Single-pulse Shock Tube Studies of Hydrocarbon PyrolysisPart 5.-Pyrolysis of NeopentaneBY JOHN N. BRADLEY* AND KENNETH 0. WESTDepartment of Chemistry, University of Essex, Colchester, EssexReceived 29th January, 1975The thermal decomposition of neopentane has been studied in the single-pulse shock tube over thetemperature range 1030-1300 K. The experimental results are fully explained by the generalizedmechanism for alkane pyrolysis proposed by Bradley (Proc. Roy. Suc. A, 1974, 337, 199) providedthe same special features are included i.e. " forbidden " radical isomerizations, hydrogen atom attackon olefins, and '' high " rates for methyl radical abstraction reactions. Optimization by computerof the critical reaction parameters leads to the following rate constants(1)CH3+CSH12 + CH4+CSH11 (2)H+ C4Hs -+ C4H9 (14)neo-C5H12 -+ t-C4H, + CH3k, = 3.3 x 1OI6 exp( - 336 kJ mol-'/RT) s-lkz = 6.6 x 10" exp( - 90 kJ moI-'/RT) dm3 mol-' s-lkI4 = 1.6 x 1O'O exp( - 6.3 kJ rn~l-~/RT) dm3 mol-I s-IThe reaction behaviour is also sensitive to the reactionsCH3 + C4Hs -+ CH4 + CIH, (19)C4H9 -+ C3H6 + CH, (6)C4H9 + CzH4+CzH5 (32)although absolute rate constants obtained for them cannot be regarded as reliable.A general mechanism for the pyrolysis of alkanes at high temperatures has beenproposed previously, using isobutane as an appropriate hydrocarbon to test thearguments.The experimental technique and the computational procedures havesince been improved and the investigation has been extended to neopentane.Neopentane was selected because, as with isobutane, there is only one likelyinitiation step, i.e.,CH3 CH3I II ICH3 CH3CH3-C-CH3 + CH3-C* + *CH3but it is somewhat simpler than isobutane in having all its hydrogen atoms identical.The absence of any additional complexities means that it provides an ideal test of themechanism proposed previously and particularly of the importance of the " forbidden "modes of radical decomposition.EXPERIMENTALThe apparatus and experimental procedure have been described in previous papers inGeneral purpose grade neopentane (>98 %), supplied by B.D.H.Ltd,, was purified bythe s e ~ i e s . ~ - ~J . N. BRADLEY AND K . 0. WEST 9bulb-to-bulb distillation in vacuo followed by preparative g.1.c.using an alumina column.G.1.c. analysis demonstrated that the purified neopentane contained less than O.OO0 1 % ofother C1-C5 hydrocarbons and mass spectral analysis showed that higher hydrocarbons didnot exceed 0.01 %. Experiments were conducted with mixtures of 0.1, 0.5 and 1.0 % neo-pentane in high purity argon. For some runs 100,500 or 10oO p.p.m. of oxygen were addedto the 1 %mixtures.The experiments were conducted on all these m'xtures over the range 1050-1300 K. Thereaction times fell in the range 0.6-1.6 ms and reflected shock pressures in the range 300-400kN m-'.The products were analyzed by g.1.c. using a Porapak N column. This column wasunable to separate propane from propylene and absence of the former was confirmed bysplitting the product stream in two and analyzing one stream on an alumina colum.Integration of the kinetic equations by computer was carried out using the Gear predictor-corrector method as before, but in this case the integration procedure formed a sub-routine within an optimization program which manipulated the rate constants until a satis-factory fit to the experimental data had been achieved.This progrm is basically that devel-oped by Broyden for the solution of simultaneous non-linear differential equations, withmodifications by Dr. J. Ford of the Computing Centre, University of Essex. The mostimportant of these is that the rate constants are converted to trigonometric functions in orderto keep them within prescribed bounds. In this example, with about thirty rate equations andfive adjustable rate parameters, optimization typically required 80-100 iterations, with amaximum of about 200.RESULTSOverall rate constants for neopentane pyrolysis have been calculated from the lossof the parent hydrocarbon assuming reaction orders of 5, 1 and 3.No significant3.0nrl L-Y,--.