is however another point which I consider to be important. If the reaction takes place in a very high pressure range say G-H then the first type of collision must lead to an equilibrium between the energized molecules M* and the ordinary molecules M i.e. X + M + M* + X (collisions of the 1st type). GENERAL DISCUSSION 217 I n the same pressure region collisions of the second type must lead to an equilibrium between the " activated " molecules MS and the ordinary molecules M i.e.X + M $ MS + X (collisions of the 2nd type). One has to realize that any type of " activated " molecules MZ can be produced either by collision of the 1st type via M* or directly by collision of the 2nd type. If the latter modes of activation leads to the state of equilibrium between ordinary molecules M and activated species MZ then i t would be expected that the activation through collisions of the first type should also lead to an equilibrium between these two species (this follows from the fact that both types of collisions occur in the same gas phase). Consequently the following scheme should be valid for a reaction resulting from collisions of the first type M + X + M* + X MZ -f Products Px I Jr and the equilibrium concentration of MZ should be the same as in the case of collisions of the 2nd type.That however would contradict the as- sumptions on which Fig. z was obtained. FIG. 2. One can argue that the collisions of the second type do not lead to an equilibrium concentration of Mi since their rate of decomposition is too high being of the order of 1013 (MS) sec.-l. This seems to be quite correct. Indeed collisions of the second type can never be of any significance and if they have to maintain a first-order character of decomposition a pressure of the order of 105 atm. would be required. It seems that the best method of approaching the problem is t o sum up aZZ the contributions arising from all possible collisions (appropriately weighed). The reaction would be of the second order a t low pressures and would attain the first-order character at sufficiently high pressures In the latter case the activation energy of decomposition would be given by the activation energy of the process M +M$.Dr. F. J. Stubbs and Prof. C. N. Hinshelwood (Oxford) (communic- ated) Given the hypothesis that the nitric-oxide inhibited reaction is a definite molecular reaction (which as we have explained there is a good prima facie case for accepting) the changes of order referred to by Dr. Szwarc are the basic experimental facts upon which our views about the kinetically composite nature of the reaction are founded. The tentative interpretation which we have advanced is not in any way built upon a preconceived theory and may be modified in the light of further experi- ments now in progress.In the meantime we would prefer not to regard the facts about the order as deductions from a theory as Dr. Szwarc 218 GENERAL DISCUSSION implies but as observations calling for explanation. What we have suggested is that the normal theory requires some degree of amplification. Prof. S. W. Benson (California) (conznzunicated) The data presented by Ingold Stubbs and Hinshelwood and quoted here seem to indicate that even in the " fully inhibited " pyrolysis there is a surface effect which increases with decreasing temperature and is virtually independent of pressure. It seems to me that until this effect is further elucidated i t is difficult to accept the discussion and rate constants quoted as anything but tentative.There still remains the possibility that there is a heterogeneous reaction initiated on the walls which starts chains which also end at the wall. Rice and Herzfeld 1 have recently shown that under such conditions i t is possible that depending on the mechanism the rate may show any de- pendence on the surface volume ratio between complete independence and direct proportionality. Even in a direct decomposition to molecular products i t is conceivable that heterogeneous decomposition competes with homogeneous decom- position. In such case the activation energy of the two would be ex- pected to be different and since the relative amount of each would be expected to show a monotonic if not simple pressure dependence there would be observed an effect of pressure on experimental activation energy.For n-pentane i t is difficult to believe that when the experimental activation energy is found to vary between 80 and g o kcal./mole in the pressure range 50-150 mm. Hg (inhibited reaction) t h a t the free radical mechanism of splitting of the C-C bond which has an activation energy of 80-85 kcal. is not proceeding at a competitive rate. Finally it is difficult to reconcile the observed decrease of activation energy with increasing pressure with any reasonable behaviour of mole- cules. In the Rice-Ramsperger-Kassel theories of the unimolecular decomposition the activation energy is supposed to decrease at the lower pressures where the higher energy states can no longer make their com- plete contribution to the overall rate.The explanation offered by the authors to the effect that there may be a single mode which requires a very long time to reach activation even when the energy is present in other modes seems completely ad hoc and unsupported by any other experimental or theoretical evidence. It also leans upon an interpreta- tion of a molecular collision that seems in immediate conflict with the law of microscopic reversibility. That is there seems to be an implicit assump- tion that an active molecule can only lose energy on collision and never gain it. If there is a mode which is only slowly activated then the chances are that i t is just as slowly deactivated and collisions do not necessarily play any decisive role in the time of decomposition.The example of ethane where the kinetics seem to go through a second to first order transition while the activation energy changes but little (or possibly is higher a t the lower pressures) seems a striking example of the above-mentioned contradiction. I doubt very much that these pyrolyses will be very tractable theoreticallv until the nature of the surface efiects and the inhibition axe better understood. Dr. F . J. Stubbs and Prof. C. N. Hinshelwood (0,vford) (conz~lzunic- nted) Chain reactions which are initiated by a decomposition are not at all likely to start and stop at a wall. If the molecule dissociates on the wall radicals are likely t o remain adsorbed. In any case chain reactions staxting at a wall should be subject to the usual kinds of inhibition just like any other chain reactions.This assumption of Prof. Benson seems to us to be improbable. We are unable to understand his fourth paragraph. In pentane we have always believed the chain reaction normally to proceed a t a " competitive rate " but we state a prima facie case for concluding that 1 Symposium on Abnormalities in Reaction Kinetics Amer. Chem. SOC. Minn. hfinn. Spring 1950. GENERAL DISCUSSION I 80 CH3-CH3 1 219 this competing reaction has in the conditions of the experiments been suppressed. In his penultimate paragraph we do not recognize our views very clearly and as to the last paragraph we can assure him that our approach continres to be essentially experimental and that all the details of these admittedly rather puzzling phenomena are under examination still and from several new points of view.Dr. L. Bateman (Welwyn Garden City) (communicated) The form and relative position of the paraffin and olefin decomposition rate curves in Fig. I of Stubbs and Hinshelwood’s paper are just as would have been predicted on the basis of A H changes on breaking the most easily broken bonds so that some reference to this mode of primary bond scission is perhaps of interest. The AH changes (in round figures) for the C,-C compounds are summarized in the following Table. Ethylene should be I00 80 < 60 much more stable than ethane propane and propylene should be of similar stability and the higher members of the two series should exhibit a difference reflecting the partial utilization of allylic resonance energy in the bond-breaking process.These features are evident experimentally but ethane and I-butene appear to be relatively more stable than might be expected and may serve to illustrate how secondary processes affect the overall decomposition. The common factor is the primary formation of methyl radicals and not higher alkyl radicals capable of ready decom- position into smaller fragments. Hence in ethane recombination will be specially favoured since the only alternative reaction is the rather difficult hydrogen exchange In Me I-butene + CH the CH corresponding . CH,Me -f MeH reaction + CH to the CH latter .CH . Me ( E - 5 kcal.) is much more facile but the resulting substituted allyl radical like that initially produced will be far less active than an alkyl radical in inducing further decomposition. An important aspect of these considerations is that all A1-olefins are required to yield simple allyl radicals as primary fission products. The reactivity of these radicals under the decomposition conditions their ultimate products and possibly the estimation of their extent of forma- tion might well be stcdied experimentally by carrying out co-decomposi- tions with I 5-hesadiene as a foreign source of allyl radicals. Informa- tion on the third point would establish the contribution of allylic bond primary scission to the evidently composite overall decompositi m.Dr. R. G. Partington (Oxford) (communicated) In their paper Stubbs and Hinshelwood state that Steacie and Folkins suppnrt the view that the reaction taking place In the presence of sufficient nitric oxide to reduce the rate to a minimum (the “ residual reaction ”) is “ the primary process of what in the absence of inhibitors would be a chain reaction ”. This is I think a mistaken interpretation of Steacie and Folkins’ conclusions I 1 slightly ( 8 0 CH . CH . CH,-H CH3. CH,-CH3 CH CH-H 1 due to second- CH CH . CH,-CH3 60 CH3. CH,-CH,. CH3 CH . CH,-CH . CH . CH J order effects CH CH . CH,-CH . CH3 ’lightly Me + C,H + MeH + C,H (E N 10-15 kcal.). Steacie and Folkins Can. J . Res. B 1940 18 I. CH,-CH,-CH,-CH,-CH3 CH3+N0 = CH3N0 CH3-CH2-CH,-CH,+ 2 20 which read '' the addition of nitric oxide then merely diminishes the chain length without completely suppressing the chains ".Had the suggestion been made that all the initidly formed radicals reacted with nitric oxide one would have to consider the possibility of the formation of such a molecule as (I) in the following scheme e.g. (1) which might well participate in further reaction (starting by its decom- position) in a similar way to the large radical itself. A further paper by Steacie and Folkins 8 is of interest in this connection. It is concerned with the decomposition of n-butane initiated by free C A FIG. 3 . (19391 P. 700. There are many possible sources and convenient methods of producing the appropriate radicals either thermally or photochemically and in connection with the possibility of making such a study some experiments of mine on the chemical analysis of small concentrations of nitric oxide in gas samples of the size usually handled in work of this character might be of value.A gas sample (say 50ml. of gas at N.T.P.) containing say Q to I yo of nitric oxide is enclosed in the holder A (Fig. 3 ) . This is then con- nected to a flask B containing air and small amount of an aqueous solution of sodium or potassium hydroxide free of nitrite. On opening the wide- bore stop-cock C and gently shaking the reactions take place. Reaction ( 3 ) is slow and the conversion to nitrite is almost complete. The solution of nitrite can then be analyzed with great accuracy by the standard colorimetric method^.^ The conversion factor NO -f nitrite is reproducible for a given nitric oxide concentration and its value could be determined over the desired range of concentrations by making up mixtures.Can. J . Res. B 1939 17 gg. 4 Trotman-Dickenson and Steacie J . Amer. Ckem. SOC. 1950 72 2310 ; J . Chem. Physics 1950 18 1097 and following papers. 6 See for example Vogel A Text Book of Quantitative Inorganic Analysis (3) (4 2NO + 0 = 2N0 . (NO + NO,) + 2NaOH = 2NaN0 + H,O . - (4) GENERAL DISCUSSION = CH,+CH,-CH2-CH2-CH3 NO = CHs-CH~-CH~-CH2-NO radicals from ethylene oxide and the authors conclude that their results " cast some suspicion upon the idea that maximum inhibition by nitric oxide in all cases corresponds to complete suppression of chains *'.The whole problem of the action of nitric oxide as ail inhibitor (and the foregoing remarks were largely included to justify the opinion that the problem still exists) seems to depend on the relative rates of the two reactions viz. reaction of radicals with the hydrocarbons e.g. * R + RH = R,H + R . and reaction of radicals with the inhibitor e.g. * (1) R + NO = R,NO. . Reactions of the type (I) have recently been studied and similar specific attention to reactions of the type ( z ) would undoubtedly be of the greatest value in helping to solve the above problem and might be less subject to ambiguity than experiments on a system in which reactions of this type are among the very many taking place.* 22 I GENERAL DISCUSSION . - (20) * CH + (CH,),CO -+ CH + residue. Since the agreement of values of El based on the two separate assumptions was good one conclusion was that the two sets of results were consistent. The more recent value of Dodd (E20 = 10.7 f 0.5 kcal.) is simply not reconcilable with the older value (8.6 kcal.) of Grahame and Rollefson. It is consistent with the results of Trotman-Dickenson and Steacie only if it is assumed that our measurements of E differences are inaccurate to about 2 kcal. We are cognizant of the difficulties of this work. Never- theless the new evidence presented by us does support a high value of El and Dodd’s results emphasize that fact. Dr. A. F. Trotman-Dickenson (Manchester) said I am not clear why Anderson Davison and Burton have disregarded the recent work of Dodd in selecting 8.6 kcal.as the best available value for the activation energy of the reaction CH + CH,CHO -f CH + CH,CO. Dodd has shown that the results of all previous workers (including those of Rollefson and Grahame) may be interpeted as giving a value for the activation energy of 10.7 f 0.5 kcal. This value makes the agree- ment between the acetaldehyde and acetone results much worse (15.8 and 13.2 kcal. respectively). Dr. Steacie and I recently reviewed the data on the reaction CH + H +CH4 + H El = 8.8 kcal. steric factor = 4 f 2 x I O - ~ (oR2 = 2-8 A ucas = 3.5 A) and concluded that the best work (three separate determinations) gave and K / K t at 182O C = 50 x 10-1 molecules-* cm.812 sec.-4.The results of Anderson Davison and Burton give K,/kt2 at 182O C = I x 10-1 molecules-* ~ m . ~ / e set.-* if the same collision diameters were used as above. No satisfactory reasons for these serious discrepancies have been given. Prof. M. Burton (Notre Dame) (communicated) Mr. Trotman- Dickenson mLst not have realized that the full paper by Dodd did not reach our laboratory until after the paper by Anderson Davison and myself was submitted for publication. Otherwise we would certainly have commented on it. Our actual work was concerred with establishment of activation energy differences. Calculation of E for - (1) CH,+H + CH4+H . involved also assumptions of reliability of E values of Grahame and Rollefson for CH + CH,CHO -+ CH + CH,CO and of Trotman-Dickenson and Steacie for CH + H = CH + H .(18) Regarding the relative merits of the results of Dodd and of Grahame and Rollefson we have no firm opinion. However we may note that E, >El* is a new idea. The reverse order of activation energies ac- cepted prior to the work of Dodd is consistent with the usual notion that since in reaction (20) a formyl H is involved while in reaction (18) it is a methyl H reaction (20) probably has a lower E,. Mr. Trotman-Dickenson’s remark concerning discrepancies between values of kl/kl24 from old experimental data and from calculation involving our El and s emphasizes the dangers inherent in acceptance of old cal- culations based on experiments which may contain innate error e.g.the general difficulty of interpretation of experiments of involved mechanism. Prof. W. A. Noyes Jr. (Rochester) said It is interesting to note in the paper by Prof. Burton and co-workers that for the reaction (1) Dodd J . Chem. Physics 1950 18 234 ; Trans. Faraday SOC. 1951 47 56. 