G. B. KISTIAKOWSKY AND A. GORDON NICKLE 187 THE PYROLYSIS OF DIBENZYL BY C. HORREX* AND S. E. MILES? Received 5th February, 1951 The pyrolysis of dibenzyl has been studied a t low partial pressures and The products have been shown to be in accord using fractional decompositions. with the equation 3PhCH,CH,Ph -+ 2PhCH, + P h C H d H P h + PhH + P h C H d H , and the process has been discussed in terms of a primary dissociation into benzyl radicals with a first order rate constant k(sec.-l) = 1oSs3 exp (- 48.0 kcal.)/RT. If the energy of activation is taken to indicate the dissociation energy of the central carbon-carbon bond it agrees with anticipations based on a resonance energy of 24.5 kcal. for the benzyl radical ; the temperature independent factor is abnormally low. The possible secondary reactions have been outlined.The determination of the bond dissociation energy for the central C-C bond in dibenzyl is a problem of considerable interest in view of the recent work leading to a resonance energy of 24-5 kcal. €or the benzyl radical.1 It is known that this central bond is shorter than the normal carbon-carbon bond by 0.06 and Szwarc, in a recent discussion of these points, used Skinner’s data on the relationship between bond length and bond strength to deduce that the bond is strengthened by 13-7 kcal. * Present address : Chem. Dept., The University, St. Andrews. -f Present address : Thornton Res. Centre, Shell Refining and Marketing Co., Szwarc, Faraday SOC. Discussion, 1947, 2, 39 ; J . Chem, Physics, 1948, Jeffrey, Proc.Rqy. SOC. A , 1947, 188, 222. Skinner, Trans. Faraduy SOC., 1945, 41, 645. Ltd. 16, 138.I 88 PYROLYSIS OF DIBENZYL From the heats of formation of toluene and dibenzyl, together with the bond dissociation energy for C-H in toluene, he deduced that 47-1 kcal. is the dissociation energy of the central C-C bond of dibenzyl. This value is 11 kcal. greater than that deduced by subtracting z x 24-5 kcal. from the normal C-C bond strength of 85 kcal. in ethane, and Szwarc attributed this 11 kcal. to the bond shortening effect, thus obtaining reasonable agreement between the two methods of estimation. Before the publication of his work, an attempt at a direct determin- ation of the bond dissociation energy in dibenzyl had been started. This was based on earlier observation of Horrex and Szwarc on the pyrolysis of benzyl iodide, that the benzyl radical had considerable stability.Other relevant qualitative data have been summarized by Szwarc.1 The pyrolysis of dibenzyl has not been the subject of kinetic study under conditions where it might be Fossible to derive the mechanism of the decomposition. Barbier 4 distilled dibenzyl through a red hot tube and decomposed it into toluene and stilbene, with smaller amounts of phenanthrene. The molecular ratio toluene/stilbene was observed to be approximately 211 and this was confirmed by Graebe.6 The most suitable experimental conditions for the investigation of the primary step in the pyrolysis should be where only small fractional decompositions result (cp. Butler and Polanyi 6 for organic iodides). A conventional flow technique is required for such an investigation and the mechanism has to be derived from the analysis of stable end-products.Since the benzyl radical has a high degree of stability, it was considered worth while to see if the probable primary dissociation of dibenzyl into two benzyl radicals could be observed more directly. For this purpose, very low pressures and moderate temperatures would be favourable, and the use of a quartz fibre manometer in a heated closed system con- taining dibenzyl vapour was therefore considered. Results derived from the application of both techniques are described below. The manometric technique was applied first and showed that secondary reactions were proceeding ; this necessitated a change to the flow technique.Experimental 1. Quartz Fibre Manometer Experiments Haber and Kerschbaum' showed that the rate of damping of the vibra- tions of a quartz fibre in a gas a t a pressure such that the mean free path of the molecules was much greater than the fibre diameter, was given by the equation b/t = u +pM*, where t is the time 01 half damping, a