-0rlM0 -2.00.8103 KIT1.0FIG. 1 .-Arrhenius plot of the apparent overall first-order rate constants for neopentane pyrolysis. +, 0.1 % neopentane ; x , 0.5 neopentane ; 0, 1.0 % neopentane ; V, 1 % neopentane with100 p.p.m. oxygen; A, 1 % neopentane with 500 p.p.m. oxygen; H, 1 % neopentane with 1OOOp.p.m. oxygen10 PYROLYSIS OF NEOPENTANEdifference could be detected in the standard deviations obtained by least-squaresanalysis of Arrhenius plots based on each of the three sets.An Arrhenius plot ofthe first-order rate constants is shown in fig. 1 and the data are fitted best by theexpression k = 2.88 x lo9 exp(- 159 kJ mol-'/RT) s-l although this should not betaken to imply that the reaction obeys first-order kinetics. Addition of oxygen tothe neopentane had no measurable effect on the nature and extent of the decomposi-tion although similar amounts are known to affect the pyrolysis of fluorinated hydro-carbons.The major products of the reaction were isobutene, ethane and methane, withsmaller quantities of allene, methylacetylene, propylene, acetylene and ethylene. Noattempt was made to analyze for hydrogen although mass balance considerations showthat it must have been formed.At conversions greater than 50 %, minor quantitiesof pentene and of isobutane were also detected. The amounts were too small to allowidentification of the pentene isomers. The product yields changed with temperatureand to a lesser extent with time, the most significant change being the increase inconcentration of ethylene and propylene with temperature at the expense of theisobutene. These effects of temperature on the product yields are illustrated in fig. 2.20113M)0-I100 12000I ICO 1200temperature/KFIG. 2.-Variation of product yield with temperature at a constant reaction time of 1 ms.In the absence of a unique reaction order, any data averaging must be treated assuspect and for the purpose of subsequent analysis ten representative experiments wereselected which covered the range of the experimental conditions.All the experimentschosen were free from measurable shock attenuation and showed product recoveriesin excess of 90 % of the original reactant. These experimental data are listed intable 1J . N. BRADLEY AND K . 0. WEST I1COMPUTER SIMULATIONIn an earlier paper,l the following schematic mechanism for alkane pyrolysis wasproposed :initiation : R-R’ + R+R’(rupture of C-C only)(attack by CH3 and H on parent)propagation : CH3+RH + CH,+RH + RH + H2 + Rradical decomposition(to H or lower radical + olefin)ethane reactionsR -+ olefin+ R’H atom addition to olefins H+olefin + R.On this basis the detailed mechanism for neopentane pyrolysis can be represented byreactions (1) to (16) in table 2.As one of the main objectives of the present studywas to assess the importance of the so-called “ forbidden ” isomerizations, thesereactions were excluded from the mechanism in the first instance.tempera-turc/K1295126512301220120511801170114011301100TABLE 1 .