7 Trotman-Dickenson and Steacie J . Physic. Chem. (in press). GENERAL DISCUSSION * 222 the activation energy is 13-14 kcal. and the steric factor about I O - ~ . These results are based on studies in which reaction (I) could compete with reactions of the type CH + RH = CH + R. . The activation energies for (2) depend on R but generally fall in the region of 8-10 kcal. (e.g. if R is CH,COCH,- at about 9-7 kcal.).Thus if steric factors for (I) and (2) were identical (2) would proceed about 102 times as f a s t as (I) at equal pressures of H and of RH at temperatures of 200-300° C. A ratio of rates as high as 102 would invalidate most conclusions due to attendant experimental difficulties. The very fact of successful com- petition under experimental conditions used by most authors indicates quite strongly that steric factors for (2) are usually 10-1000 times lower than for (I). (3) (2) This fact and other related facts raises questions about “ hot ” radical conclusions. If a radical separated from a parent molecule by absorption of energy much greater than that necessary to break the bond retains energy solely as kinetic energy i t could hardly undergo 103-105 collisions without being reduced essentially to thermal equilibrium with its sur- roundings.One must conclude either that hot radical effects would be unobservable or that the so-called steric factor is a very pronounced function of kinetic energy. Theory concerning this matter is very vague and qualitative although one is led to predict an increase in steric factor with increase in kinetic energy. An examination of published data on hot radical effects for reactions of the type of (2) reveals that positive conclusions are not warranted. Uncertainties exist either due to scatter in the data themselves or to effects other than hot radicals which afford equally satisfactory explanations. This is true particularly in those cases such as CH,COCH3 and Hg(CH,) where the full nature and yield of the primary photochemical process have not been elucidated as a function of temperature and other variables.A word should be said also about the reaction CH + CH = CpH6. . All workers seem agreed that activation energies for radical combination reactions of the type of (3) are very low. Less agreement is found con- cerning the steric factor but the safest conclusion seems to be that i t is high and probably near unity. The necessity for a third body has not been clearly shown but an examination of the data a t sufficiently low pressures will indicate a trend which might show a third-body effect More data on this point will be found in a forthcoming article by Dr. A. J. C. Nicholson as well as in a recent article by Gomer and myself.* A third body may be introduced in several ways but i t is not satisfactory to introduce a mere triple collision.One way would be (4) 1 4 * CH + X c Z ks CH,X CH,X + CH = C,H6 + X . in which case the rate of ethane formation would be ( 5 ) Under conditions of high radical concentrations (high intensities) eqn. (6) reduces to an equation which is satisfactory for part of the ethane formation during photolysis of mercury dimethyl but which is not satisfactory under con- ditions so far studied in acetone. The trend in acetone at low pressures may indicate the necessity for a three-body collision. J . Amer. Chem. SOL 1950 71 101. GENERAL DISCUSSION * (8) 223 The most satisfactory way to obtain the right rate equation for the majority of experimental conditions is that used by Gomer and others viz.the reversible formation of an intermediate complex from two methyls the intermediate complex being stabilized to ethane by collisions. At sufficiently high pressures the rate-controlling step is the formation of the intermediate complex a step whose rate depends only on (CH3)2. One comes to the conclusion therefore that the nature of the third body may determine the way in which i t acts in agreement with the classical experiments of Rabinowitch and Wood on the recombination of bromine atoms. Under most conditions the rate of ethane formation is given by RQG = k(CH,)' in which h may be a complex constant which indicates nevertheless the rate of (3) to be very high.Dr. P. Torkington (Brit. Rayon Res. Assoc.) (corutmuvzicated) Is the steric factor temperature-dependent ? If so its variation might possibly be related to the proportion of molecules (in this case radicals) undergoing in the reaction CH3- + H, it might be thought that approach along the a mode of vibration more favourable to reaction than other modes. Thus line C . . . H-H coinciding with opening of the CH " umbrella " at the correct stage (the hydrogen molecule lying on the opposite side to the hydrogen atoms of the methyl group) would have a high probability of successful reaction. I n this case then the favourable vibration is the symmetrical breathing frequency of the methyl group. The argument is not essentially altered if the group is planar as has been fairly recently suggested though the favouring would not be so noticeable.Possibly the principle could be applied generally. As regards feasibility with a frequency of 1500 cm.-l and amplitude of the order 10-g cm. the hydrogen atoms in the breathing mode of a methyl group have a mean velocity of the order 105 cm./sec. ; the root-mean-square velocity of hydrogen molecules at 2 5 O C is about 2 x 105 cm./sec. The example shows that vibrations might co-operate. Dr. George Porter (Cambridge) (communicated) The high value for the bimolecular rate constant of methyl radical combination at room temperature of about 5 x 1013 cm.a mole-I sec.-l quoted by Sztvarc and others in the discussion now seems well established having been obtained by several workers using the sector method There appears a discrepancy however with the results of mirror experiments which cannot be dis- regarded on the grounds that radical removal under these conditions is known to be a wall reaction.Whatever the mechanism by which the radicals disappear it is the rate-determining one and thus sets an upper limit for the homogeneous bimolecular rate constant. If the concentration of radicals is low as in the photochemical experi- ments of Norrish and Porter O the rate constant is found to be high and no discrepancy appears but as the concentration is increased so is Kmax in- creased for the radical lifetime does not decrease proportionately. Thus Forsyth's values 10 give kmax = 6 x 1012 cm.3 g .mole-1 sec.-l and although other workers have not recorded concentration measurements specifically i t seems certain from a consideration of mirror removal times that even higher concentrations were involved in the experiments of Rice Johnston and Evering l1 and of Paneth Hofeditz and Wunsch l2 who also increased the life-time to 0.1 sec. by decreasing the rate of the wall reaction. As the rate-determining reaction is heterogeneous i t appears that the velocity of the bimolecular gas phase reaction determined in this way is consider- ably less than that corresponding to unit collision efficiency. Norrish and Porter Faraday SOC. Discussions 1947 2 97. Forsyth Trans. Faruday Soc. 1941 37 312. l1 Rice Johnston and Evering J. Amer. Chem. Soc.1932 54 3529. la Paneth Hofeditz and Wunsch J. Chem. Soc. 1935 372. GENERAL DISCUSSION The value of kmax is obtained from t* and the concentration the latter being determined from the weight of mirror removed or metal alkyl formed in unit time which can be measured to within 10 yo without great difficulty. The half-life is also a reasonably accurate determination and the well-known difficulties of the method such as mirror poisoning and irreproducible wall conditions can produce only a scatter and not a general trend of this kind. It appears therefore that this discrepancy must be considered seriously unless a reason can be given for doubting the basic assumptions of the mechanism of mirror removal. The bimolecular rate constant must eventually decrease at low pressures when the collision life becomes greater than the lifetime of the collision complex.A possible explanation therefore lies in the lower pressure of the mirror experiments which is about I mm. whereas the pressure in the sector experiments was usually much higher and always above 5 mm. Dr. E. J. Bowen (Oxford) said The importance of the amount of delocalization energy in the products of a dissociation affecting the ease with which it occurs is also shown by work carried out by Miss Rohatgi on the photochemical reaction of anthracene with liquid chlorinated hydrocarbons. In light absorbed by anthracene the reaction is AX + RC1 -+ ARC1 (derivative of dihydroanthracene). . 0.018 The following quantum efficiencies were found Carbon tetrachloride CC Pentachlorethane CHCl,CCl .Tetrachlorethylene CCl,=CCl Tetrachlorethane CHCl ,CHCl . Trichlorethylene CHC1-CC1 a . Ethylene dichloride CHCl=CHCl . Chloroform CHCl . 0'4 . 0.27 . 0.07 . 0.008 . 0.0076 . 0.0032 High quantum efficiencies appear only where the CCl radical might be involved. Prof. S. W. Benson (California) said It seems rather strange that the Br atoms liberated in the initial step do not undergo addition and hydrogen abstraction reactions with the cyclohexene. I wonder if Dr. Robb has any information on this point. Dr. E. C. Kooyman (Amsterdam) (communicated) It seems likely that the main differences between the mechanisms of the addition of bromotrichloromethane and that of carbon tetrachloride to olefins arise from the far greater reactivity of the bromine atom in CBrCl as compared with that of the chlorine atoms in CCl,.Kharasch and Friedlander la found bromotrichloromethane to react rapidly with styrene at 50°C under the influence of ultra-violet light. However little reaction occurred at 20' C ; at this temperature styrene was found to inhibit the addition of CBrCl to other olefins. These facts were interpreted by Kharasch and Friedlander as resulting from the low reactivity of the benzyl type radical formed by preferential addition of CCl to the styrene double bond -CCl + CH,=CH-Ph -+ CC1 ,CH,CH-Ph. These data suggest the abstraction of a bromine atom from CBrCl to be a slow reaction at low temperatures when the attacking radical is re1 atively stable.In detailed analyses of the reaction between cyclohexene CCl and benzoyl peroxide at 78" C Kooyman and Farenhorst l4 found 60-70 ?& of 3 3-dicyclohexenyl on the basis of peroxide decomposed. In our in- vestigation on a-methylenic reactivity l5 the " retardation constant '' for cyclohexene was found to be 11 x I O - ~ ; in view of the value 0 - 2 reported l 3 Kharasch and Friedlander J . Org. Chem. 1949 14 239. l4 Kooyman and Farenhorst Rec. trav. clzim. 1951 (in press). 15 Kooyman this Discussion. GENERAL DISCUSSION 225 by Kharas3i and Friedlander and by Kharasch and Sage l6 for the ratio of the addition rates of CC1 radicals to the double bond in cyclohexene and n-octene respectively this suggests the abstraction of a-hydrogen atoms in cyclohexene by trichloromethyl radicals to proceed at a rate which is not much smaller than the rate of addition.Finally attention is called to the experiments of Kharasch and Fried- lander with respect to the overall kinetics of the photochemical addition of bromotrichloromethane to various olefins including cyclohexene at 10-50OC. The authors found the reaction to be zero order with respect to olefin. Olefin determinations were made by direct titration with bromide-bromate solution rather than by dilatometry. On the bases of these arguments it is suggested that the rate constants reported by Melville and his co-workers may contain contributions from cyclohexenyl radicals rather than to apply to the trichloromethyl radicals only.Thus termination might consist both of dimerization of trichloro- methyl radicals and of dimerization of cyclohexenyl radicals. The in- fluence of the latter will be of course more pronounced a t lower halide/ olefin ratios. I f 2-5 x IO* 16 Kharasch and Sage J . Org. Chern. 1949 14 537. Prof. H. W. Melville Dr. J. C. Robb and Mr. Tutton (Birminghum) (communicated) The kinetic evidence presented in our paper entirely pre- cludes atatck by trichloro-methyl radicals on the a-methylenic hydrogen atoms to form chloroform and a stable cyclohexenyl radical as suggested by Dr. Kooyman since under all our conditions the rate of reaction is very accurately proportional to (rate of initiation) 4 and also to the concentration of olefin. If such a reaction did occur exclusively as a termination CC1 + CgHlo +CHCl + CBHg .. . . k' the rate of reaction would be given by which is of course not the case. If on the other hand both processes i.e. reactions given by k' and R were operative then the intensity ex- ponent would lie between 0.5 and I. The fact that i t is so accurately 0.5 supports the original kinetic scheme suggested in our paper. Since the discussion an experiment has been done in which cyclo- hexene in presence of benzoyl peroxide as catalyst with excess of carbon tetrachloride was placed in a sealed dilatometer tube. By irradiating with light of suitable wavelength the photo-catalyzed reaction was studied and in agreement with the work reported by Dr. Kooyman this was shown to be dependent directly on the light intensity thus establishing that under his conditions termination is indeed first order with respect to radical concentration.The rate of reaction under these conditions is very much less than when CC1,Br is used. This then raises the point of the reason for the different mechanism under the two different experimental conditions but further experiments are necessary in order to establish the nature of the differences in the mechanism involved and more precise information is required regarding the velocity constants for all possible steps in the reaction. 2k6 Prof. H. W. Melville Dr. J. C. Robb and Mr. Tutton (Birmingham) said Since the paper on the reaction of trichloromethyl radicals with cyclohexene was submitted further values have been obtained for the proposed kinetic steis.These are listed below. 1.3 x 1 0 2 1. mole-1 sec.-l J I R (30" C) k (40" C) 1-5 x I O ~ k (5oOC) 1-90 x I O ~ Eovers1l = E - &Es = 4-5 kcal./mole. 2.0 x 10' 1. mole-1 sec.-l. 2kS H GENERAL DISCUSSION 226 This gives a value of 4 = 5.5 where 2k6 = +V2k4 . 2k5. This low value is in accord with those recorded elsewhere in the dis- cussion by Bateman Gee Morris and Watson. Further attempts to obtain a value for the energy of activation of the termination step by means of experiments conducted using the technique of intermittent illumination at various temperatures has resulted in failure to detect any activation energy for this step. It should be pointed out however that this technique is not sufficiently sensitive or accurate to measure small activation energies less than 3 or 4 kcal.over the rather small temperature range normally available in experiments of this kind. Dr. M. Magat (Paris) said I would like to mention some additional evidence concerning two points raised by Dr. Kooyman in his paper. First concerning so-called " stable " i.e. non-dimerizable radicals some caution is advisable. Dr. Chapiro in our laboratory has for instance observed that the reputedly stable radical I I-diphenyl-2-picrylhydrazyl reacts with double bonds of polymerizable vinyl compounds forming molecules of a molecular weight higher than that of the radical dimer. We are now investigating the kinetics of this process and the nature of the molecules formed.Secondly concerning the reaction of radicals with poly cyclic com- pounds we have investigated the effect of addition of cancerogenic hydro- carbons and their homologues on the rate of thermal polymerization of styrene at 37" C. If the usually present traces of peroxides are destroyed by a preheating under vacuum at this temperature for 2-3 weeks all these cancerogenic compounds slow down the polymerization. It can be shown by persistent fluorescence of the polymer that these compounds do enter the chain. It is remarkable that the absorption and fluorescent spectra are not significantly shifted. The slowing-down efficiency decreases in the order 3 4-benzpyrene eo-methylcholantrene phenantrene a-methylanthracene I 2 5 6-di- benzanthracene.Chrysene and pyrene have no effect at a1l.l' Dr. K. S . Pitzer (Washington D.C.) said Prof. Kistiakowsky has reported in his paper some exceptionally fine experimental work and the minor point of interpretation which I am about to raise does not in any way detract from the principal results of the work. In the text just after Table IV i t is stated that the presently unknown corrections for anharmonicity in the vibrations would raise the calculated heat capacity of ethane more than that of ethylene. I would urge caution in accepting this estimate as certain although i t may well be correct. While i t is true that ethane has more vibrations in a given frequency range than ethylene i t could be that the effect of anharmonicity in the torsional vibrations overshadows all others at moderate temperatures.The present treatment as a restricted internal rotation accounts for the anharmonicity in the torsional motion in ethane while the more highly restricted torsion of ethylene is presently treated as a harmonic oscillator. Thus i t seems possible that the anharmonicity correction for the torsional motion in ethylene might be large enough to make the total for ethylene exceed that for ethane. The same statements can also be applied to propane and propylene. Prof. E. A. Guggenheim (Reading) (comr~zunicated) It j s clear that the measurements of Kistiakowsky and Nickle on the ethane-ethylene equilibrium are appreciably more accurate and more reliable than any previous measurements.It is particularly satisfactory to notice how well these results agree with the most recent theoretical values. It is perhaps not entirely without interest to compare the experimental values with theoretical values calculated before the experiments. The following l7 Bodme and Magat Comfit. rend. 1951,232 1657. GENERAL DISCUSSION 227 Table gives a comparison between experimental measurements made in 1942 and here reported and the values calculated according to a formula la published in 1941 based on Kistiakowskfs own experimental value for the heat of hydrogenation a t 82’ C. 450’ c . EQUILIBRIUM VALUES OF @CIQB[2/p,-,Hb IN ATM. Measured Temperature 1942 . . 5-13 f 0.13 x 10-4 380’ C 4-04 f 0.17 x 10-6 4’3 x 10-6 Calculated 1941 .. 