and b are constants characteristic of the fibre and the temperature, M is the molecular weight of the gas and p its pressure. For use as a manometer, direct calibration of t against a McLeod gauge has to be made and a and b determined empirically. It seems that no use has been made of the instrument a t elevated temperatures. was constructed and calibrated for pressures of 0-2400 p air and a t temperatures from room t o 1000' K.The detailed results, which are t o be given and discussed elsewhere,O show that the instrument has distinct disadvantages for our purpose in that a and b vary markedly with temperature, the variation of a being particularly serious in regions which we expected to be of interest to us (Fig. I). At con- stant temperature the instrument obeyed the above equation w7ell up to pres- sures of 50 p of dry air a t which the mean free path is about 10 times the fibre diameter. In using the calibration with air when measuring pressures of other gases, the accommodation coefficient was assumed substantially identical for all particles involved in momentum transfer with the vibrating fibre. A bifilar type of the manometer as described by Coolidge Barbier, Compt.rend., 1873, 1770. Haber and Kerschbaum, 2. Elektrochem., 1914. 20, 296. Coolidge, I Amer. Chew. SOC.. 1923, 45, 1637. Horrex and Miles (to be published). 6 Graebe, Annalen, 167, 161. fi Butler and Polanyi, Trans. Faraday SOC., 1943, 39, 19.C. HORREX AND S. E. MILES 189 The tube containing the quartz fibre manometer projected from its surround- ing furnace and was joined to a sample container through a breakable seal. The samples were held a t constant temperature and the products of pyrolysis could be removed, if required, through a valve. As a check on the technique, vapour pressures of iodine and dibenzyl a t ordinary temperatures were deter- mined and the dissociation of iodine was followed using furnace temperatures from 523 t o 983' K and vapour pressures of 41 to 26 p.The observation of t and use of the relevant a and b values for that temperature permitted PM*, and hence M , to be evaluated. The progressive dissociation of the iodine a t increasing temperatures was clearly shown and the calculated equilibrium constants were in broad agreement with Bodenstein's values, but the heat of dissociation calculated from our data was some 30 yo low. Two sources of error probably contributed to this. The previously mentioned assumption of constant accom- modation coefficients cannot be satisfactory a t high percentages of dissociation ; also the derived value of the equilibrium constant was very sensitive to small ob- servational errors in t . It was concluded from these experiments with iodine that the dissociation process could be shown with sufficient accuracy t o warrant experiments with dibenzyl, but that any heat of dissociation would be subject to considerable uncertainties.Initial experiments with di- benzyl were made by raising the furnace temperature in steps and taking frequent readings at each temperature and using pressures of dibenzyl of about 5 p. These readings showed that up to 460 & I O O C the values of pM*, calculated with the aid of the FIG. I.-Va,riation of a for the quartz-fibre manometer. appropriate u and b values (Fig. I), remained steady, but above this temperature a marked increase in $M* was evident and this was more pronounced the higher the temperature and the longer the time of contact. The increase was irreversible and unaffected by subsequent lowering of the temperature below 460°C.A simple dissoci- ation process should have given a decrease as with iodine and it was concluded that new molecules were being produced by pyrolysis and each was contributing its part to the sum ZflM4. An attempt was made to discover whether the value of pM* was lower than normal immediately after dibenzyl reached the furnace by studying the variation of its value with contact time. The quantity (pM*),/(pM*), was plotted against the time of reaction, but extrapolation was made uncertain by the difficulty of fixing the zero of time when the dibenzyl had developed its full vapour pressure after sudden admission to the furnace. It was clear, however, that any radical