-RESULTS OF EXPERIMENTS SELECTED FOR COMPUTER SIMULATIONreaction comer-timelms sion/% CH4 C2H4 C& C2H2 C3H6 C3H4-A C3H4-M i-Caa i-C4Hlo i-CsHlo1.05 58.3 8.9 14.7 16.3 1.9 5.5 10.5 11.5 29.6 0.8 0.31.1 51.7 7.6 9.4 16.8 1.6 4.5 9.0 7.5 43.1 0.4 0.10.85 25.7 5.3 3.0 16.8 0.7 2.0 4.5 2.2 65.0 - -0.7 16.5 4.7 1.9 16.7 0.3 1.5 2.9 1.2 70.8 - -0.8 15.4 4.6 0.9 16.1 0.2 0.9 1.5 0.6 75.2 - -1.0 14.6 4.9 0.4 15.5 0.1 0.6 1.1 0.2 77.2 - -0.95 32.9 5.8 4.0 17.1 0.9 2.8 5.1 3.7 60.6 - -1.2 24.5 5.3 1.8 16.1 0.3 1.3 3.1 0.9 71.2 - -0.9 14.2 5.3 0.4 15.2 0.1 0.6 1.0 0.2 77.2 - -0.6 6.1 4.9 0.2 15.2 - 0.3 0.4 - 79.0 - -The yields are quoted as percentages of the total quantity of neopentane decomposed.C3H4-A and C3H4- M denote allene and methylacetylene respectively.To avoid confusion, no attempt was made in the model to distinguish the con-centrations of the different radical isomers present.This means that C4H9 denotesboth t- and iso-species and C3H7 both iso- and n-species. This gives the mechanismadded flexibility and removes some unwanted “ degeneracy ” which arises whenseveral reactions lead to a kinetically-identical result.As it stands the mechanism is inadequate to explain the experimental results sinceit contains no routes leading to allene, methylacetylene and acetylene. Acetylenewas only a minor product and it seemed reasonable to make the assumption that itwould be formed either from ethylene, e.g., via the sequenceCH3 + C2H4 + C2H3 + CH4H+ C2H4 + C2H3 + H2C2H3 + C2H2 +H,or from a precursor of ethylene, e.g., via the ethyl radicalC2H5 + C2H3+H2CZH3 + C2H2 + H12 PYROLYSIS OF NEOPENTANERather than complicating the mechanism further, the small acetylene yield wastherefore added to the amount of ethylene formed.A corresponding assumption was not considered suitable to deal with the alleneand methylacetylene because of the larger quantities involved and it was thereforenecessary to incorporate in the mechanism a reaction sequence leading to C3H4, nodifferentiation being made between the two isomers. It is believed that C3H4 isformed from the product, isobutene, either by direct decompositionC4Hs -+ C3H5 +CH, (17)C3H5 -+ C3H4+H (1 8)TABLE 2.-sUMMARY OF REACTIONS AND RATE DATA USED IN INITIAL SIMULATION-13-14IIII111III1IIIII1133-1I1I1reactionAl Kassel(dm3 mol-1 s-1 or s-1) E/kJ mol-1 correction12 -+ C4H9+CH3 (1) 1 . 1 7 ~ 10171 -+ C4Hs+ CH3 (4) 1 .Ox 1014-I9 -+ C3Hs+CH3 (6) 1 .6 ~ 101437 4 C2H4-k CH3 (8) 4'0 x 101312 -+ CH++CSHl1 (2) 5 . 0 ~ 10"12 -+ Hz+CSHll (3) 1 . 0 ~ lo1'39 + Cq.Hs+H (5) 4 . o ~ 10133 7 -+ CsH6-I-H (7) 2 .0 ~ 10143 5 -+ CJI4+H (9) 3 . o ~ 1013H6 -+ CH4+CzHS (10) 5 . 0 ~ 10"& j -+ Hz+CzH5 (11) 1.oX 10"36 S CHjfCH3 (12,13) 5 . 0 ~38 C4H9 (14) 3 . 2 ~ lolo34 + C2H5 (16) 9 . 3 ~ 1O1O-+ C3H4+H (18) 3 . 2 ~ loz3H7 + C3H4+ CH3 (21) 1 .Ox lOI436 -+ C3H7 (15) 7 . 2 ~ 10938 -+ C3H5$CH3 (17) 6.3 X3 8 4 CH4+C4H7 (19) 5 . 0 ~ 10''38 Hz+C4H7 (20) 5 . 0 ~ 10''H6 -+ CH*+C3H5 (22) 6.3 X 10''k&j -+ 8 2 + CBHS (23) 5.0 X 10"12 4 C4H1o+CsHll (24) 5 . 0 ~ 10":12 -+ C ~ H ~ + C ~ H ~ I (25) 5 . 0 ~ 1O'l112 4 CZH6+C5Hll (26) 2 . 0 ~ 10'33 --* GHiz (27) 1.ox 10''33 * C4H10 (28) 1 . 0 ~ 1O'O34310915718013717313817038.589.938.56.35.011.737036720110920320.932.220.910910952.700or by the abstraction of hydrogenCH3 +C4H8 + C4H7 + CH4H + C4H8 -+ C4H7 +H,C4H7 -+ C3H4+CH30.9 --0.90.50.50.230.230.16--0.13i --0.50.23 --0.6----I --classIIII11111I11VVI11II1I, 1IVVIVIVII1I11VVIVVVVV(19)(20)(21)and both sequences were added to the reaction mechanism.Abstraction frompropylene was also included as a possible source, e.g.,CH3 + C3H6 -+ CH4 + C3H5 (22)H+C,H, -+ H P + C ~ H ~ . (23)Formation of C3H5 radicals directly from propyleneC3H6 + CSHS +J . N. BRADLEY AND K . 0. WEST 13is a much slower process becay$e it involves rupture of a C-H rather than a C--Cbond.Although previous work had demonstrated that attack by radicals other than Hand CH3 on the parent hydrocarbon was unimportant, it was considered desirable toreassess this hypothesis, particularly for C3H5, and reactions (24) to (26) representingattack by C4H9, C3H5 and CzH5 respectively were included in the initial analysis.In the same way, reactions (27) and (28), representing radical combination, wereincorporated even though the earlier work had illustrated that methyl radical combi-nation