5-2 x 10-4 Dr. L. Bateman (Welwyn Garden City) (commu.nicated) Horrex and Miles’s reference to the bond shortening in dibenzyl requires revision. Cruickshankl9 has re-analyzed Jeffrey’s data and finds that they are actually indicative of only half the contraction quoted. Moreover recent isomerization equilibrium measurements by Dr. J. I. Cunneen and myself 20 fail completely to reveal any chemical effect (AG differences < 0.1 kcal.) which could be attributed to bond shortening in I 5-dienes. I suggest therefore that the agreement claimed between the experi- mental and ‘ I calculated ” bond dissociation energies needs critical reconsideration. Three uncertainties are apparent ((i) the correctness of the observed activation energy absolutely and as regards its identity with the energy rather than say CH .CH,-CH . CH as the “ non-resonance ” refer- of primary bond scission ; (ii) the appropriateness of using CH,-CH ence compound for P h . CH,-CH,. P h ; (iii) the magnitude of the resonance energy of the benzyl radical. Horrex and Miles present con- vincing evidence for the approximate validity of their conclusions con- cerning (i) and this receives support from Bolland and Orr’s 21 investigation into the analogous breakdown of aliphatic I 5-dienes. I question however whether f I kcal. is a fair estimate of the overall uncertainv in the activation energy in view of the complexity of the decomposition process the analytical intricacies and the spread of the points in Fig. 7.Any error associated with (ii) is undoubtedly small but will be such as to tend to reduce the “ standard ” CC-bond energy (by about 1-2 kcal.). The third factor is thus left as the main source of uncertainty. Szwarc has derived the value of 24.5 kcal. by subtracting the activation energy for the pyrolytic decomposition of toluene (this being identified with the energy required to break an acyclic CH-bond) from the CH-bond dissoci- ation energy in methane. In fact there is sufficient uncertainty in this derivation for the formerly accepted value of 19 kcal. still to be tenable. Szw-arc reports the activation energy as 77-5 f 1.3 kcal. but statistical regression analysis of all his tabulated data (for uniform surface condi- tions) leads to the result E = 80.0 f 4-2 kcal.(95 yo limit). Further apart from the small uncertainty in D, in methane (101 f I kcal.) i t would seem more legitimate to compare with an alkane €We where D, in the Me group is undoubtedly lower,2s rather than with the sym- metrical methane molecule. dissociation energy of dibenzyl simply as (84 - 2 x 19) = 46 kca1.-in For the present therefore i t would seem preferable to calculate the satisfactory agreement with the thermochemical and kinetic data- rather than as (85 + 11 - z x 24.5) = 47 kcal. in which a large other- wise unrecognized bond energy term has to be invoked in order to com- pensate for a higher resonance energy which is not definitely specified experimentally. To be published shortly. l9 Cruickshank Acta Cvyst.1949 2 65. l 8 Trans. Favaday Soc. 1940 37 272 a1 Bolland and On I.R.I. Tram. 1943 21 133. 22 Szwarc J . Chem. Physics 1948 16 138. Stevenson J . Chem. Physics 1942 10 291 ; Anderson and Van Artsdalen J Chem. Physics 1944 IS 479. GENERAL DISCUSSION Dr. M. Szwarc (Munchester) (communicated) Dr Bateman's remarks enables me to clear up some details of calculation of activation energy in the pyrolysis of toluene.24 The technique elaborated in this investiga- 0.01 yo. This seems to be the lower limit permissible in this technique tion makes i t possible to measure the rate of pyrolysis down to about and indeed runs 89 88 87 and 94 (crosses in Fig. 4) show that losses of products start to be noticeable when the limiting case of 0.01 yo is reached. Therefore these four results were ignored in calculating activation energy and were omitted in the graph which was presented in the paper but they were included in the Table giving all the results.The value of 80 kcal./mole obtained by Dr. Bateman results from including these four runs; their omission leads to an activation energy of 77.5 kcal./mole as quoted in the original paper. I take this oppor- tunity to include in Fig. 4 the results of pyrolysis of toluene obtained in 1948 by Dr. J. S . Roberts and in 1949 by Mr. J. Murawski. Pyrolysis of toluene. 0 szxTm2 (1947) A Roberts (1948) E = 77'5 kcal./mole. x runs No. 89 88 87 94. Murawski (1949) 228 C-C FIG. 4. In his contribution Dr. Bateman deals with the possible connection existing between the length and the dissociation energy of the central bond in the molecule of dibenzyl.This problem has been discussed previously,26 and here I would like to clarify further certain points which need additional emphasis. (i) The dissociation energy of a bond linking two atoms depends not only on the nature of the two atoms but i t is also greatly influenced by the molecular environment of the bond in question (e.g. Table I in ref. (2)). (ii) The factors which influence the magnitudes of bond dissociation energies can be divided into two groups those connected with the struc- ture of the undissociated molecule and those due to the nature of radicals produced on dissociation. On the whole i t is not possible to ascertain which factor and to what extent is responsible for the observed changes in bond dissociation energies (see however ref.(3) in which an exceptional case is discussed). (iii) Further treatment of the problem of variations in bond dissoci- ation energies requires therefore the introduction of certain simplifying t4Szwarc J. Chem. Physics 1948 16 128. 25 Szwarc Faraday SOC. Discussions 1947 2 39. GENERAL DISCUSSION 229 assumptions. It has been assumed tentatively 26 that the variations in the C-H bond dissociation energies in molecules of the type R . H are due entirely to the factors arising from the nature of the radical R (i.e. i t has been assumed that these variations reflect the changes in the stabilities of various radicals R). Taking the value of D(CH,-H) as the point of reference it is possible zo to build up a system of experimental resonance energies " for various radicals R defined as D(CH,-H) -D(R-H).(iv) It follows from the examination of heats of formation of the relevant compounds that variations in other bond dissociation energies cannot be accounted for by the experimental resonance energies only. For example D(CH,-CH,) - D(R-R,) -+ [D(CH,-H) - D(R-H)] + [D(CH,-H) - D(R,-H)]. I would like to emphasize that this inequality arises from thermochemical data only i.e. the value of [D(CH,-H) - D(R-H)] + [D(CH,-H) - D(R,-H)] - [D(CH,-CH,) - D(R-R,)] is independent of the values of dissociation energies used in this expression. (v) I t has been assumed that the above-mentioned value the " strengthening effect ",26 is related to the length of the relevant R-R bond.For example its value for dibenzyl has been calculated as about 11 kcal./mole and i t has been suggested 2 5 ~ 26 that this value is related to the bond shortening in the molecule of dibenzyl. shortening of the central C-C bond in the molecule of dibenzyl. Since as has been said above the value of II kcal./mole is independent of the value of the C-H bond dissociation energy in toluene the uncertainty (iij) quoted in the communication by Dr. Bateman is irrelevant €or the problem of the C-C (vi) The relation between the C-H bond dissociation energy in toluene and the C-C bond dissociation energy in dibenzyl is given by the heats of formation of toluene dibenzyl and the H atom i.e.ZD(CeH5. CH2-H) - D(C6H5. CH,-CH,. C6H5) = ZAH,(H) -ZAHf(C6H5 . CH,) +AHI(C6HQ . CH . CH . C6H5) = 106 kcal. /mole. Hence the value of D(C6H5. CH,-H) = 77-5 kcal./mole,28 requires D(C6H5. CH,-CH . C6H,) to be 49 kcal./mole while the value of the latter C-C bond dissociation energy is experimentally estimated by Horrex and Miles 29 as 48 kcal./mole. If D(C,H5 . CH,-H) = 80 kcal./mole as Dr. Bateman suggests then D(C6H5 CH,-CH . C6H5) must be 54 kcal./mole i.e. the values D(C,H . CH,-H) = 80 kcal./mole and D(C6H5 . CH,-CH . C6H5) = 46 kcal./mole suggested by Dr. Bateman are incompatible. Dr. B. G . Gowenlock (Swansea) (communicated) Horrex and Miles postulate the reaction (e) PhCHCH,Ph -f PhCH=CHPh + H-(128 - Q) kcal. I among the reactions resultant upon the production of PhCHCH,Ph I radicals by the attack of benzyl radicals upon dibenzyl.This reaction is endothermic to the extent of about 58 kcal. being taken as 70 kcal. and therefore an activation energy of at least 58 kcal. will be required for this reaction. This activation energy is greater than that observed for the primary reaction. From the data given in Table I and assuming a 26 Szwarc J . Chem. Physics 1950 18 1660. 27 Szwarc and Taylor Trans. Furaduy SOC. (in press). 28 Szwarc J . Chem. Physics 1948 16 128. 29 Horrex and Miles this Discussion. GENERAL DISCUSSION I 230 normal value for the temperature independent factor ( 1 0 ~ ~ sec.-l) i t can be shown that for the lower temperature data (630-700’ C) decomposition of the PhCHCH,Ph radical will take place to the extent of only 10-50 yo.Similar coksiderations apply to reaction (i). Reaction ( f ) 2PhCHCH,Ph -+ PhCH=CHPh + PhCH,CH,Ph + (29- 128) kcal. on the other hand is exothermic and should have a much lower activation energy. In contrast to reaction (e) i t will produce only one molecule of stilbene per two PhCHCH,Ph radicals. Therefore on the basis of the authors’ reaction mkchanism a variation of the toluene/stilbene and toluene/styrene ratios should be obtained with variation in temperature. Investigation of the pyrolysis of mixtures of benzyl iodide and dibenzyl a t about 5 0 0 O C should therefore lead to the virtual elimination of re- actions (e) and (i) together with their dependent reactions (g) ( h ) and ( j ) and thus lead to toluene and stilbene as the sole reaction products in the ratio of 2/1.Dr. C. Horrex and Dr. S . E. Miles (St. Andrew) (communicated) In reply to Dr. Gowenlock we wish to point out that our analyses showed no significant variation in the composition of the products with the tem- perature of reaction. When introducing our paper we drew attention to the endothermicities of reactions (e) and (i) and pointed out that these decompositions would have to have normal temperature-independent factors of about 1 0 1 ~ in order to be significant in our conditions. We consider that the uncertainty in the value of Q makes i t unprofitable to pursue such calculations at present since an error of 4 kcal. alters the estimated rate of (e) or (i) at 630’ C by a factor of 10.Since the reactions appear to be feasible and a homogeneous chain sequence improbable we prefer to await the result of further work which one of us is carrying out. It must be noted that although the primary decomposition is not a heterogeneous reaction the experimental conditions favour access to the wall and we cannot at present assess the part i t may play in radical decompositions. The extent of a homogeneous disproportionation re- action which requires the encounter of two radicals is very dependent on the concentrations of the latter and hence on the rate constants of (d) (e) and ( 2 ) . It may be significant to note that the early work which reported stilbene and toluene as main products was done with much higher concentrations than we employed ; this would result in decreased accessibility of the wall to the radicals and reactions (e) or (i) which might occur there would be less prominent.Investigations in progress include work on the reaction of benzyl radicals with other molecules. With reference to the points made by Dr. Bateman we have stated that we evaluated the best straight line by use of the high partial pressure data since in our opinion the analytical precision in the other experiments was impaired by the small amounts of products available to us. We consider this procedure legitimate and a closer examination of our Fig. 7 shows that the spread of points in the data used is small (with the noted exception of one point which by any test must be regarded as a faulty experiment).The point concerning the revision of the length of the central C-C bond in dibenzyl has also been drawn to our attention by Prof. Cox. The amount of the shortening and its relationship to the strength of that link are of importance and in summarizing Szwarc’s arguments we may have given the impression that the strengthening is dependent OR the value for this shortening. Actually any “predicted ” value for the dissociation energy of the central bond in dibenzyl depends on the evalu- ation of the heat of formation of the benzyl radical. This has been done in several independent ways a summary of which has been given recently.30 30 Quart. Rev. 1951 5 42. 23 1 GENERAL DISCUSSION The agreement shown by these methods appears to us to be convincing evidence for the higher value for the resonance energy of the benzyl radical.Dr. W. A. Waters (Oxfovd) said In connection with the paper by Horrex and Miles on the pyrolysis of dibenzyl I should like to draw attention to the rather different conclusions of Dr. A. F. Bickel and myself 31 concerning the free benzyl radical which we prepared by de- composing w w'-azotoluene in boiling decalin solution at about 200' C. We found that even at this low temperature the benzyl radicals did not just recombine to form dibenzyl but underwent about 35 yo dispropor- tionation t o toluene and stilbene. We did not detect either benzene or styrene which may be secondary pyrolysis products of stilbene. Though our results could perhaps be attributed to a very rapid de- hydrogenation of dibenzyl by benzyl radicals i t was significant that the decalin solvent was not dehydrogenated at all.The alternative ex- planation would therefore require both an abnormally high lability of the C-H groups of dibenzyl and a very low probability of recombina- tion of benzyl radicals. Unlike Szwarc we see no fundamental objection to postulating the simple reaction . 2Ph. CH . + Ph . CH + Ph . CH. It is interesting also to note that the activation energy for the vapour phase dissociation of dibenzyl accords with the experimental data given by Ziegler 3 2 for liquid phase dissociations of many similar compounds some of which give radicals that undergo disproportionation at temper- atures as low as 1 5 0 O C.Dr. M. Szwarc (Manchester) (communicated) Dr. Waters suggests the possibility of disproportionation of benzyl radicals i.e. D(C6H5 . CH-H)-D(C,H The activation energy of this reaction should certainly be less than 5 kcal./mole if reaction ( I ) is to compete successfully with dimerization. This follows from the approximate equality of A factors for both dis- proportionation and dimerization (since the activated complexes for both reactions are very alike and in the liquid phase the deactivation of " hot " djbenzyl molecules cannot be the rate determining step). The activation energy of a process must be a t least equal to its endothermicity i.e. . CH2-H) < 5 kcal./mole D(C6H . CH-H) < 82.5 kcal./mole. CH3). On the other hand D(C6H5 .CH,-H) +D(C,H5 . CH-H) +D(C,H,-CH) +D(C-H) = 3AHf(H) + f AHf(C6H6) -AHf(C6H5. Taking D(C6H5. CH,-H) = 77-5 kcal./mole D(C,H5. CH-H) < 82.5 kcal/mole D(C-H) = 80 kcal. ; AH,(H) = 52 kcal./mole ; AH,(C) = L ( L being the heat of sublimation of carbon) and finally AH,(C6H5) ~3.68 kcal. /mole (the latter value being derived from a plausible assumption D(C6H5-H) m IOO kcal./mole) one derives D(CsH,-CH) > L - 28 kcal./mole. The heat of sublimation of carbon is probably not less than 136 kcal./mole and therefore D(C6H,-CH) > 108 kcal. /mole. The latter value appears to be much too high. It might be reasonable if the activation energy of (I) is about 25 kcal./mole but then this re- action could not be observed.3l Rec. truu. cbim. 1950 69 316. 32 Annalen 1942 551 161 ; 1950 567 134. GENERAL DISCUSSION 232 It seems to me that the results of Bickel and Waters could easily be interpreted on similar lines to those of Horrex and Miles. The reason why benzyl radicals do not dehydrogenate decalin seems to be simple the C-H bond dissociation energy in decalin is higher than D(CQH6. CHZ-H) and thus the activation energy for dehydrogenation is too high. Dr. C . Eorrex (St. Andrews) (communicated) In reply to Dr. Waters I think the products obtained from the interaction of benzyl radicals depend on the experimental conditions. The conditions in solution and in dilute gaseous systems are distinctly different and even in our work with gaseous systems we have observed differences in the proportions of the products when the only substantial change has been the pressure of the inert carrier gas.Dr. Ruth Lapage when using toluene as an acceptor for methyl radicals at 500 to 600" C in a flow technique with 3 to 8 mm. nitrogen found the main solid product was dibenzyl with a small amount of stilbene. With 600 mm. nitrogen however the product was stilbene. Mr. C. B. Cowan has obtained similar data with phenyl radicals and toluene at 600" to 700" C. In the toluene pyrolysis Mr. J. 0. McCrae and Mr. R. B. Cundall examined the solid products from an experiment a t 850°C when using the pyrolysis technique as published by Dr. Szwarc. By use of u.