dissociation products underwent quite rapid secondary reactions even a t the low pressures employed.Some evidence as t o the nature of the products produced was obtained by noting, in an experiment a t 725" C, that pM* fell from 306 to 26 when liquid air was applied to the sample tube. By measuring the pressure of this non- condensible gas by a McLeod gauge, and correcting for the expansion, the value of p in the manometer system was found to be r 1-7 p and hence M = 4-9. Thus a great part of the gas must have been hydrogen. To obtain further information on the nature and rate of the decomposition a flow technique was adopted. 11. Flow Technique Experiments Procedure.-The apparatus shown diagrammatically in Fig. z was con- structed of Pyrex glass with the exception of the furnace tubes which were madeI 90 PYROLYSIS OF DIBENZYL of silica.In essence, a stream of pure nitrogen picked up tne vapour of di- benzyl, transported it through a furnace, was freed from the products of reaction and unchanged dibenzyl, and returned via a device for measurement of its rate of flow to the circulating pump. Cylinder nitrogen was cooled to the temperature of liquid oxygen to free it from volatile matter and passed very slowly over sodium in two traps heated to 300'C in order to free i t from oxygen. The purified material was stored in sufficient quantity to provide a supply for all the experiments. The purified nitrogen a t about 2 to 3 mm. pressure was circulated by the mercury pump P and freed from mercury vapour by passage through a liquid air trap followed by an arrangement W permitting alternate heating and cooling of the gas in order to remove any slight mercury mist.The dibenzyl could be placed in U-tube 3, or in a metal boat in the preheated section Q when higher partial pressures were required. The partial pressures of dibenzyl employed were I O - ~ to 4 x 10-1 mm. and to prevent condensation in the inlet tubes the latter were arranged to be suitably heated. The reaction vessel (diam. 2-7 cm.) was in a steel tube furnace with tapped windings arranged to provide a length of 32 cm. with a temperature constant to 2 O C and with a sharp fall a t each 1 FIG. 2.-Diagram of apparatus. extremity. The use of a constant voltage source on the furnace windings per- mitted the mean temperature of a run to be maintained constant to f I' C by hand control. The time of contact (0.25 to I sec.) in the furnace and tem- perature were chosen to give decompositions ranging in the main from I to 15 yo.Limitations of the analytical techniques developed prevented smaller decom- positions being employed and runs lasted about Q to 2 hr. The gas issuing from the furnace passed through trap I maintained at a tem- perature of oo C to - 22' C, which was suitable for the removal of the less volatile products from the stream. All the remaining products save hydrogen and methane were removed in trap 2 cooled in liquid air. The hydrogen and methane passed through a furnace containing copper oxide, the temperature of this being adjustable so that the combustion of H, could be achieved whle pyrolysis pro- ceeded. The methane was determined separately by raising the temperature and recirculating later.The products of these combustions were removed in U-tube 4, and the flow rate of the carrier gas deduced from the pressure drop across a suitable capillary system D inserted in the return line to the circulation pump. Constant flow rates were achieved by operating this pump a t a suitable constant temperature by using a Woods metal bath with electrical heating con- trolled by a Sunvic energy regulator. Since the dibenzyl was not separated by taps from the furnace, checks were made on the loss sustained during the preliminary evacuation of the apparatus with the dibenzyl at room temperature and of the loss due to diffusion during a period of 70 min. in presence of 3 mm. N, but without circulation of the latter.The figures were 0-4 yo and 0-16 yo of the amount transferred through the furnace during the experiments. Thus the error introduced a t the beginning and end of an experiment was negligible. Analysis of Products .