to ethane was the only reaction of this type which had to be considered.Rate constants for reactions (4), (5), (6), (7), (8), (9), (lo), (12), (f4), (15), (16),(22), (23) and (26) were taken directly from the references quoted in table 2.kl wasobtained by averaging the results of Tsang l4 and of Halstead et aZ.l Because of thepresent uncertainty in the rates of methyl radical abstraction reactions, k, was treatedas a variable throughout, the value quoted in table 2 merely serving as a suitablestarting point.k3 was estimated by combining the activation energy quoted by Trostand Steacie l6 with the pre-exponential factor for the corresponding reaction withi~0butane.l~ kl has been measured by several groups of workers and an averageof the literature values was emp10yed.l~’~~ k13 and k24 were derived from the corre-sponding forward reactions k,, and kl and the appropriate equilibrium constants.k17 has been measured in this laboratory : 2o separation into Arrhenius parameterswas achieved by estimating the activation energy from thermodynamic data. kl 8 andkZ1 were obtained by considering the entropy changes involved in the transition statein order to predict the pre-exponential factors, the activation energies being derivedfrom the overall thermochemistry. k19 and k,, were arbitrarily set equal to k, andkzo to k23.kZ7 and k28 were assigned the value of lolo dm3 mol-l s-l typical ofradical recombination reactions.The complete system was then integrated by computer for several sets of experi-mental conditions. For comparison with the experimental data, it was assumed thatmethyl radicals present at the end of the reaction time combined to form ethane : thiswas borne out by computer simulation of the expansion process. The other radicalconcentrations were sufficiently low to be neglected. Each rate constant in turn wasmultiplied by a series of factors between 50 and 0.02 and also by zero.Histogramsfrom the computations are illustrated in fig. 3 for a typical experiment conducted at1200 K with a reaction time of 900 p s . From such histograms it is possible to cate-gorize the importance of each step in terms of the classification suggested in theprevious paper :(I) reactions whose rates determine the overall kinetics ; (11) reactions which havelittle effect on the overall rate but influence the product distribution; (111) reactionswhose rates are unimportant, above a certain minimum, but are nevertheless essentialto the mechanism ; (IV) reactions which definitely occur but which have little influenceon the reaction; (V) reactions which do not occur to any significant extent.The classification obtained has been included in table 2.It will be immediately apparent that the arguments of the earlier paper (a) thatattack on the parent hydrocarbon by radicals other than CH3 and H, and (6) thatradical recombinations other than methyl + methyl are unimportant, are confirmed bythe present work.It is therefore possible to omit reactions (24) to (28) from furtherconsideration.The analysis also shows that the C3H4 compounds arise due to the abstractionof a hydrogen atom from the isobutene product (1 9) and (20) and do not involve eitherdirect decomposition of isobutene (17) or hydrogen atom abstraction from propylen14 PYROLYSIS OF NEOPENTANE(22), (23). It is worth commenting that all subsequent reactions of propylene appearto be unimportant although the same cannot be said of isobutene.It is thus possible to associate the characteristic reaction behaviour with the ratesof six reactions and to a considerable extent to correlate specific features with specificrate constants, thus :reaction (1) controls the overall reaction rate ; reactions (12) and (1 3) control themethyl radical concentration responsible for propagation ; reaction (2) controls thepropagation rate and the [CH,] yield; reaction (19) controls the [CH,] and [C,H4]yields ; reaction (14) controls the [C,H,] yield ; reaction (1 1) controls the [C,H4] yield.GOLO20030201001050302010030 r90 r7050302 1 6 8 10 12 1L 16 18 20 22 U 26 28reaction numberFIG.3.