-v. absorption spectroscopy' they' found the solids contained dibenzyl stilbene and anthracene in the ratio 380 7 I.Thus Dr Szwarc is substantially correct in speaking of dimerization only under his conditions. It was also found that using 1-5 mm. toluene in 5 mm. nitrogen carrier gas gave even smaller amounts of stilbene and anthracene at 850" C. In toluene pyrolysis the benzyl radicals are formed at temperatures where dibenzyl would be rapidly decomposed and dimerization probably occurs at the exit of the furnace. Since the gases are cooled quickly there will be little opportunity' for attack by the radicals on the first fractions of the dibenzyl formed. In the work of Lapage and Cowan the temperatures required should permit dibenzyl to be formed within the furnace and remain unpFolyzed. If stilbene was formed from di- benzyl by radical attack it seems that higher inert gas pressures materially assist in the formation of the dibenzyl.If benzyl radicals disproportionate I would have expected the reaction to be rapid at the temperatures of toluene pyrolysis and more stilbene produced than has been found. Dr. Miles used w a'-azotoluene as well as benzyl iodide to provide benzyl radicals when examining the attack of these radicals on dibenzyl. The products in both cases were the same and similar to those obtained from dibenzyl alone on pyrolysis. I would expect the abstraction of a hydrogen atom from decalin by a benzyl radical to be a distinctly endothermic process and the data given by Dr. Kooyman in his Tables I and I1 support the idea that hydrogen abstraction from dibenzyl is an easier process.With reference to Ziegler's data I would like to add that his tem- perature independent factors are generally high. Prof. S . W. Benson (California) said The data of Horrex and Miles seem to indicate a steric factor which is unusually low for a unimolecular decomposition involving the rupture of a single bond and an activation energy lower than that which would be expected for a C-C bond if the proper length of the bond were employed. In addition the dependence of the rate on the surface/volume ratio shows a strong temperature effect. Thus the ratio of rates in a packed to unpacked flask fall regularly from a factor of 3 at 635O C to about 1-4 at 720° C . The authors make the assumption that the ratio of heterogeneous to homogeneous reaction is proportional to the surface/volume ratio.This is by no means necessarily the case and must certainly depend on the rate and energy of adsorption of reactant on the surface as well as on the overall GENERAL DISCUSSION 233 mechanism. In particular the dependence may vary from complete independence to direct proportionality according to the conditions. Further the pressure range investigated (about 0-1 mm. Hg reactant) is that in which wall collisions proceed a t a rate comparable to intermolec- ular collisions. In the light of these uncertainties i t would seem premature to accept with any certainty the value of 48 kcal./mole and the low frequency factor for the unimolecular homogeneous decomposition of dibenzyl. Dr.C. Horrex and Dr. S. E. Miles (St. Andrews) (communicated) Prof. Benson is in error in stating that in our conditions of 0.1 mm. Hg reactant pressure the wall collisions are about as frequent as intermolec- ular ccl'isions. He appears to be ignoring the presence of the carrier gas which impedes diffusion to the wall considerably. Calculation of the average number of collisions made by a molecule diffusing to the wall 33 gives about 105 for the pressures and reaction vessel used. By considering his comments on surface effects in conjunction with details given in his previous contribution on this topic we note that the effects of changes in surface/volume ratios are difficult to interpret where chains are initiated and end at the walls. In our case we see no reasonable evidence for a chain mechanism and consider the relatively small influence of the large change in surface implies little heterogeneity in the primary reaction.This is not in conflict with the possibility (mentioned in reply to Dr. Gowenlock) that some radicals produced in secondary reactions may reach the walls and undergo reaction there. Prof. Benson's assumption that the expected bond dissociation energy depends on the length assigned to the central C-C bond has been dealt with above. We agree of course that the temperature independent factor is unusually low in our experi- mental conditions. It seems to us that it is desirable to have present an acceptor other than dibenzyl for the removal of the benzyl radicals ; data on the thermal stability of potentially suitab!e substances is being completed.In view of findings on the pyrolysis of methyl iodide in this pressure range given later we are also checking to see if the high pressure limiting rate has not been attained. Dr. C. Horrex and Dr. Ruth Lapage (St. Andrews) (communicated) Some points disclosed by a detailed examination of the pyrolysis of methyl iodide seem to be of interest in connection with the kinetic difficulties in this pyrolysis work a t low pressures and particularly since C-H and other bond dissociation energies have been derived from published work on iodides. Butler and Polanyi 34 report five experiments on the pyrolysis of CH31 at 493'-495" C and by calculating first order constants and using log k (sec.-l) = 13 - E/4*57T they derived E = 54 kcal.With this value assigned to the C-I bond dissociation energy they deduced a C-H value €or CH which proved in agreement with later determinations by other methods. We have used the same techniques and conditions and found that their quoted values of k can only be realized a t temper- atures about rooo C higher. The use of the correct temperature in the above equation would give a value of E of about 60 kcal. and this seems too high for the C-I dissociation energy. The validity' of assuming first-order behaviour and a 1013 factor under such experimental con- ditions was therefore examined. With 0.07 to 0.3 mm. CH,I in 3 mm. nitrogen carrier gas the kinetics approached second order and changed towards first order above I mm.of CH31. In these conditions the re- action was mainly' homogeneous and not inhibited by iodine the free methyl radicals could be trapped by toluene but in its absence they were decomposed quantitatively' a t the wall to methane and carbon. The rate of reaction was increased by increase in pressure of the carrier gas. s3 Bursian and Sorokin 2. $lay&. Chew. B 1939 12 247. ,* Butler and Polanyi Trans. Faraday SOC. 1943 39 19. GENERAL DISCUSSION g mm. N Goo mm. N z mm. C,H 3 mm. N 1-1-3 mm. 3 mm. N 0.3 mm. 1-2 mm. CH,I CH,I 1-1.3 mm. CH31 0.0066 0.028 CH,I 0.0056 A A na rp A A 234 By using 600 mm. Nz z mm. toluene and 1-2 mm. CH,I we obtained rate constants showing a temperature variation given by k (sec.-l) = 2.6 x 1olS exp -547oo/RT.The strong influence of experimental conditions on the rates of decomposi- tion can be seen from the following values of k calculated on a first order basis. Conditions K (sec.-l) at M~ I + 0.2 MM. 1,) - FIG. 5. 0.0016 Soo0 K The presence of a radical acceptor was essential in experiments at high total pressures since with the decreased accessibility of the wall to the radicals a recombination reaction became important. An analvsis of our data has beeimade which considers activation of the CHJ molecules as occur- ring by collisions with CH31 or N mole- cules the latter having about one-fifth the efficiency of the former. Applying the con- ventional test for data on second- to first- order transitions of plotting the reciprocal of a first-order constant against the recip- rocal of the concentrations (appropriately weighted) we obtain Fig.5 The experi- mental point on the vertical axis clearly represents the high pressure limiting rate for the reaction. Details of this work are being prepared for publication. These findings while not altering the C-I bond dissociation energy make the initial derivation of the 54 kcal. value seem rather fortuitous. Similar troubles may exist with other substances in the Butler- Polanyi conditions although one might expect the fall off in first-order constants for more comdicated molecules to occur a t lower pressures than for CHJ. A number of C-H bond energy values which rest on the original iodide pyrolysis determinations may' be uncertain for these reasons.A 2.0 53 &A A A /;o Dr. Horrex and J. 0. McCrae (St. Andrews) said Dr. Kooyman may be interested in our preliminary results on the pyrolysis of diphenyl- methane which show that the products are essentially tetraphenylethane fluorene and hydrogen. The technique of investigation has been similar to that used for dibenzyl (see this Discussion). We assume the primary process is the breaking of the methylenic C-H link and that fluorene production may proceed through the planar diphenylmethyl radical radical. The calculation of the rate of this primary process via the hydrogen production (allowing for fluorene formation) yields a rate constant k = 10'~ exp (- 73ooo/RT).We are extending the range of experimental variables but suggest that this energy of activation can be considered as the C-H bond dissociation energy Prof. 6. W . Benson (California) said There seems to be some question in the work of Rowley and Steiner of the role played by each of the three possible simultaneous reactions cis plus cis ; cis plus trans and trans plus trans (butadiene). W e can write a general rate expression hobs = a2k + z a ( ~ - a)kd + (I - a)'LKtl where a is the fraction of the cis form and K, K,t and Kg are the specific rate constants for the individual reactions of the species indicated. If i t GENERAL DISCUSSION 23 5 is assumed that the active complex is the same for all of these cases then it can be easily shown that the observed activation energy is not dependent on the extent to which the different species react since a = e -AE/RT ; E, = Ell - zAE = E,# - AE and the temperature dependence of the 01 exactly cancels the different temperature dependence o f the different rate constants.In this case the above expression reduces to kobs = ktt = a'ku. If however there is more than one possible complex for the system and the steric factors are different for the different isomers then we should expect a different contribution at different temperatures from the different species. Thus i t may be quite possible that a cyclic and linear complex both exist having different activation energies and different frequency factors. In this case the system becomes much more complicated and alternative explanations must be considered for interpreting the data.Dr. M. Magat (Paris) said There is one point that strikes me as not being taken enough into account in all theoretical discussion of butadiene cyclization reaction. It is the fact pointed out by Aston Szasz Wooley and Brickwedde 35 that the trans configuration is 2.5 kcal./mole more stable than the cis configuration. On the other hand at least one of the two molecules must have the cis configuration in order to make the cyclic complex possible while the linear complex can be realized with any one of the two. Mr. B. Eisler and Dr. A. Wasserman (London) (communicated) Experiments have been carried out which enable a comparison to be made of the activation energies E of Diels-Alder diene associations of the following type (1) Butadiene + dienophile -+ adduct Cyclopentadiene + dienophile + adduct .* (11) The difference between E and EII can be accounted for,ss if it is assumed that the formation of the transition state of (I) involves the conversion of trans-butadiene into the cis form. In the reaction discussed by Steiner (14 2 Butadiene -+ vinylcyclohexene 2 Cyclopentadiene -+ dicyclopentadiene and collaborators. . * the dienophile of (I) is butadiene. A comparison of the E values of (Ia) and (IIa) indicates that both butadiene molecules are converted into the cis form before the transition state of the reaction (Ia) is taken up. . . . (IIa) This is due firstly to steric requirements and secondly to effects which operate generally in Diels-Alder reactions and which bring about the formation of relatively closely packed transition Dr.H. Steiner (Petrocarbon Ltd. Manchester) (communicated) I n reply to the remarks by Benson Magat and Wasserman a statistical rate calculation of the butadiene dimerization using the values for the thermodynamic functions of butadiene obtained by Aston Szasz Woolley and Brickwedde,s* may give an answer to some of the questions raised. Since the interconversion of trans- into cis-butadiene is endothermic by only 2-5 kcal. /mole whereas the activation energy of the dimerization is some 25 kcal./mole the initial state which has to be assumed in such a calculation is most likely' an equilibrium mixture of trans- and cis- butadiene at the appropriate temperature.It may be that the deviations from the Arrhenius function which we have observed are connected with the gradual shift of the cis-trans-butadiene equilibrium particularly at low temperatures. Unfortunately our statistical calculations were carried out using approximate values for the partition function of buta- diene only. I hope to recalculate this prablem using the data of Aston 85 J . Chem. Physics 1946 14 67. 36 Detailed considerations and experimental results to be reported elsewhere. 37 J . Chem. Soc. 1935. 833 1512a8 1936 432 ; Trans. Faraday SOC. 1936 J Chem. Physics 1946 14 67. 35 841. D. ROWLEY AND H. STEINER 213 GENERAL DISCUSSION 11. HYDROCARBON REACTIONS A.THERMAL REACTIONS Map I point out rather an interesting correlation between results quoted in papers by Sir Cyril Hinshelwood and Prof. Glockler ? The latter shows that the C-H and C-C bond energies in n-paraffin chains increase in passing from the centre towards the ends ; the former states that in the thermal chain fission of n-paraffins the probability of C-C fission (and Dr. P. Torkington (Brit. Rayon Res. Assoc.) (communicated) 214 GENERAL DISCUSSION hence also of C-H fission accompanying H-transfer) also increases to-wards the ends. I suggested earlier in this Discussion that Prof. Glockler’s results might be interpreted as evidence for the existence of molecular orbitals (overall but not necessarily delocalized in the Sense of implying electron mobility) favouring the chain ends.The correlation here implies that the presumably higher terminal electron density is associated with greater activity i.e. electron mobility an interesting negative analogy with true bimolecular reactions. Regarding the concentration of energy via vibrational coupling I should like to draw attention to another previous remark of mine following Prof. Ubbelohde’s paper. The envisaged concentration of energy could surely never occur with “ perfect ” coupling ; it seems that if the occur-rence of energy concentration by vibrational coupling is accepted as proved then so also must the occurrence of two or more collisions pro-viding the energy sources. Transfer from two or more such sources by coupling might well lead to a concentration of energy favourable to re-action by’ normal interference.It seems quite probable that a long-chain molecule might be involved in several simultaneous collisions but against fundamental principles for a concentration of energy to arise from a single collision. Finally is it possible for a little more non-mathematical commonsense to make a little more headway against a little more formalism ? How does symmetry lead to electro delocalization 9 Why because one can write functions 4 = ,Zai$i because of symmetry does an electronic system become mobile ? Perhaps too much n-bond theory has preceded a study’ of normal molecules. In an unsymmetrically-substituted benzene de-rivative there is no symmetry but still the same delocalization as in the parent compound. Dr.G. Gee (Welwyn Garden City) said I should like to outline a possible explanation of the observed thermal decomposition products of paraffins. Considering first radical mechanisms to a first approxima-tion the thermal dissociation of a long paraffin chain will occur randomly at any link. The resulting radicals will then tend to degrade by the facile process of losing a series of ethylene molecules from the end, and this process will continue until a methyl or ethyl radical remains. Thus primary dissociation of long chains will result mainly in the produc-tion of ethylene and ethyl and methyl radicals and the ratio of these primary products will not be markedly temperature dependent. The most probable secondary reaction of a methyl or ethyl radical appears to be the dehydrogenation of a paraffin molecule a process which again will occur nearly randomly along the chain (3) and will be followed by dissociation of the resulting radical (4) CH3* + CH,(CH,)nCH + CH + CH,(CHa),*CH(CHa),-,,CH ( 3 ) CH3(CH2).*CH(CH2),-,,CH3 + CH,(CH,),CH=CH,+CH,(CH,),-, (4) The situation now becomes more complex in that the olefin produced will be more reactive toward further methyl and ethyl radicals than is the original paraffin.Without seeking to follow o u t these processes in detail i t can be seen that a further series of degradative processes will follow from the secondary radicals in which the principal final products will again be ethylene methane and ethane. Thus i t is easy to see why these products predominate and also why their relative yields are largely independent of conditions.It also follows from these considerations that in presence of nitric oxide the main hydrocarbon product to be expected from radical dissoci-ation is ethylene since nitric oxide would compete more successfully with The point is obvious and not worth developing. CH,(CHz)nCH + CH3(CHa)z + CH,(CHa)L - - ( 1 ) CH3(CH,),* + CH3(CH2)g2 + CH*CH2 * (2 GENERAL DISCUSSION 215 (3) than with ( 2 ) . The fact that methane and ethane are still produced in presence of nitric oxide therefore argues in favour of Stubbs and Hinshelwood’s contention that a molecular reaction occurs. The mechanism of this reaction does not appear very clear but it may be plausibly sug-gested that ethylene could be eliminated from the chain at any point through the formation of a cyclobutane type of activated complex : CH,-R CH,R I -+ I + CHeCH,+R-R .