-For convenience these are divided into three groups : (i) gaseous, (ii) liquid and (iii) solid. (i) THE GASEOUS PRODUCTS include H,, CH, and any C, hydrocarbons. The latter would condense in trap 2 when their partial pressures exceeded about 0.05 mm., which in the known volume of the apparatus corresponds to 10-5C. HORREX AND S. E. MILES mole. By isolating trap 2 a t the end of an experiment, raising the mercury to mark C and replacing the liquid air by a bath a t - 78" C, the rise in pressure in the calibrated volume (after correction for temperature change), permitted an estimation of the condensed C, hydrocarbons.In expt. 39 where 5-55 x I O - ~ mole dibenzyl were passed through the furnace and 15 yo was decomposed to toluene, only 5-2 x I O - ~ mole C , hydrocarbons were found, i.e. less than 0-2 yo decomposition. The amount of products combusted by the copper oxide could be determined by isolating and warming U-tube 4 successively to - 78" C and room tem- perature and noting the pressure developed in the calibrated volume. In almost all cases the hydrogen production in the reaction was small and saturation pressures of water vapour were not reached. A check on the CH, production was made by determinations of the total pressure rise in the circulation system at the end of an experiment using carefully standardized conditions of com- parison with initial conditions.The results by both methods were in fair agreement, showing methane production to be negligible. (ii) THE LIQUID PRODUCTS collected in trap 2 and were transferred to the detachable limb R by vacuum distillation. By raising the mercury to A and placing baths a t a series of temperatures on R the vapour pressure curve could be obtained and served as an indication of substances which might possibly be present. By lowering the mercury to B or C the liquid could, in some cases, be completely vaporized and the pressure exerted in the fixed volume recorded. By recondensing the liquid in R the latter could be weighed and the total amount of liquid products found. The quantity available from a single run was about By consideration of the probable modes of fission of dibenzyl the liquid pro- ducts might contain benzene, toluene, ethyl benzene and styrene.The vapour pressure curve from - 38°C to 25°C lay about midway between the vapour pressure against temperature curves of benzene and tduene, and well above those of ethyl benzene and styrene. A further qualitative indication was that the melting point of the mixture seemed to be over the range - 120' to - 108O C . The melting point of benzene is 5" C, toluene - 102" C, ethyl benzene - 93" C and styrene - 33"C, so that the two pieces of evidence suggested that a con- siderable amount of toluene was present and probably some benzene, but did not enable any decision on the others to be made.By arranging a suitable additional attachment a t R the liquid products were vacuum distilled into previously degassed sulphuric acid, allowed to react a t room temperature for 5 min. and then transferred back to the original con- tainer. By measurement of the pressure of the completely vaporized liquid before and after such treatment i t was found that some I I to 25 % of the total had been absorbed by the acid. Blank tests on benzene, toluene and ethyl- benzene showed less than 2 to 5 "/p loss. The result suggested that styrene might be present and further qualitative indications were obtained from re- fractive index data. The refractive index of the liquid products was in the range 1-5050-1-5136 ; styrene has a value of 1-5379 and the benzene, toluene, ethyl benzene values are rather similar and lower, viz.1-500g-1.4g70 a t 17.2" C. The use of synthetic mixtures suggested that the styrene proportion was definitely 20 yo or more. By using bromine in glacial acetic acid under conditions ( I hr. at room temperature in the dark) such that styrene was estimated with 95 yo accuracy and reaction with the saturated hydrocarbons was negligible, it was found that the styrene present in the liquid products was 21 f 4 yo weightlweight. These analyses covered the whole range of decompositions studied and no significant trend in analysis with the percentage decomposition was evident. Morton and Mahoney's lo technique for fractional