-The effect on the computed product yields of varying the values of the (non-optimized) rateconstants for a typical experiment at 1200 K with a reaction time of 900 ps (shaded area denotesx 10, open area denotes x 0.1).Attempts were then made to optimize the mechanism by varying the rate constantsk l , kZ, klz, k14 and kI9 until the computed yields of the major products matched thoseobserved experimentally. The rate of abstraction of a hydrogen atom from an alkanewas further assumed to be linearly dependent on the number of hydrogen atoms avail-able. k l l was then allotted a value equal to one-half the quoted value of k3 and k l J . N. BRADLEY AND K . 0. WEST 15was varied with k2 to maintain the relation k,, = 0.5 k2. k,, was tied to k12 viathe equilibrium constant.As with isobutane, all attempts failed to match the experimental yield of ethylene.The only way this problem could be overcome within the bounds of the mechanismwas to relax the constraint that kll should equal 0.5 k3.However optimizationrequired the quite unacceptable situation that kll should be two orders of magnitudegreater than k3 and even then the mechanism failed to account for the decrease inrelative yield of isobutene with temperature.The only acceptable solution was to permit the occurrence of "forbidden"isomerizations as in the isobutane pyrolysis. The available reactions areThe yield of pentene is quite sinall so that reaction (29) is unlikely to be of significance.Reactions (30) and (31) both generate C2 and C3 hydrocarbons in equal amounts andwould not provide the necessary " flexibility " in the ethylene yield.Reaction (32)serves the purpose admirably since it provides a unique source of ethylene : furthermorethe addition of hydrogen atoms to isobutene by reaction (14) generates radicals withexcess energy in two isomeric forms : t-butyl (CH3)3C- and isobutyl (CH3)2CHCH2*Rearrangement of excited isobutyl radicals certainly seems more feasible than the otherforbidden isomerizations.With the inclusion of k32 as an additional variable parameter, it was necessary tospecify the value of one of the other rate constants and that of k12 was chosen ashaving the most reliable literature value. It was then possible to determine rateconstant expressions for the six reactions which satisfied the complete set of experi-mental data.These are collected in table 3.TABLE 3 .-RATE CONSTANTS OBTAINED BY COMPUTER OPTIMIZATIONreaction rate constant expressionC5H12 -+ C4H9+CH3 (1) kl = 5 . 0 ~ 1015 expCH3+C5HI2 -+ CH4+C5H11 (2) k2 = 6 . 6 ~ lo1' expH+C4Hs + C4H9 (14) k14 = 1 . 6 ~ 1O'O expCH3+C4H8 -+ CH4+C4H7 (19) kI9 = 2 . 6 ~ loi3 exp(-318 kJrn~l-~/RT)s-l(- 90 kJ mol-'/RT) dm3 rnol-' s-'(- 6.3 kJ rnol-'/RT) dm3 rnol--l s-'(- 11 1 kJ rnoP1/RT) dm3 rnol-I s-'(- 92.1 kJ mol-l/RT)C4H9 3 C3H6+CH3 (6) kS2/k6 = 7 . 2 ~ lo3 expC4Hg -+ C2H4+ CzHS (32) }standarddeviationin loglo k0.1 10.220.290.820.12It should be noted that, as the number of variable rate constants in the computersimulation was chosen to match the quantity of independent experimental data avail-able, the fit was exact, i.e., to within the rounding errors.Uncertainty in the experi-mental measurmeents then appeared as scatter on the Arrhenius plots used to providethe complete rate constant expressions. The standard deviations listed in table 3shows that the rate constant expressions for k , , k , and k4 are very sntisfxtory bu16 PYROLYSIS OF NEOPENTANEthat the reaction for (19), which is a secondary reaction involving