- ( 5 ) CH,-R CH,-.R, where R R are alkyl radicals or hydrogen. It is not a t first sight clear why this reaction should lead to methane and ethane rather than to larger paraffins but it appears likely that increase in the size of the radicals R and R would lead to steric difficulties. If this is indeed the case each successive step in the degradation of a long chain (by removal of ethylene) will lead to a more readily dissociable paraffin and-as in the radical decomposition-the principal final products will be ethylene methane and ethane. Dr. F. J. Stubbs and Prof. C. N. Hinshelwood (Oxford) (co.mununi-cated) In connection with Dr. Gee’s interesting suggestions we may state for comparison the views which we have formed about the reason for the similarity of products from a chain reaction and a molecular process.For the typical case of n-hexane the molecular process would be CH,CH,CH,CH,CH,CH -+ CH + CH CHCH,CH,CH, where El and El’ are the activation energies of the two reactions. Initia- CH,CH,CH,CH,CH,CH 3 X + Y (1) (1’) (2) CH;+CH,CH,CH,CH,CH,CH +- CH4+’CH,CH2CH,CH,CH,CH, (2’) C,H; +CH,CH,CH,CH,CH,CH -f C,H,+’CH,CH,CH,CH,CH,CH, where e, e, etc. are the respective activation energies. El CH,CH,CH,CH,CH,CH -z C,H + CH CHCH,CH, El’ The various steps in the chain reaction would be tion { X+CH,CH,CH,CH,CH,CH + XH+’CH,CH,CH,CH,CH,CH, DCH,CH2CH2CH,CH,CH -+ CH;+CH CHCH,CH,CH, *CH,CH,CH,CH,CH,CH -+ C,Hi+CH CHCH,CH, With ling chains the ielative proportions of methane and ethane formed will depend only on the repeating steps ( I ) and (1’) in the above scheme and their sequels ( 2 ) and (2’).The nature of the initiating re-action is unimportant for this purpose. For the products from the molec-ular process and chain reaction to be in the same proportions all that is necessary is that (&-El’) should equal (el-el’). This seems quite likely since the alternative steps in the molecular process are closely parallel with the steps (I) and (1’) respectively the two positions of rupture being in the same relative position in the radical CH,CH,CH,CH,CH,CH; as in the molecule CH,CH,CH,CH,CH,CH,. Dr. M. Szwarc (Manchester l7niversit-y) said Stubbs and Hinshelwood proposed in their paper an interesting mechanism for a unimolecular decomposition.I would like to examine some of the conclusions which, it seems to me follow from their assumptions. Let us consider a system composed of two components viz. an inert gas X and a reacting gas M which may decompose unimolecularly. Furthermore let u s choose the experimental conditions in such a way that the energy supplied to molecules M arises from collisions between M and X. According to Hinshelwood and Stubhs we can observe two kinds of collisions. (if One type of collisions between M and X lead to the formation of energized molecules M* i.e. molecular species which contain an amount of energy at least sufficient for their decompositio 2 16 GENERAL DISCUSSION but distributed unfavourably amongst various internal degrees of freedom.This process we represent by the equation X + M + M* + X (collision of the 1st type). The decomposition of energized molecules requires a reshuffling of their energy which has to be accumulated eventually in the reacting centre, i.e. M* +- MS + Products, MS denoting an " activated " molecule i.e. a species which contains at least the necessary amount of energy localized in the reacting centre. Although collisions of the first type are frequent the probability for the energy redistribution is low and consequently the reaction caused by these collisions attains a first-order character at comparatively low pressure of the gas X. Plotting the rate of this decomposition as a function of partial pressure of gas X we obtain the curve I shown in Fig.I . (ii) Collisions of the second type lead directly to the formation of I ' activated " species MS X + M -t MS + X (collisions of the 2nd type). Since the chance of the occurrence of such a collision is very low and the probability for the subsequent decomposition of Mt is very high, the reaction caused by collisions of the second type retains the second-order character up to comparatively high pressures of the gas X. Plotting the rate of the latter reactions as a function of pressure of the gas X we obtain the curve 2 shown in Fig. I. FIG. I. Fig. I illustrates the features of the decomposition claimed by Hinshelwood and Stubbs. There is a comparatively low pressure region (C-D) in which the first type of collision predominates and in which the reaction obeys first-order kinetics.There is a comparatively high pressure region (E-F) in which collisions of the second type are predominant and conseqfiently in this region the reaction obeys second-order kinetics. However in addition to these two regions there is a very low pressue region (A-B) in which both types of collision call for a decomposition of the second order and finally there is a very high pressure region (G-H) in which both types of collision demand first-order kinetics of decomposition. If we plot therefore the order of reaction as a function of the partial pressure of gas X we obtain a curve represented in Fig. 2. I would like to ask Prof. Hinshelwood if he agrees that this behaviour of decomposition is a consequence of his assumption.There is however another point which I consider to be important. If the reaction takes place in a very high pressure range say G-H then the first type of collision must lead to an equilibrium between the energized molecules M* and the ordinary molecules M i.e., X + M + M* + X (collisions of the 1st type) GENERAL DISCUSSION 217 I n the same pressure region collisions of the second type must lead to an equilibrium between the " activated " molecules MS and the ordinary molecules M i.e. X + M $ MS + X (collisions of the 2nd type). One has to realize that any type of " activated " molecules MZ can be produced either by collision of the 1st type via M* or directly by collision of the 2nd type. If the latter modes of activation leads to the state of equilibrium between ordinary molecules M and activated species MZ , then i t would be expected that the activation through collisions of the first type should also lead to an equilibrium between these two species (this follows from the fact that both types of collisions occur in the same gas phase).Consequently the following scheme should be valid for a reaction resulting from collisions of the first type : M + X + M* + X Jr MZ -f Products, and the equilibrium concentration of MZ should be the same as in the case of collisions of the 2nd type. That however would contradict the as-sumptions on which Fig. z was obtained. I Px FIG. 2. One can argue that the collisions of the second type do not lead to an equilibrium concentration of Mi since their rate of decomposition is too high being of the order of 1013 (MS) sec.-l.This seems to be quite correct. Indeed collisions of the second type can never be of any significance and if they have to maintain a first-order character of decomposition a pressure of the order of 105 atm. would be required. It seems that the best method of approaching the problem is t o sum up aZZ the contributions arising from all possible collisions (appropriately weighed). The reaction would be of the second order a t low pressures and would attain the first-order character at sufficiently high pressures In the latter case the activation energy of decomposition would be given by the activation energy of the process M +M$. Dr. F. J. Stubbs and Prof.C. N. Hinshelwood (Oxford) (communic-ated) Given the hypothesis that the nitric-oxide inhibited reaction is a definite molecular reaction (which as we have explained there is a good prima facie case for accepting) the changes of order referred to by Dr. Szwarc are the basic experimental facts upon which our views about the kinetically composite nature of the reaction are founded. The tentative interpretation which we have advanced is not in any way built upon a preconceived theory and may be modified in the light of further experi-ments now in progress. In the meantime we would prefer not to regard the facts about the order as deductions from a theory as Dr. Szwar 218 GENERAL DISCUSSION implies but as observations calling for explanation. What we have suggested is that the normal theory requires some degree of amplification.Prof. S. W. Benson (California) (conznzunicated) The data presented by Ingold Stubbs and Hinshelwood and quoted here seem to indicate that even in the " fully inhibited " pyrolysis there is a surface effect which increases with decreasing temperature and is virtually independent of pressure. It seems to me that until this effect is further elucidated i t is difficult to accept the discussion and rate constants quoted as anything but tentative. There still remains the possibility that there is a heterogeneous reaction initiated on the walls which starts chains which also end at the wall. Rice and Herzfeld 1 have recently shown that under such conditions i t is possible that depending on the mechanism the rate may show any de-pendence on the surface volume ratio between complete independence and direct proportionality.Even in a direct decomposition to molecular products i t is conceivable that heterogeneous decomposition competes with homogeneous decom-position. In such case the activation energy of the two would be ex-pected to be different and since the relative amount of each would be expected to show a monotonic if not simple pressure dependence there would be observed an effect of pressure on experimental activation energy. For n-pentane i t is difficult to believe that when the experimental activation energy is found to vary between 80 and g o kcal./mole in the pressure range 50-150 mm. Hg (inhibited reaction) t h a t the free radical mechanism of splitting of the C-C bond which has an activation energy of 80-85 kcal.is not proceeding at a competitive rate. Finally it is difficult to reconcile the observed decrease of activation energy with increasing pressure with any reasonable behaviour of mole-cules. In the Rice-Ramsperger-Kassel theories of the unimolecular decomposition the activation energy is supposed to decrease at the lower pressures where the higher energy states can no longer make their com-plete contribution to the overall rate. The explanation offered by the authors to the effect that there may be a single mode which requires a very long time to reach activation even when the energy is present in other modes seems completely ad hoc and unsupported by any other experimental or theoretical evidence.It also leans upon an interpreta-tion of a molecular collision that seems in immediate conflict with the law of microscopic reversibility. That is there seems to be an implicit assump-tion that an active molecule can only lose energy on collision and never gain it. If there is a mode which is only slowly activated then the chances are that i t is just as slowly deactivated and collisions do not necessarily play any decisive role in the time of decomposition. The example of ethane where the kinetics seem to go through a second to first order transition while the activation energy changes but little (or possibly is higher a t the lower pressures) seems a striking example of the above-mentioned contradiction. I doubt very much that these pyrolyses will be very tractable theoreticallv until the nature of the surface efiects and the inhibition axe better understood.Dr. F . J. Stubbs and Prof. C. N. Hinshelwood (0,vford) (conz~lzunic-nted) Chain reactions which are initiated by a decomposition are not at all likely to start and stop at a wall. If the molecule dissociates on the wall radicals are likely t o remain adsorbed. In any case chain reactions staxting at a wall should be subject to the usual kinds of inhibition just like any other chain reactions. This assumption of Prof. Benson seems to us to be improbable. In pentane we have always believed the chain reaction normally to proceed a t a " competitive rate " but we state a prima facie case for concluding that We are unable to understand his fourth paragraph.1 Symposium on Abnormalities in Reaction Kinetics Amer. Chem. SOC., Minn. hfinn. Spring 1950 GENERAL DISCUSSION CH3-CH3 CH3. CH,-CH3 CH3. CH,-CH,. CH3 CH . CH,-CH . CH . CH, 219 80 CH CH-H I00 80 60 < 60 1 slightly ( 8 0 CH . CH . CH,-H 1 due to second- CH CH . CH,-CH3 J order effects CH CH . CH,-CH . CH3 ’lightly this competing reaction has in the conditions of the experiments been suppressed. In his penultimate paragraph we do not recognize our views very clearly and as to the last paragraph we can assure him that our approach continres to be essentially experimental and that all the details of these admittedly rather puzzling phenomena are under examination still and from several new points of view. Dr. L. Bateman (Welwyn Garden City) (communicated) The form and relative position of the paraffin and olefin decomposition rate curves in Fig.I of Stubbs and Hinshelwood’s paper are just as would have been predicted on the basis of A H changes on breaking the most easily broken bonds so that some reference to this mode of primary bond scission is perhaps of interest. The AH changes (in round figures) for the C,-C, compounds are summarized in the following Table. Ethylene should be I I much more stable than ethane propane and propylene should be of similar stability and the higher members of the two series should exhibit a difference reflecting the partial utilization of allylic resonance energy in the bond-breaking process. These features are evident experimentally, but ethane and I-butene appear to be relatively more stable than might be expected and may serve to illustrate how secondary processes affect the overall decomposition.The common factor is the primary formation of methyl radicals and not higher alkyl radicals capable of ready decom-position into smaller fragments. Hence in ethane recombination will be specially favoured since the only alternative reaction is the rather difficult hydrogen exchange Me + C,H + MeH + C,H, In I-butene the corresponding reaction to the latter (E N 10-15 kcal.). Me + CH CH . CH,Me -f MeH + CH CH . CH . Me ( E - 5 kcal.) 1 is much more facile but the resulting substituted allyl radical like that initially produced will be far less active than an alkyl radical in inducing further decomposition.An important aspect of these considerations is that all A1-olefins are required to yield simple allyl radicals as primary fission products. The reactivity of these radicals under the decomposition conditions their ultimate products and possibly the estimation of their extent of forma-tion might well be stcdied experimentally by carrying out co-decomposi-tions with I 5-hesadiene as a foreign source of allyl radicals. Informa-tion on the third point would establish the contribution of allylic bond primary scission to the evidently composite overall decompositi m. Dr. R. G. Partington (Oxford) (communicated) In their paper Stubbs and Hinshelwood state that Steacie and Folkins suppnrt the view that the reaction taking place In the presence of sufficient nitric oxide to reduce the rate to a minimum (the “ residual reaction ”) is “ the primary process of what in the absence of inhibitors would be a chain reaction ”.This is, I think a mistaken interpretation of Steacie and Folkins’ conclusions, Steacie and Folkins Can. J . Res. B 1940 18 I 2 20 GENERAL DISCUSSION which read '' the addition of nitric oxide then merely diminishes the chain length without completely suppressing the chains ". Had the suggestion been made that all the initidly formed radicals reacted with nitric oxide one would have to consider the possibility of the formation of such a molecule as (I) in the following scheme e.g., CH,-CH,-CH,-CH,-CH3 = CH,+CH,-CH2-CH2-CH3 CH3+N0 = CH3N0 CH3-CH2-CH,-CH,+ NO = CHs-CH~-CH~-CH2-NO (1) which might well participate in further reaction (starting by its decom-position) in a similar way to the large radical itself.A further paper by Steacie and Folkins 8 is of interest in this connection. It is concerned with the decomposition of n-butane initiated by free radicals from ethylene oxide and the authors conclude that their results " cast some suspicion upon the idea that maximum inhibition by nitric oxide in all cases corresponds to complete suppression of chains *'. The whole problem of the action of nitric oxide as ail inhibitor (and the foregoing remarks were largely included to justify the opinion that the problem still exists) seems to depend on the relative rates of the two reactions viz., reaction of radicals with the hydrocarbons e.g., and reaction of radicals with the inhibitor e.g., A C R + RH = R,H + R .* (1) R + NO = R,NO. . * (4 Reactions of the type (I) have recently been studied and similar specific attention to reactions of the type ( z ) , would undoubtedly be of the greatest value in helping to solve the above problem and might be less subject to ambiguity than experiments on a system in which reactions of this type are among the very many taking place. There are many possible sources and convenient methods of producing the appropriate radicals either thermally or photochemically and in connection with the possibility of making such a study some experiments of mine on the chemical analysis of small concentrations of nitric oxide in gas samples of the size usually handled in work of this character might be of value.A gas sample (say 50ml. of gas at N.T.P.) containing say Q to I yo of nitric oxide is enclosed in the holder A (Fig. 3 ) . This is then con-nected to a flask B containing air and small amount of an aqueous solution of sodium or potassium hydroxide free of nitrite. On opening the wide-bore stop-cock C and gently shaking the reactions FIG. 3 . 