distillation of small amounts of liquid was next applied. The fractionating tubes used were 7.5 cm.x 1.5 mm. and packed with powdered glass wool, and the boiling point of the distillate samples was determined by Emich's method. The results, examples of which are given in Fig. 3 and 4, were of the same type whether the pyrolysis temperature had resulted in 15 yo or 0.5 yo decomposition, or if even lower temperatures had been used with a source of benzyl radicals present. For comparison purposes, blank runs on synthetic mixtures containing various proportions of benzene, toluene, styrene and ethylbenzene were carried out. The distillation curves of the pyrolysis products indicated that benzene (b.p. 80-81'), toluene (b.p. 108-IIO"), and styrene (b.p. 144-147' ) might be present, loMorton and Mahoney, Ind. Eng. Chem. (Anal.), 1941, 13, 494. 0'1 to 0'2 g.192 PYROLYSIS O F DIBENZYI, but the possibility of some ethylbenzene (b.p.134-136') being in the last 30 yo or SO of the distillate could not be eliminated. The results of blank experiments using 49.7 % benzene + 29.2 % ethylbenzene + 21.1 % styrene, and 50-5 % ben- zene + 49.5 yo ethylbenzene are superimposed in Fig. 4 and show that the central portion of the I' product data " is given by a genuine fractionation separating - /40 . B.Pt C 'C E - 0 f40 1 B.Pt - "C - /20 I FIG. 4. 0 Run 58 ; - 50'5 "/o C6H6, 49'5 O/o CSHI,. 49'7 "/o c6H6, 29'2 "/o C ~ H I ~ , 21.1 "/o CeH,. - - - - FIGS. 3 and +-Micro fractionation curves for liquid products and synthetic mixtures. toluene a t about 108~-112~ C. This is emphasized in Fig. 3 by the curve given by a blank of composition 20 yo benzene + 59 yo toluene + 21 % styrene, and it also appears from comparison with the latter that the materials in the pyrolysis products which were less volatile than toluene might amount to 25 yo.Due to the small amounts of materials involved (0*020-0.040 g.) and errors caused by fractionation during the determination of the boiling points of the samples,C. HORREX AND S. E. MILES 1 93 there is distinct scatter in the observations but comparisons with blanks sug- gested that the composition of the liquid products of pyrolysis was 20 f 5 Yo benzene, 25 f 5 yo styrene (with ethylbenzene), with residual toluene 55 & 10 yo uy weight. At a late stage in the work a Beckman Model D.U. photoelectric spectro- photometer became available and the u.-v.absorption curves of Fig. 5 for typical tiquid pyrolysis products were obtained. Also included on the graph are the curves given by a blank containing by weight 19 Yo benzene, 59 yo toluene and 21 yo styrene and another for 30 yo benzene, 40 yo toluene and 30 yo ethyl- benzene. The correspondence between the solution containing styrene and the unknowns is close ; due to the much higher extinction coefficient of styrene the characteristics of the first three graphs are essentially those of styrene and the small halt a t 2680 A is the main indication of the presence of other hydro- carbons. It is presumably due to the peak present in that region for both toluene and ethylbenzene. To be assured of the accuracy of these deductions, the calculated optical density of the known solution was compared with the observed a t suitable intervals, using calibration curves foi the pure component obtained on the same instrument.Good agreement resulted ; a t 2680 wheIe the contribution of toluene to the total is proportionally greatest, the result was D (obs.) = 0.14, while D (calc.) = 0-00084 (benzene) f 0.0522 (toluene) + 0.0889 (styrene) = 0.1419. The two unknown solutions show an absorption above 2940 A which is not found in the blank. It is known that the solid products, from which the liquid products had been separated by high-vacuum distillation, contained stilbene which absorbs strongly in this region. The presence of a trace of this would produce an error in evaluating styrene concentrations from the absorption maxima a t 2810 and 2910 A.As styrene does not absorb at 3100 A, a stilbene estimation (c = 20,ooo) giving 0~000108 g./L tor Iun 80, indicated that the 0.0348 g./L liquid products present in the solution contained 0-29 yo stilbene as impurity. At 2910 and 2810 di this would lead, with E (styrene) 446 and 740 and (stilbene) 21,400 and 19,050, to weight % styrene values of 26.6 and 25.3 for the liquid product. Without this correction for stilbene, and assuming the total absorption a t the two peaks was due to styrene, an average value of 34 weight Yo was obtained. Run 81 gave 20 yo with the correction and 25 yo without it. Since the evidence from refractive indices, bromination and u.-v. data favoured a styrene concentration of 20 to 25 weight yo in the liquid products, i t was concluded that the 25 f 5 yo fraction found by micro-fractionation was largely styrene and the ethylbenzene was not appreciable.(iii) THE SOLID PRODUCTS of pyrolysis condensed with a large excess of unchanged dibenzyl as a white solid ; there was no evidence of carbon in this nor in the reaction zone. In view of the earlier work of Graebe, and some evidence concerning the action of methyl radicals on dibenzyl, stilbene was considered as a possible product. Due to its low concentration in the solid fraction from normal runs, crystallization and sublimation brought about little separation. From runs where a high proportion of dibenzyl was decomposed (by means of benzyl radicals produced from benzyl iodide), stilbene was isolated and checked by mixed melting point.Bxomine in glacial acetic acid was used to estimate the unsaturated fraction in the solid products, blank tests having been done on dibenzyl and phen- anthrene under the same conditions. The u.-v. absorption curve for the solid products in ethanol (Fig. 6) combined characteristics of stilbene and dibenzyl. Phenanthrene, which absorbs in the same region as these substances, continues to absorb up to 3750 A and the absence of its characteristic peaks after 3400 A was taken to indicate its absence from the solid. At 3940 the absorption of the ethanol solution of the solid (0.318 g./1.) from Run 80 was ascribed to stilbene (E molar = z3,5co) and this gave 3-5 weight yo stilbene in the solid. Bromination of the same solid gave a value of 3-1 yo while for Run 81 the u.-v.absorption data gave 1-6 yo and bromin- ation 1-5 yo. The analyses based on bromination have been used for the results quoted below as the spectrophotometer was not available sufficiently early in the work. From the composition of the liquid products we derive a molecular ratio of 1-0 f 0*23/2-3 f 0*42/o-g2 f 0.2 for benzene/toluene/styrene. The toluene/ stilbene ratio varied more widely, from 1*51/1 to 311, and we ascribe this to difficulties in determining the small amounts of stilbene by bromination. Since GJ 91 PYROLYSIS OF DIBENZYL the gaseous products were negligible we conclude that the overall decomposj tion can be represented by 3PhCHzCHzPh -+ 2PhCH, + PhH + PhCH=CH, + PhCH=CHPh. In order to justify the assumption that toluene is formed by benzyl radicals attacking dibenzyl, runs were conducted with the flow technique as previously described but using lower temperatures and dibenzyl plus benzyl iodide as the reacting system.From previous work i t is known that even at 450' C the benzyl iodide is 60 yo decomposed in the time which produced about 2 yo de- composition of dibenzyl a t 700° C. Using equimolecular proportions of dibenzyl and benzyl iodide at 701' C (partial pressures 0.045 mm.), the liquid products increased 25-fold and seemed by distillation characteristics to be the same as noted for dibenzyl alone, toluene being a major constituent. The use of a carrier gas containing 53 yo nitric oxide in nitrogen instead of pure nitrogen produced no significant change in the rate of decomposition of dibenzyl a t 752' C and was not further investigated. I I t I I FIG.5.-U.-v. absorption curves for liquid products and synthetic mixtures in ethanol. FIG. 6.-U.-v. absorption curves for solid products in ethanol. Kinetic Data on the Reaction.