2NO + 0 = 2N0 . * (3) (NO + NO,) + 2NaOH = 2NaN0 + H,O . - (4) take place. Reaction ( 3 ) is slow and the conversion to nitrite is almost complete. The solution of nitrite can then be analyzed with great accuracy by the standard colorimetric method^.^ The conversion factor NO -f nitrite is reproducible for a given nitric oxide concentration and its value could be determined over the desired range of concentrations by making up mixtures.Can. J . Res. B 1939 17 gg. 4 Trotman-Dickenson and Steacie J . Amer. Ckem. SOC. 1950 72 2310 ; 6 See for example Vogel A Text Book of Quantitative Inorganic Analysis J . Chem. Physics 1950 18 1097 and following papers. (19391 P. 700 GENERAL DISCUSSION 22 I Dr. A. F. Trotman-Dickenson (Manchester) said I am not clear why Anderson Davison and Burton have disregarded the recent work of Dodd in selecting 8.6 kcal. as the best available value for the activation energy of the reaction CH + CH,CHO -f CH + CH,CO. Dodd has shown that the results of all previous workers (including those of Rollefson and Grahame) may be interpeted as giving a value for the activation energy of 10.7 f 0.5 kcal. This value makes the agree-ment between the acetaldehyde and acetone results much worse (15.8 and 13.2 kcal.respectively). Dr. Steacie and I recently reviewed the data on the reaction CH + H +CH4 + H and concluded that the best work (three separate determinations) gave El = 8.8 kcal. steric factor = 4 f 2 x I O - ~ (oR2 = 2-8 A ucas = 3.5 A) and K / K t at 182O C = 50 x 10-1 molecules-* cm.812 sec.-4. The results of Anderson Davison and Burton give K,/kt2 at 182O C = I x 10-1, molecules-* ~ m . ~ / e set.-* if the same collision diameters were used as above. No satisfactory reasons for these serious discrepancies have been given. Prof. M. Burton (Notre Dame) (communicated) Mr. Trotman-Dickenson mLst not have realized that the full paper by Dodd did not reach our laboratory until after the paper by Anderson Davison and myself was submitted for publication.Otherwise we would certainly have commented on it. Our actual work was concerred with establishment of activation energy differences. involved also assumptions of reliability of E values of Grahame and Rollefson for and of Trotman-Dickenson and Steacie for Since the agreement of values of El based on the two separate assumptions was good one conclusion was that the two sets of results were consistent. The more recent value of Dodd (E20 = 10.7 f 0.5 kcal.) is simply not reconcilable with the older value (8.6 kcal.) of Grahame and Rollefson. It is consistent with the results of Trotman-Dickenson and Steacie only if it is assumed that our measurements of E differences are inaccurate to about 2 kcal.Never-theless the new evidence presented by us does support a high value of El and Dodd’s results emphasize that fact. Regarding the relative merits of the results of Dodd and of Grahame and Rollefson we have no firm opinion. However we may note that E, >El* is a new idea. The reverse order of activation energies ac-cepted prior to the work of Dodd is consistent with the usual notion that, since in reaction (20) a formyl H is involved while in reaction (18) it is a methyl H reaction (20) probably has a lower E,. Mr. Trotman-Dickenson’s remark concerning discrepancies between values of kl/kl24 from old experimental data and from calculation involving our El and s emphasizes the dangers inherent in acceptance of old cal-culations based on experiments which may contain innate error e.g.the general difficulty of interpretation of experiments of involved mechanism. Prof. W. A. Noyes Jr. (Rochester) said It is interesting to note in the paper by Prof. Burton and co-workers that for the reaction Calculation of E for CH,+H + CH4+H . - (1) CH + CH,CHO -+ CH + CH,CO . - (20) CH + (CH,),CO -+ CH + residue. * (18) We are cognizant of the difficulties of this work. CH + H = CH + H . (1) Dodd J . Chem. Physics 1950 18 234 ; Trans. Faraday SOC. 1951 47 56. 7 Trotman-Dickenson and Steacie J . Physic. Chem. (in press) 222 GENERAL DISCUSSION the activation energy is 13-14 kcal. and the steric factor about I O - ~ . These results are based on studies in which reaction (I) could compete with reactions of the type The activation energies for (2) depend on R but generally fall in the region of 8-10 kcal.(e.g. if R is CH,COCH,- at about 9-7 kcal.). Thus if steric factors for (I) and (2) were identical (2) would proceed about 102 times as f a s t as (I) at equal pressures of H and of RH at temperatures of 200-300° C. A ratio of rates as high as 102 would invalidate most conclusions due to attendant experimental difficulties. The very fact of successful com-petition under experimental conditions used by most authors indicates quite strongly that steric factors for (2) are usually 10-1000 times lower than for (I). This fact and other related facts raises questions about “ hot ” radical conclusions. If a radical separated from a parent molecule by absorption of energy much greater than that necessary to break the bond retains energy solely as kinetic energy i t could hardly undergo 103-105 collisions without being reduced essentially to thermal equilibrium with its sur-roundings.One must conclude either that hot radical effects would be unobservable or that the so-called steric factor is a very pronounced function of kinetic energy. Theory concerning this matter is very vague and qualitative although one is led to predict an increase in steric factor with increase in kinetic energy. An examination of published data on hot radical effects for reactions of the type of (2) reveals that positive conclusions are not warranted. Uncertainties exist either due to scatter in the data themselves or to effects other than hot radicals which afford equally satisfactory explanations.This is true particularly in those cases such as CH,COCH3 and Hg(CH,), where the full nature and yield of the primary photochemical process have not been elucidated as a function of temperature and other variables. CH + RH = CH + R. . * (2) A word should be said also about the reaction CH + CH = CpH6. . (3) All workers seem agreed that activation energies for radical combination reactions of the type of (3) are very low. Less agreement is found con-cerning the steric factor but the safest conclusion seems to be that i t is high and probably near unity. The necessity for a third body has not been clearly shown but an examination of the data a t sufficiently low pressures will indicate a trend which might show a third-body effect, More data on this point will be found in a forthcoming article by Dr.A. J. C. Nicholson as well as in a recent article by Gomer and myself.* A third body may be introduced in several ways but i t is not satisfactory to introduce a mere triple collision. One way would be in which case the Under conditions reduces to ks 1 4 CH + X c Z CH,X . (4) CH,X + CH = C,H6 + X * ( 5 ) rate of ethane formation would be of high radical concentrations (high intensities) eqn. (6) an equation which is satisfactory for part of the ethane formation during photolysis of mercury dimethyl but which is not satisfactory under con-ditions so far studied in acetone. The trend in acetone at low pressures may indicate the necessity for a three-body collision.J . Amer. Chem. SOL 1950 71 101 GENERAL DISCUSSION 223 The most satisfactory way to obtain the right rate equation for the majority of experimental conditions is that used by Gomer and others, viz. the reversible formation of an intermediate complex from two methyls, the intermediate complex being stabilized to ethane by collisions. At sufficiently high pressures the rate-controlling step is the formation of the intermediate complex a step whose rate depends only on (CH3)2. One comes to the conclusion therefore that the nature of the third body may determine the way in which i t acts in agreement with the classical experiments of Rabinowitch and Wood on the recombination of bromine atoms. Under most conditions the rate of ethane formation is given by in which h may be a complex constant which indicates nevertheless the rate of (3) to be very high.Dr. P. Torkington (Brit. Rayon Res. Assoc.) (corutmuvzicated) Is the steric factor temperature-dependent ? If so its variation might possibly be related to the proportion of molecules (in this case radicals) undergoing a mode of vibration more favourable to reaction than other modes. Thus, in the reaction CH3- + H, it might be thought that approach along the line C . . . H-H coinciding with opening of the CH " umbrella " at the correct stage (the hydrogen molecule lying on the opposite side to the hydrogen atoms of the methyl group) would have a high probability of successful reaction. I n this case then the favourable vibration is the symmetrical breathing frequency of the methyl group.The argument is not essentially altered if the group is planar as has been fairly recently suggested though the favouring would not be so noticeable. Possibly the principle could be applied generally. As regards feasibility with a frequency of 1500 cm.-l and amplitude of the order 10-g cm. the hydrogen atoms in the breathing mode of a methyl group have a mean velocity of the order 105 cm./sec. ; the root-mean-square velocity of hydrogen molecules at 2 5 O C is about 2 x 105 cm./sec. The example shows that vibrations might co-operate. Dr. George Porter (Cambridge) (communicated) The high value for the bimolecular rate constant of methyl radical combination at room temperature of about 5 x 1013 cm.a mole-I sec.-l quoted by Sztvarc and others in the discussion now seems well established having been obtained by several workers using the sector method There appears a discrepancy, however with the results of mirror experiments which cannot be dis-regarded on the grounds that radical removal under these conditions is known to be a wall reaction.Whatever the mechanism by which the radicals disappear it is the rate-determining one and thus sets an upper limit for the homogeneous bimolecular rate constant. If the concentration of radicals is low as in the photochemical experi-ments of Norrish and Porter O the rate constant is found to be high and no discrepancy appears but as the concentration is increased so is Kmax in-creased for the radical lifetime does not decrease proportionately.Thus Forsyth's values 10 give kmax = 6 x 1012 cm.3 g . mole-1 sec.-l and although other workers have not recorded concentration measurements specifically i t seems certain from a consideration of mirror removal times that even higher concentrations were involved in the experiments of Rice Johnston and Evering l1 and of Paneth Hofeditz and Wunsch l2 who also increased the life-time to 0.1 sec. by decreasing the rate of the wall reaction. As the rate-determining reaction is heterogeneous i t appears that the velocity of the bimolecular gas phase reaction determined in this way is consider-ably less than that corresponding to unit collision efficiency. Norrish and Porter Faraday SOC. Discussions 1947 2 97. Forsyth Trans. Faruday Soc. 1941 37 312. RQG = k(CH,)' * (8) l1 Rice Johnston and Evering J.Amer. Chem. Soc. 1932 54 3529. la Paneth Hofeditz and Wunsch J. Chem. Soc. 1935 372 GENERAL DISCUSSION The value of kmax is obtained from t* and the concentration the latter being determined from the weight of mirror removed or metal alkyl formed in unit time which can be measured to within 10 yo without great difficulty. The half-life is also a reasonably accurate determination and the well-known difficulties of the method such as mirror poisoning and irreproducible wall conditions can produce only a scatter and not a general trend of this kind. It appears therefore that this discrepancy must be considered seriously unless a reason can be given for doubting the basic assumptions of the mechanism of mirror removal. The bimolecular rate constant must eventually decrease at low pressures when the collision life becomes greater than the lifetime of the collision complex.A possible explanation therefore lies in the lower pressure of the mirror experiments which is about I mm. whereas the pressure in the sector experiments was usually much higher and always above 5 mm. Dr. E. J. Bowen (Oxford) said The importance of the amount of delocalization energy in the products of a dissociation affecting the ease with which it occurs is also shown by work carried out by Miss Rohatgi on the photochemical reaction of anthracene with liquid chlorinated hydrocarbons. The following quantum efficiencies were found : In light absorbed by anthracene the reaction is AX + RC1 -+ ARC1 (derivative of dihydroanthracene).Carbon tetrachloride CC . 0'4 Pentachlorethane CHCl,CCl . . 0.27 Tetrachlorethylene CCl,=CCl . . 0.07 Tetrachlorethane CHCl ,CHCl . 0.018 Trichlorethylene CHC1-CC1 a . . 0.008 Ethylene dichloride CHCl=CHCl . . 0.0076 Chloroform CHCl . 0.0032 High quantum efficiencies appear only where the CCl radical might be involved. Prof. S. W. Benson (California) said It seems rather strange that the Br atoms liberated in the initial step do not undergo addition and hydrogen abstraction reactions with the cyclohexene. I wonder if Dr. Robb has any information on this point. Dr. E. C. Kooyman (Amsterdam) (communicated) It seems likely that the main differences between the mechanisms of the addition of bromotrichloromethane and that of carbon tetrachloride to olefins arise from the far greater reactivity of the bromine atom in CBrCl as compared with that of the chlorine atoms in CCl,.Kharasch and Friedlander la found bromotrichloromethane to react rapidly with styrene at 50°C under the influence of ultra-violet light. However little reaction occurred at 20' C ; at this temperature styrene was found to inhibit the addition of CBrCl to other olefins. These facts were interpreted by Kharasch and Friedlander as resulting from the low reactivity of the benzyl type radical formed by preferential addition of CCl to the styrene double bond : -CCl + CH,=CH-Ph -+ CC1 ,CH,CH-Ph. These data suggest the abstraction of a bromine atom from CBrCl, to be a slow reaction at low temperatures when the attacking radical is re1 atively stable.In detailed analyses of the reaction between cyclohexene CCl and benzoyl peroxide at 78" C Kooyman and Farenhorst l4 found 60-70 ?& of 3 3-dicyclohexenyl on the basis of peroxide decomposed. In our in-vestigation on a-methylenic reactivity l5 the " retardation constant '' for cyclohexene was found to be 11 x I O - ~ ; in view of the value 0 - 2 reported l 3 Kharasch and Friedlander J . Org. Chem. 1949 14 239. l4 Kooyman and Farenhorst Rec. trav. clzim. 1951 (in press). 15 Kooyman this Discussion GENERAL DISCUSSION 225 by Kharas3i and Friedlander and by Kharasch and Sage l6 for the ratio of the addition rates of CC1 radicals to the double bond in cyclohexene and n-octene respectively this suggests the abstraction of a-hydrogen atoms in cyclohexene by trichloromethyl radicals to proceed at a rate which is not much smaller than the rate of addition.Finally attention is called to the experiments of Kharasch and Fried-lander with respect to the overall kinetics of the photochemical addition of bromotrichloromethane to various olefins including cyclohexene at 10-50OC. The authors found the reaction to be zero order with respect to olefin. Olefin determinations were made by direct titration with bromide-bromate solution rather than by dilatometry. On the bases of these arguments it is suggested that the rate constants reported by Melville and his co-workers may contain contributions from cyclohexenyl radicals rather than to apply to the trichloromethyl radicals only. Thus termination might consist both of dimerization of trichloro-methyl radicals and of dimerization of cyclohexenyl radicals.The in-fluence of the latter will be of course more pronounced a t lower halide/ olefin ratios. Prof. H. W. Melville Dr. J. C. Robb and Mr. Tutton (Birminghum) (communicated) The kinetic evidence presented in our paper entirely pre-cludes atatck by trichloro-methyl radicals on the a-methylenic hydrogen atoms to form chloroform and a stable cyclohexenyl radical as suggested by Dr. Kooyman since under all our conditions the rate of reaction is very accurately proportional to (rate of initiation) 4 and also to the concentration of olefin. If such a reaction did occur exclusively as a termination CC1 + CgHlo +CHCl + CBHg . . . . k', the rate of reaction would be given by which is of course not the case.If on the other hand both processes, i.e. reactions given by k' and R were operative then the intensity ex-ponent would lie between 0.5 and I. The fact that i t is so accurately 0.5 supports the original kinetic scheme suggested in our paper. Since the discussion an experiment has been done in which cyclo-hexene in presence of benzoyl peroxide as catalyst with excess of carbon tetrachloride was placed in a sealed dilatometer tube. By irradiating with light of suitable wavelength the photo-catalyzed reaction was studied and in agreement with the work reported by Dr. Kooyman this was shown to be dependent directly on the light intensity thus establishing that under his conditions termination is indeed first order with respect to radical concentration.The rate of reaction under these conditions is very much less than when CC1,Br is used. This then raises the point of the reason for the different mechanism under the two different experimental conditions but further experiments are necessary in order to establish the nature of the differences in the mechanism involved and more precise information is required regarding the velocity constants for all possible steps in the reaction. Prof. H. W. Melville Dr. J. C. Robb and Mr. Tutton (Birmingham) said Since the paper on the reaction of trichloromethyl radicals with cyclohexene was submitted further values have been obtained for the proposed kinetic steis. These are listed below. R (30" C) k (40" C) 1-5 x I O ~ J , k (5oOC) 1-90 x I O ~ I Eovers1l = E - &Es = 4-5 kcal./mole.2kS 2k6 2-5 x IO* I f 1.3 x 1 0 2 1. mole-1 sec.