-A test of the homogeneity of the reaction was made by packing the furnace with silica wool of approximate diameter 0-01 mm. and thereby increasing the surface exposed by 16 f 5 times. The products of the reaction seemed identical with those from the open tube and the liquid portion on micro-fractionation gave a curve indistinguishable from normal runs. The values of a calculated first order constant were about 60 % higher than the value derived without packing (Runs 55 and 77) ; this indicates that with the normal furnace the reaction was substantially homogeneous and rather less than 4 yo of the total rate might have been due to a surface reaction.The partial pressure of dibenzyl was varied by a factor of fifteen (o*ozz to 0.35 mm.) and the time of contact by nearly five times (0.17 to 0.80 sec.). With partial pressures of less than 0.07 mm. of dibenzyl reproducibility decreased as analysis errors became more important due to the small amounts of products available. The data were analyzed on a first order basis assuming that the toluene found was a measure of the initially formed benzyl radicals. The toluene was taken to be 55 yo of the weight of liquid products as a result of the previous work. As can be seen from Table I by comparing runs a t about the same temperature, the first order assumption seemed valid.To illustrate the effect of partial pressure variations, the following pairs of runs can be taken,C. HORREX AND S. E. MILES I95 No. 45, 61, No. 47, 74, No. 76, 68, while for variation of contact times and pressures jointly, No. 59, 58, No. 74, 80, No. 45, 44 are suitable. The data for log k against r/T0 K are plotted in Fig. 7 and the method of least mean squares has been applied to the values for experiments a t partial pressures greater than 0.07 mm. in order to obtain the best straight line. The data at lower pressures clearly are in general agreement with the other data but as indicated above the analytical uncertainties were known to be much greater. The results are repre- sented by the equation, (48.0 & 1.0 kcal.) x 103 2'303 RT log,, k = g 29 f 0'22 - FIG.7.-Variation of velocity constants with temperature. Line refers to high partial pressure data. High partial pressure- Low partial pressure- A Long contact time. v Long contact time. Short contact time. 0 Short contact time. Discussion Mechanism of the Decomposition.-Three possibilities can be con- (4 PhCH,CH,Ph -+ ZPhCH, (b) (4 The activation energy for reaction (a) js 84-4 + 13-7 - 2 x resonance energy of benzyl radical in kcal. as indicated earlier, and with the recent value of 24-5 for the last term, the dissocia.tion would require 49 kcal. For reaction (b) we consider that the activation energy will be rather less than 77-5 kcal. found for the dissociation of a hydrogen atom from toluene. In (b) we are concerned with a secondary hydrogen atom, and the radical produced should have a resonance energy of a t least the same magnitude as the benzyl radical ; moreover it appears that the benzyl radical can abstract hydrogen atoms from dibenzyl.The Ph-C bond in dibenzyl is 0.04 A shorter than the usual C-C distance which implies appreciable sidered for the primary decomposition : -+ PhCH,Ph + H(- Q kcal.) -+ Ph + - CH,CH,Ph.1 96 PYROLYSIS OF DIBENZYL strengthening and, as in toluene, the breaking of this bond should prove more difficult than the loss of a hydrogen atom. The most favourable process is therefore dissociation into two benzyl radicals and we have provided evidence €or believing that this is followed by (d) PhCH,* + PhCH,CH,Ph --t PhCH, + PhCHCH,Ph + (77-Q).The subsequent reactions produced stilbene, styrene and benzene in ap- proximately equimolecular proportions and while insufficient evidence exists to establish a mechanism firmly some features of the problem can be discussed. Stilbene could result from the radical decomposition (e) or disproportionation (f), but the latter offers no route to the production of styrene and benzene while the former does since it can be followed by the further reactions (g) and (h). PhCHCH,Ph -+ PhCH=CHPh + H - (129-Q) kcal. (4 (f) ( g ) H+PhCk,CH,Ph + PhCH,CH,. + PhH + 12 kcal. ( h ) (i) ( j ) 2PhCHCH,Ph 3 PhCH=CHPh+PhCH,CH,Ph+ (2Q- 130) kcal. PhCH,CH, --f PhCH=CH, + H - 37-3 kcal. An alternative sequence of reactions could be PhCHCH,Ph 3 PhCH=CH, + Ph - (128-Q) Ph + PhCH,CH,Ph --f PhH + PhCHCH,Ph + (101.6-Q).Since stilbene does not result from this last alternative, the reactions (2) and ( j ) can only be regarded as proceeding parallel with (e) or (f). The heats of the reactions have been deduced using standard heats