-l 2.0 x 10' 1. mole-1 sec.-l. 16 Kharasch and Sage J . Org. Chern. 1949 14 537. 226 GENERAL DISCUSSION This gives a value of 4 = 5.5 where 2k6 = +V2k4 . 2k5. This low value is in accord with those recorded elsewhere in the dis-cussion by Bateman Gee Morris and Watson. Further attempts to obtain a value for the energy of activation of the termination step by means of experiments conducted using the technique of intermittent illumination at various temperatures has resulted in failure to detect any activation energy for this step. It should be pointed out, however that this technique is not sufficiently sensitive or accurate to measure small activation energies less than 3 or 4 kcal.over the rather small temperature range normally available in experiments of this kind. Dr. M. Magat (Paris) said I would like to mention some additional evidence concerning two points raised by Dr. Kooyman in his paper. First concerning so-called " stable " i.e. non-dimerizable radicals, some caution is advisable. Dr. Chapiro in our laboratory has for instance, observed that the reputedly stable radical I I-diphenyl-2-picrylhydrazyl reacts with double bonds of polymerizable vinyl compounds forming molecules of a molecular weight higher than that of the radical dimer. We are now investigating the kinetics of this process and the nature of the molecules formed. Secondly concerning the reaction of radicals with poly cyclic com-pounds we have investigated the effect of addition of cancerogenic hydro-carbons and their homologues on the rate of thermal polymerization of styrene at 37" C.If the usually present traces of peroxides are destroyed by a preheating under vacuum at this temperature for 2-3 weeks all these cancerogenic compounds slow down the polymerization. It can be shown by persistent fluorescence of the polymer that these compounds do enter the chain. It is remarkable that the absorption and fluorescent spectra are not significantly shifted. The slowing-down efficiency decreases in the order 3 4-benzpyrene, eo-methylcholantrene phenantrene a-methylanthracene I 2 5 6-di-benzanthracene. Chrysene and pyrene have no effect at a1l.l' Dr. K. S . Pitzer (Washington D.C.) said Prof.Kistiakowsky has reported in his paper some exceptionally fine experimental work and the minor point of interpretation which I am about to raise does not in any way detract from the principal results of the work. In the text just after Table IV i t is stated that the presently unknown corrections for anharmonicity in the vibrations would raise the calculated heat capacity of ethane more than that of ethylene. I would urge caution in accepting this estimate as certain although i t may well be correct. While i t is true that ethane has more vibrations in a given frequency range than ethylene i t could be that the effect of anharmonicity in the torsional vibrations overshadows all others at moderate temperatures. The present treatment as a restricted internal rotation accounts for the anharmonicity in the torsional motion in ethane while the more highly restricted torsion of ethylene is presently treated as a harmonic oscillator.Thus i t seems possible that the anharmonicity correction for the torsional motion in ethylene might be large enough to make the total for ethylene exceed that for ethane. The same statements can also be applied to propane and propylene. Prof. E. A. Guggenheim (Reading) (comr~zunicated) It j s clear that the measurements of Kistiakowsky and Nickle on the ethane-ethylene equilibrium are appreciably more accurate and more reliable than any previous measurements. It is particularly satisfactory to notice how well these results agree with the most recent theoretical values. It is perhaps, not entirely without interest to compare the experimental values with theoretical values calculated before the experiments.The following l7 Bodme and Magat Comfit. rend. 1951,232 1657 GENERAL DISCUSSION 227 Table gives a comparison between experimental measurements made in 1942 and here reported and the values calculated according to a formula la published in 1941 based on Kistiakowskfs own experimental value for the heat of hydrogenation a t 82’ C. EQUILIBRIUM VALUES OF @CIQB[2/p,-,Hb IN ATM. 380’ C 4-04 f 0.17 x 10-6 Temperature . 450’ c Calculated 1941 . . 5-2 x 10-4 4’3 x 10-6 Measured 1942 . . 5-13 f 0.13 x 10-4 Dr. L. Bateman (Welwyn Garden City) (commu.nicated) Horrex and Miles’s reference to the bond shortening in dibenzyl requires revision.Cruickshankl9 has re-analyzed Jeffrey’s data and finds that they are actually indicative of only half the contraction quoted. Moreover, recent isomerization equilibrium measurements by Dr. J. I. Cunneen and myself 20 fail completely to reveal any chemical effect (AG differences < 0.1 kcal.) which could be attributed to bond shortening in I 5-dienes. I suggest therefore that the agreement claimed between the experi-mental and ‘ I calculated ” bond dissociation energies needs critical reconsideration. Three uncertainties are apparent ((i) the correctness of the observed activation energy absolutely and as regards its identity with the energy of primary bond scission ; (ii) the appropriateness of using CH,-CH, rather than say CH . CH,-CH . CH as the “ non-resonance ” refer-ence compound for P h .CH,-CH,. P h ; (iii) the magnitude of the resonance energy of the benzyl radical. Horrex and Miles present con-vincing evidence for the approximate validity of their conclusions con-cerning (i) and this receives support from Bolland and Orr’s 21 investigation into the analogous breakdown of aliphatic I 5-dienes. I question, however whether f I kcal. is a fair estimate of the overall uncertainv in the activation energy in view of the complexity of the decomposition process the analytical intricacies and the spread of the points in Fig. 7. Any error associated with (ii) is undoubtedly small but will be such as to tend to reduce the “ standard ” CC-bond energy (by about 1-2 kcal.). The third factor is thus left as the main source of uncertainty.has derived the value of 24.5 kcal. by subtracting the activation energy for the pyrolytic decomposition of toluene (this being identified with the energy required to break an acyclic CH-bond) from the CH-bond dissoci-ation energy in methane. In fact there is sufficient uncertainty in this derivation for the formerly accepted value of 19 kcal. still to be tenable. Szw-arc reports the activation energy as 77-5 f 1.3 kcal. but statistical regression analysis of all his tabulated data (for uniform surface condi-tions) leads to the result E = 80.0 f 4-2 kcal. (95 yo limit). Further, apart from the small uncertainty in D, in methane (101 f I kcal.) i t would seem more legitimate to compare with an alkane €We where D, in the Me group is undoubtedly lower,2s rather than with the sym-metrical methane molecule.For the present therefore i t would seem preferable to calculate the dissociation energy of dibenzyl simply as (84 - 2 x 19) = 46 kca1.-in satisfactory agreement with the thermochemical and kinetic data-rather than as (85 + 11 - z x 24.5) = 47 kcal. in which a large other-wise unrecognized bond energy term has to be invoked in order to com-pensate for a higher resonance energy which is not definitely specified experimentally. Szwarc l 8 Trans. Favaday Soc. 1940 37 272 l9 Cruickshank Acta Cvyst. 1949 2 65. a1 Bolland and On I.R.I. Tram. 1943 21 133. 22 Szwarc J . Chem. Physics 1948 16 138. J Chem. Physics 1944 IS 479. To be published shortly. Stevenson J . Chem. Physics 1942 10 291 ; Anderson and Van Artsdalen 228 GENERAL DISCUSSION Dr.M. Szwarc (Munchester) (communicated) Dr Bateman's remarks enables me to clear up some details of calculation of activation energy in the pyrolysis of toluene.24 The technique elaborated in this investiga-tion makes i t possible to measure the rate of pyrolysis down to about 0.01 yo. This seems to be the lower limit permissible in this technique, and indeed runs 89 88 87 and 94 (crosses in Fig. 4) show that losses of products start to be noticeable when the limiting case of 0.01 yo is reached. Therefore these four results were ignored in calculating activation energy and were omitted in the graph which was presented in the paper but they were included in the Table giving all the results. The value of 80 kcal./mole obtained by Dr.Bateman results from including these four runs; their omission leads to an activation energy of 77.5 kcal./mole as quoted in the original paper. I take this oppor-tunity to include in Fig. 4 the results of pyrolysis of toluene obtained in 1948 by Dr. J. S . Roberts and in 1949 by Mr. J. Murawski. Pyrolysis of toluene. A Roberts (1948) E = 77'5 kcal./mole. x runs No. 89 88 87 94. Murawski (1949) 0 szxTm2 (1947) FIG. 4. In his contribution Dr. Bateman deals with the possible connection existing between the length and the dissociation energy of the central C-C bond in the molecule of dibenzyl. This problem has been discussed previously,26 and here I would like to clarify further certain points which need additional emphasis.(i) The dissociation energy of a bond linking two atoms depends not only on the nature of the two atoms but i t is also greatly influenced by the molecular environment of the bond in question (e.g. Table I in ref. (2)). (ii) The factors which influence the magnitudes of bond dissociation energies can be divided into two groups those connected with the struc-ture of the undissociated molecule and those due to the nature of radicals produced on dissociation. On the whole i t is not possible to ascertain which factor and to what extent is responsible for the observed changes in bond dissociation energies (see however ref. (3) in which an exceptional case is discussed). (iii) Further treatment of the problem of variations in bond dissoci-ation energies requires therefore the introduction of certain simplifying t4Szwarc J.Chem. Physics 1948 16 128. 25 Szwarc Faraday SOC. Discussions 1947 2 39 GENERAL DISCUSSION 229 assumptions. It has been assumed tentatively 26 that the variations in the C-H bond dissociation energies in molecules of the type R . H are due entirely to the factors arising from the nature of the radical R (i.e. i t has been assumed that these variations reflect the changes in the stabilities of various radicals R). Taking the value of D(CH,-H) as the point of reference it is possible zo to build up a system of experimental resonance energies " for various radicals R defined as D(CH,-H) -D(R-H). (iv) It follows from the examination of heats of formation of the relevant compounds that variations in other bond dissociation energies cannot be accounted for by the experimental resonance energies only.For example, D(CH,-CH,) - D(R-R,) -+ [D(CH,-H) - D(R-H)] + [D(CH,-H) - D(R,-H)]. I would like to emphasize that this inequality arises from thermochemical data only i.e. the value of [D(CH,-H) - D(R-H)] + [D(CH,-H) - D(R,-H)] - [D(CH,-CH,) - D(R-R,)] is independent of the values of dissociation energies used in this expression. (v) I t has been assumed that the above-mentioned value the " strengthening effect ",26 is related to the length of the relevant R-R, bond. For example its value for dibenzyl has been calculated as about 11 kcal./mole and i t has been suggested 2 5 ~ 26 that this value is related to the shortening of the central C-C bond in the molecule of dibenzyl.Since, as has been said above the value of II kcal./mole is independent of the value of the C-H bond dissociation energy in toluene the uncertainty (iij) quoted in the communication by Dr. Bateman is irrelevant €or the problem of the C-C bond shortening in the molecule of dibenzyl. (vi) The relation between the C-H bond dissociation energy in toluene and the C-C bond dissociation energy in dibenzyl is given by the heats of formation of toluene dibenzyl and the H atom i.e. ZD(CeH5. CH2-H) - D(C6H5. CH,-CH,. C6H5) = ZAH,(H) -ZAHf(C6H5 . CH,) +AHI(C6HQ . CH . CH . C6H5), = 106 kcal. /mole. Hence the value of D(C6H5. CH,-H) = 77-5 kcal./mole,28 requires D(C6H5. CH,-CH . C6H,) to be 49 kcal./mole while the value of the latter C-C bond dissociation energy is experimentally estimated by Horrex and Miles 29 as 48 kcal./mole.If D(C,H5 . CH,-H) = 80 kcal./mole as Dr. Bateman suggests then D(C6H5 CH,-CH . C6H5) must be 54 kcal./mole i.e. the values D(C,H . CH,-H) = 80 kcal./mole and D(C6H5 . CH,-CH . C6H5) = 46 kcal./mole suggested by Dr. Bateman are incompatible. Dr. B. G . Gowenlock (Swansea) (communicated) Horrex and Miles postulate the reaction (e), I PhCHCH,Ph -f PhCH=CHPh + H-(128 - Q) kcal., among the reactions resultant upon the production of PhCHCH,Ph radicals by the attack of benzyl radicals upon dibenzyl. This reaction is endothermic to the extent of about 58 kcal. being taken as 70 kcal., and therefore an activation energy of at least 58 kcal. will be required for this reaction. This activation energy is greater than that observed for the primary reaction.From the data given in Table I and assuming a I 26 Szwarc J . Chem. Physics 1950 18 1660. 27 Szwarc and Taylor Trans. Furaduy SOC. (in press). 28 Szwarc J . Chem. Physics 1948 16 128. 29 Horrex and Miles this Discussion 230 GENERAL DISCUSSION normal value for the temperature independent factor ( 1 0 ~ ~ sec.-l) i t can be shown that for the lower temperature data (630-700’ C) decomposition of the PhCHCH,Ph radical will take place to the extent of only 10-50 yo. Similar coksiderations apply to reaction (i). Reaction ( f ) , 2PhCHCH,Ph -+ PhCH=CHPh + PhCH,CH,Ph + (29- 128) kcal., I on the other hand is exothermic and should have a much lower activation energy. In contrast to reaction (e) i t will produce only one molecule of stilbene per two PhCHCH,Ph radicals.Therefore on the basis of the authors’ reaction mkchanism a variation of the toluene/stilbene and toluene/styrene ratios should be obtained with variation in temperature. Investigation of the pyrolysis of mixtures of benzyl iodide and dibenzyl a t about 5 0 0 O C should therefore lead to the virtual elimination of re-actions (e) and (i) together with their dependent reactions (g) ( h ) and ( j ) , and thus lead to toluene and stilbene as the sole reaction products in the ratio of 2/1. Dr. C. Horrex and Dr. S . E. Miles (St. Andrew) (communicated) : In reply to Dr. Gowenlock we wish to point out that our analyses showed no significant variation in the composition of the products with the tem-perature of reaction.When introducing our paper we drew attention to the endothermicities of reactions (e) and (i) and pointed out that these decompositions would have to have normal temperature-independent factors of about 1 0 1 ~ in order to be significant in our conditions. We consider that the uncertainty in the value of Q makes i t unprofitable to pursue such calculations at present since an error of 4 kcal. alters the estimated rate of (e) or (i) at 630’ C by a factor of 10. Since the reactions appear to be feasible and a homogeneous chain sequence improbable we prefer to await the result of further work which one of us is carrying out. It must be noted that although the primary decomposition is not a heterogeneous reaction the experimental conditions favour access to the wall and we cannot at present assess the part i t may play in radical decompositions.The extent of a homogeneous disproportionation re-action which requires the encounter of two radicals is very dependent on the concentrations of the latter and hence on the rate constants of (d) (e) and ( 2 ) . It may be significant to note that the early work which reported stilbene and toluene as main products was done with much higher concentrations than we employed ; this would result in decreased accessibility of the wall to the radicals and reactions (e) or (i) which might occur there would be less prominent. Investigations in progress include work on the reaction of benzyl radicals with other molecules. With reference to the points made by Dr.Bateman we have stated that we evaluated the best straight line by use of the high partial pressure data since in our opinion the analytical precision in the other experiments was impaired by the small amounts of products available to us. We consider this procedure legitimate and a closer examination of our Fig. 7 shows that the spread of points in the data used is small (with the noted exception of one point which by any test must be regarded as a faulty experiment). The point concerning the revision of the length of the central C-C bond in dibenzyl has also been drawn to our attention by Prof. Cox. The amount of the shortening and its relationship to the strength of that link are of importance and in summarizing Szwarc’s arguments we may have given the impression that the strengthening is dependent OR the value for this shortening.Actually any “predicted ” value for the dissociation energy of the central bond in dibenzyl depends on the evalu-ation of the heat of formation of the benzyl radical. This has been done in several independent ways a summary of which has been given recently.30 30 Quart. Rev. 1951 5 42 GENERAL DISCUSSION 23 1 The agreement shown by these methods appears to us to be convincing evidence for the higher value for the resonance energy of the benzyl radical. Dr. W. A. Waters (Oxfovd) said In connection with the paper by Horrex and Miles on the pyrolysis of dibenzyl I should like to draw attention to the rather different conclusions of Dr. A. F. Bickel and myself 31 concerning the free benzyl radical which we prepared by de-composing w w'-azotoluene in boiling decalin solution at about 200' C.We found that even at this low temperature the benzyl radicals did not just recombine to form dibenzyl but underwent about 35 yo dispropor-tionation t o toluene and stilbene. We did not detect either benzene or styrene which may be secondary pyrolysis products of stilbene. Though our results could perhaps be attributed to a very rapid de-hydrogenation of dibenzyl by benzyl radicals i t was significant that the decalin solvent was not dehydrogenated at all. The alternative ex-planation would therefore require both an abnormally high lability of the C-H groups of dibenzyl and a very low probability of recombina-tion of benzyl radicals. Unlike Szwarc we see no fundamental objection to postulating the simple reaction .2Ph. CH . + Ph . CH + Ph . CH. It is interesting also to note that the activation energy for the vapour phase dissociation of dibenzyl accords with the experimental data given by Ziegler 3 2 for liquid phase dissociations of many similar compounds, some of which give radicals that undergo disproportionation at temper-atures as low as 1 5 0 O C. Dr. M. Szwarc (Manchester) (communicated) Dr. Waters suggests the possibility of disproportionation of benzyl radicals i.e. The activation energy of this reaction should certainly be less than 5 kcal./mole if reaction ( I ) is to compete successfully with dimerization. This follows from the approximate equality of A factors for both dis-proportionation and dimerization (since the activated complexes for both reactions are very alike and in the liquid phase the deactivation of " hot " djbenzyl molecules cannot be the rate determining step).The activation energy of a process must be a t least equal to its endothermicity i.e. D(C6H5 . CH-H)-D(C,H . CH2-H) < 5 kcal./mole, D(C6H . CH-H) < 82.5 kcal./mole. On the other hand D(C6H5 . CH,-H) +D(C,H5 . CH-H) +D(C,H,-CH) +D(C-H) Taking D(C6H5. CH,-H) = 77-5 kcal./mole D(C,H5. CH-H) < 82.5 kcal/mole D(C-H) = 80 kcal. ; AH,(H) = 52 kcal./mole ; AH,(C) = L ( L being the heat of sublimation of carbon) and finally AH,(C6H5) ~3.68 kcal. /mole (the latter value being derived from a plausible assumption D(C6H5-H) m IOO kcal./mole) one derives : D(CsH,-CH) > L - 28 kcal./mole.The heat of sublimation of carbon is probably not less than 136 kcal./mole and therefore, D(C6H,-CH) > 108 kcal. /mole. The latter value appears to be much too high. It might be reasonable if the activation energy of (I) is about 25 kcal./mole but then this re-action could not be observed. = 3AHf(H) + f AHf(C6H6) -AHf(C6H5. CH3). 3l Rec. truu. cbim. 1950 69 316. 32 Annalen 1942 551 161 ; 1950 567 134 232 GENERAL DISCUSSION It seems to me that the results of Bickel and Waters could easily be interpreted on similar lines to those of Horrex and Miles. The reason why benzyl radicals do not dehydrogenate decalin seems to be simple : the C-H bond dissociation energy in decalin is higher than and thus the activation energy for dehydrogenation is too high.Dr. C . Eorrex (St. Andrews) (communicated) In reply to Dr. Waters I think the products obtained from the interaction of benzyl radicals depend on the experimental conditions. The conditions in solution and in dilute gaseous systems are distinctly different and even in our work with gaseous systems we have observed differences in the proportions of the products when the only substantial change has been the pressure of the inert carrier gas. Dr. Ruth Lapage when using toluene as an acceptor for methyl radicals at 500 to 600" C in a flow technique with 3 to 8 mm. nitrogen found the main solid product was dibenzyl with a small amount of stilbene. With 600 mm. nitrogen however the product was stilbene. Mr. C. B. Cowan has obtained similar data with phenyl radicals and toluene at 600" to 700" C.In the toluene pyrolysis Mr. J. 0. McCrae and Mr. R. B. Cundall examined the solid products from an experiment a t 850°C when using the pyrolysis technique as published by Dr. Szwarc. By use of u.-v. absorption spectroscopy' they' found the solids contained dibenzyl stilbene and anthracene in the ratio 380 7 I. Thus Dr Szwarc is substantially correct in speaking of dimerization only under his conditions. It was also found that using 1-5 mm. toluene in 5 mm. nitrogen carrier gas gave even smaller amounts of stilbene and anthracene at 850" C. In toluene pyrolysis the benzyl radicals are formed at temperatures where dibenzyl would be rapidly decomposed and dimerization probably occurs at the exit of the furnace.Since the gases are cooled quickly there will be little opportunity' for attack by the radicals on the first fractions of the dibenzyl formed. In the work of Lapage and Cowan the temperatures required should permit dibenzyl to be formed within the furnace and remain unpFolyzed. If stilbene was formed from di-benzyl by radical attack it seems that higher inert gas pressures materially assist in the formation of the dibenzyl. If benzyl radicals disproportionate I would have expected the reaction to be rapid at the temperatures of toluene pyrolysis and more stilbene produced than has been found. Dr. Miles used w a'-azotoluene as well as benzyl iodide to provide benzyl radicals when examining the attack of these radicals on dibenzyl. The products in both cases were the same and similar to those obtained from dibenzyl alone on pyrolysis.I would expect the abstraction of a hydrogen atom from decalin by a benzyl radical to be a distinctly endothermic process and the data given by Dr. Kooyman in his Tables I and I1 support the idea that hydrogen abstraction from dibenzyl is an easier process. With reference to Ziegler's data I would like to add that his tem-perature independent factors are generally high. Prof. S . W. Benson (California) said The data of Horrex and Miles seem to indicate a steric factor which is unusually low for a unimolecular decomposition involving the rupture of a single bond and an activation energy lower than that which would be expected for a C-C bond if the proper length of the bond were employed.In addition the dependence of the rate on the surface/volume ratio shows a strong temperature effect. Thus the ratio of rates in a packed to unpacked flask fall regularly from a factor of 3 at 635O C to about 1-4 at 720° C . The authors make the assumption that the ratio of heterogeneous to homogeneous reaction is proportional to the surface/volume ratio. This is by no means necessarily the case and must certainly depend on the rate and energy of adsorption of reactant on the surface as well as on the overall D(CQH6. CHZ-H) GENERAL DISCUSSION 233 mechanism. In particular the dependence may vary from complete independence to direct proportionality according to the conditions. Further the pressure range investigated (about 0-1 mm. Hg reactant) is that in which wall collisions proceed a t a rate comparable to intermolec-ular collisions.In the light of these uncertainties i t would seem premature to accept with any certainty the value of 48 kcal./mole and the low frequency factor for the unimolecular homogeneous decomposition of dibenzyl. Dr. C. Horrex and Dr. S. E. Miles (St. Andrews) (communicated) : Prof. Benson is in error in stating that in our conditions of 0.1 mm. Hg reactant pressure the wall collisions are about as frequent as intermolec-ular ccl'isions. He appears to be ignoring the presence of the carrier gas which impedes diffusion to the wall considerably. Calculation of the average number of collisions made by a molecule diffusing to the wall 33 gives about 105 for the pressures and reaction vessel used.By considering his comments on surface effects in conjunction with details given in his previous contribution on this topic we note that the effects of changes in surface/volume ratios are difficult to interpret where chains are initiated and end at the walls. In our case we see no reasonable evidence for a chain mechanism and consider the relatively small influence of the large change in surface implies little heterogeneity in the primary reaction. This is not in conflict with the possibility (mentioned in reply to Dr. Gowenlock) that some radicals produced in secondary reactions may reach the walls and undergo reaction there. Prof. Benson's assumption that the expected bond dissociation energy depends on the length assigned to the central C-C bond has been dealt with above.We agree of course, that the temperature independent factor is unusually low in our experi-mental conditions. It seems to us that it is desirable to have present an acceptor other than dibenzyl for the removal of the benzyl radicals ; data on the thermal stability of potentially suitab!e substances is being completed. In view of findings on the pyrolysis of methyl iodide in this pressure range given later we are also checking to see if the high pressure limiting rate has not been attained. Dr. C. Horrex and Dr. Ruth Lapage (St. Andrews) (communicated) : Some points disclosed by a detailed examination of the pyrolysis of methyl iodide seem to be of interest in connection with the kinetic difficulties in this pyrolysis work a t low pressures and particularly since C-H and other bond dissociation energies have been derived from published work on iodides.Butler and Polanyi 34 report five experiments on the pyrolysis of CH31 at 493'-495" C and by calculating first order constants and using log k (sec.-l) = 13 - E/4*57T they derived E = 54 kcal. With this value assigned to the C-I bond dissociation energy they deduced a C-H value €or CH which proved in agreement with later determinations by other methods. We have used the same techniques and conditions and found that their quoted values of k can only be realized a t temper-atures about rooo C higher. The use of the correct temperature in the above equation would give a value of E of about 60 kcal. and this seems too high for the C-I dissociation energy.The validity' of assuming first-order behaviour and a 1013 factor under such experimental con-ditions was therefore examined. With 0.07 to 0.3 mm. CH,I in 3 mm. nitrogen carrier gas the kinetics approached second order and changed towards first order above I mm. of CH31. In these conditions the re-action was mainly' homogeneous and not inhibited by iodine the free methyl radicals could be trapped by toluene but in its absence they were decomposed quantitatively' a t the wall to methane and carbon. The rate of reaction was increased by increase in pressure of the carrier gas. s3 Bursian and Sorokin 2. $lay&. Chew. B 1939 12 247. ,* Butler and Polanyi Trans. Faraday SOC. 1943 39 19 234 GENERAL DISCUSSION By using 600 mm. Nz z mm. toluene and 1-2 mm.CH,I we obtained rate constants showing a temperature variation given by The strong influence of experimental conditions on the rates of decomposi-tion can be seen from the following values of k calculated on a first order basis. Conditions Goo mm. N g mm. N 3 mm. N 3 mm. N, k (sec.-l) = 2.6 x 1olS exp -547oo/RT. z mm. C,H 1-1.3 mm. 1-1-3 mm. 0.3 mm. 1-2 mm. CH,I CH31 CH,I CH,I K (sec.-l) at 0.028 0.0066 0.0056 0.0016 Soo0 K The presence of a radical acceptor was essential in experiments at high total pressures since with the decreased accessibility of the wall to the radicals a recombination reaction became important. An analvsis of our data has beeimade which considers activation of the CHJ molecules as occur-ring by collisions with CH31 or N mole-cules the latter having about one-fifth the efficiency of the former.Applying the con-ventional test for data on second- to first-A order transitions of plotting the reciprocal of a first-order constant against the recip-rocal of the concentrations (appropriately weighted) we obtain Fig. 5 The experi-mental point on the vertical axis clearly represents the high pressure limiting rate for the reaction. Details of this work are being prepared for publication. These findings while not altering the C-I bond dissociation energy make the initial derivation of the 54 kcal. value seem &A rather fortuitous. Similar troubles may exist with other substances in the Butler-A Polanyi conditions although one might expect the fall off in first-order constants for more comdicated molecules to occur a t A A na 53 rp A A A M~ I + 0.2 MM.1,) - lower pressures than for CHJ. A number /;o 2.0 of C-H bond energy values which rest on the original iodide pyrolysis determinations may' be uncertain for these reasons. Dr. Horrex and J. 0. McCrae (St. Andrews) said Dr. Kooyman may be interested in our preliminary results on the pyrolysis of diphenyl-methane which show that the products are essentially tetraphenylethane, fluorene and hydrogen. The technique of investigation has been similar to that used for dibenzyl (see this Discussion). We assume the primary process is the breaking of the methylenic C-H link and that fluorene production may proceed through the planar diphenylmethyl radical radical. The calculation of the rate of this primary process via the hydrogen production (allowing for fluorene formation) yields a rate constant k = 10'~ exp (- 73ooo/RT).We are extending the range of experimental variables but suggest that this energy of activation can be considered as the C-H bond dissociation energy Prof. 6. W . Benson (California) said There seems to be some question in the work of Rowley and Steiner of the role played by each of the three possible simultaneous reactions cis plus cis ; cis plus trans and trans plus trans (butadiene). hobs = a2k + z a ( ~ - a)kd + (I - a)'LKtl, where a is the fraction of the cis form and K, K,t and Kg are the specific rate constants for the individual reactions of the species indicated. If i t FIG. 5. W e can write a general rate expression GENERAL DISCUSSION 23 5 is assumed that the active complex is the same for all of these cases then it can be easily shown that the observed activation energy is not dependent on the extent to which the different species react since a = e -AE/RT ; E, = Ell - zAE = E,# - AE and the temperature dependence of the 01 exactly cancels the different temperature dependence o f the different rate constants.In this case the above expression reduces to If however there is more than one possible complex for the system and the steric factors are different for the different isomers then we should expect a different contribution at different temperatures from the different species. Thus i t may be quite possible that a cyclic and linear complex both exist having different activation energies and different frequency factors. In this case the system becomes much more complicated and alternative explanations must be considered for interpreting the data. Dr. M. Magat (Paris) said There is one point that strikes me as not being taken enough into account in all theoretical discussion of butadiene cyclization reaction. It is the fact pointed out by Aston Szasz Wooley and Brickwedde 35 that the trans configuration is 2.5 kcal./mole more stable than the cis configuration. On the other hand at least one of the two molecules must have the cis configuration in order to make the cyclic complex possible while the linear complex can be realized with any one of the two. Mr. B. Eisler and Dr. A. Wasserman (London) (communicated) : Experiments have been carried out which enable a comparison to be made of the activation energies E of Diels-Alder diene associations of the following type : kobs = ktt = a'ku. Butadiene + dienophile -+ adduct . * (1) Cyclopentadiene + dienophile + adduct . * (11) The difference between E and EII can be accounted for,ss if it is assumed that the formation of the transition state of (I) involves the conversion of trans-butadiene into the cis form. In the reaction discussed by Steiner the dienophile of (I) is butadiene. A comparison of the E values of (Ia) and (IIa) indicates that both butadiene molecules are converted into the cis form before the transition state of the reaction (Ia) is taken up. 2 Cyclopentadiene -+ dicyclopentadiene . . (IIa) This is due firstly to steric requirements and secondly to effects which operate generally in Diels-Alder reactions and which bring about the formation of relatively closely packed transition Dr. H. Steiner (Petrocarbon Ltd. Manchester) (communicated) I n reply to the remarks by Benson Magat and Wasserman a statistical rate calculation of the butadiene dimerization using the values for the thermodynamic functions of butadiene obtained by Aston Szasz Woolley and Brickwedde,s* may give an answer to some of the questions raised. Since the interconversion of trans- into cis-butadiene is endothermic by only 2-5 kcal. /mole whereas the activation energy of the dimerization is some 25 kcal./mole the initial state which has to be assumed in such a calculation is most likely' an equilibrium mixture of trans- and cis-butadiene at the appropriate temperature. It may be that the deviations from the Arrhenius function which we have observed are connected with the gradual shift of the cis-trans-butadiene equilibrium particularly at low temperatures. Unfortunately our statistical calculations were carried out using approximate values for the partition function of buta-diene only. I hope to recalculate this prablem using the data of Aston and collaborators. 2 Butadiene -+ vinylcyclohexene . (14 85 J . Chem. Physics 1946 14 67. 36 Detailed considerations and experimental results to be reported elsewhere. 37 J . Chem. Soc. 1935. 833 1512a8 1936 432 ; Trans. Faraday SOC. 1936 35 841. J Chem. Physics 1946 14 67