of forma- tion for the molecules and derived radical heats of formation.ll The value of Q we estimate to be about 70 kcal. The chain reaction possi- bilities of the schemes ( e ) , (g), (h) and (i), ( j ) would seem to be excluded by the endothermicities involved. The small amount of gaseous products obtained is noteworthy, par- ticularly when compared with the data for toluene. In that case the reactions PhCH, + H -+ PhCH,- + H,, PhCH, + H -+ PhH + CH,- were involved and hydrogen and methane found as major products. Step (g) is analogous to the latter and ( k ) to the former.The absence of appreciable hydrogen formation shows that ( K ) cannot occur to any extent. (I) PhCH,CH,Ph + H + PhCH, + PhCH,. + 30 kcal. would seem to be ruled out by the fact that the proportion of toluene in the products can be accounted for by (d). As it is possible to produce the PhCHCH,Ph radical in other ways a study of its behaviour i s being made. The available evidence would appear to support the view that the initial bond fission produces two benzyl radicals and if we take the observed energy of activation as the dissociation energy of the central bond in dibenzyl we find the value agrees very well with that anticipated on the basis of a resonance energy of 24 kcal. €or the benzyl radical. It must be noted, however, that the temperature independent factor is much lower than 1013, the normal value €or a bond dissociation process.It is not proposed to discuss this feature at present as more sensitive methods of analysis may permit work in the near future with lower percentages of decomposition. But even if the present kinetic analysis was in error the magnitude of the decomposition at particular temperatures remains, and unless an extensive recombination reaction of benzyl radicals is 11 Roberts and Skinner, Trans. Farday Soc., 1949, 45, 339. ( K ) PhCH,CH,Ph + H --t PhCHCH,Ph + H, + (103-Q) kcal. The additional possibility of (I)Run 42 43 55 77 41 72 73 70 45 81 57 80 58 40 39 71 44 75 56 74 59 60 35 37 38 47 48 52 68 61 67 66 63 69 64 76 635 647 693 720 C.HORREX AND S. E. MILES Temp. "C 630 642 644 675 695 700 720 743 750 75 7 793 668 692 712 727 741 753 780 699 700 732 740 747 790 675 699 710 738 751 761 774 668 Packed Furnace :ontact Time (sec.) TABLE I 0.80 0.72 0.70 0.69 0.70 0.65 0.66 0-72 0.64 0.70 0'79 0'74 0.19 0'22 0'22 0'22 0'2 I 0'2 I 0'21 0'20 0'2 I 0'20 0'2 I 0.25 0.17 0.67 0.64 0.66 0.66 0.68 0.69 0.61 0-41 0.46 0.60 0.54 Partial Press. (mm. Hg) 0.27 0.25 0.37 0.3 I 0.30 0.3 I 0.3 I 0.3 I 0.35 0.32 0.28 0.23 0.126 0.074 0.116 0.126 0.127 0.126 0.139 0.039 0.0356 0.0326 0.0267 0-0225 0.023 0.039 0.052 0.039 0.039 0.069 0.038 0.052 0.024 0.38 0.24 0-29 % Decomposition 0.40 0.79 0.54 1-08 1-65 3'35 6-50 7-50 9-15 0.242 0.53 0.89 1-62 2'47 3-98 0.945 0'991 1-55 1.525 1-84 4'37 1'45 1-65 3-26 9'25 5'25 6-00 7-20 0.904 1.045 2-50 3'52 1'1 I 2-1 I 14.8 2'0 I 197 k (sec -1) 0.00464 0'01 I0 0.0078 0.0157 0.0159 0.0256 0.0473 0.105 0'0323 0'112 0'122 0.218 0.0124 0.0241 0.0412 0.0647 0.0925 0.119 0.193 0.045 0.049 0.062 0.072 0.0949 0'212 0-021g 0.0260 0.146 0.0795 O*Ogo9 0.123 0.0507 0'0222 0-0228 0-0422 0.0664 envisaged the primary dissociation velocity constant must be close to our value. If we use the method of Butler and Polanyi and analyze by assuming k = 1013 exp (-E/RT), and use for example our value of K at 750° C, we find E = 65 kcal. Equating this to 98 - 2 x resonance energy of the benzyl radical, we obtain 16.5 kcal. for the benzyl radical resonance energy. As this is about the value expected from theoretical calculation, the experimental value of the temperature independent factor is of considerable importance. The authors wish to thank Dr. A. S. C. Lawrence for valuable dis- cussions and t o acknowledge the award to one of them (S.E.M.) of an Ellison Fellowship during the tenure of which this work was carried out. The University, Shefield. In addition we estimate the heat of formation of the PhCH,CH, radical as - 49.4 kcal. by taking 50 kcal. as the value for the dissociation energy of the carbon-iodine bond in PhCH,CH,I (Butler, Mandel and Polanyi, Trans. Faraduy SOC., 1945, 41, 298) and 95 kcal. for the carbon-hydrogen bond. PhCH,CH,-H.