the olefin product,is far less reliable. It was possible to show that no other set of rate constants wouldfit the experimental data, assuming that the correct mechanism had been obtained.There is, of course, no way of demonstrating that the mechanism itself is unique,which is the reason why this particular approach has been adopted.Trace quantities of isobutane and pentene were observed at the highest tempera-tures.Computations showed that the former could be accounted for by reaction (24)provided k24 was about one-tenth the value of k2. This value seems reasonableconsidering the steric restrictions which inevitably accompany the reactions of t-butylradicals.Various routes to the pentenes are available. The simple radical recombinationsC3H5 +C2H5 + C5H10C4H7 +CH, + CSH10C3H7+C2H3 -+ C5H10were tested with rate constants of 3 x 1O'O dm3 mol-' s-l and failed to produce sig-nificant quantities of pentene. The isomerization of the neopentyl radical by reaction(29) is another possibility but seeins unlikely as the radical is not formed with excessenergy.A more probable route is via the addition of methyl radicals to isobutenewhich yields (CH3)2c CH2CH3 with excess energy in a reaction directly analogousto (14).CH3 + CH3--C=CH2 --+ CH3-C*-CH2-CHjI ICH3 CH3The three decomposition reactionsCH3-C-CH2-CH3 + CH2-C-CH2-CH3 + HI ICH3 C&1CH3ICH3+ CH3-C=CH-CH3 + H-+ CH3-C-CH2 + CH3all obey the P-bond breaking rule and if the first two are together responsible for 10 %of the total this would account for the yield of pentene observed.An attempt was made to simulate the small yields of acetylene by means of themechanismCH3 + C2H4 -+ CH4 + C2H3H + CzH4 --+ H2 + C2H3C2H3 + C2H2+Husing literature values for the first two rate constants 11* 21 of 2.0 x lo1' exp(-41.8kJ mol-l/RT) dm3 mol-l s-l and 1.82 x 1O'O exp( -27.6 kJ mol-l/RT) dm3 mol-' s-'respectively.As the mechanism contains no other reactions of C2H3, the third rateconstant may be chosen quite arbitrarily. It proved impossible to reproduce theacetylene yields even by increasing these rate constants by two orders of magnitude.It seems therefore that the alternative routesuggested in an earlier paperCZH, + C,H,+H,must be responsible for acetylene formationJ . N. BRADLEY AND K. 0. WEST 17DISCUSSIONThe most important conclusian of this work is that the mechanism proposed inthe investigation of isobutane pyrolysis applies to the pyrolysis of neopentane. Notonly is the same basic mechanism involved but also it is necessary to include the samethree types of process as before, i.e., radical decomposition via “ forbidden ” routes,methyl radical attack on alkanes with “ high ” reaction rates for such reactions, andaddition of hydrogen atoms to olefins.Although it is possible that other alkanes mayreveal particular problems, it is evident that the general mechanism for alkane pyrolysismay be considered proven. The work has alsO produced rate constants for six hightemperature reactions. Since the experimental technique has been refined followingthe previous work, the present findings should be more accurate.It is worth mentioning here that the previous paper implied a discrepancy of abouta factor of ten between the initiation rate of isobutane pyrolysis measured by thepresent technique and that recorded earlier by Konar, Marshall and Purnell.22 As0.8 1.0 1.2 1.410’K/TFIG.ci.-Comparison of rate constant reiations for the initiation reaction (1).this cast some doubt on the validity of the technique, it is reassuring that these authors 23have now reassessed their measurement and have reported a revised value for the ratetechnique which effectively removes the discrepancy.The optimization leads to a rate constant k, for the initiation reaction of neo-pentane pyrolysis of 5.0 x 10’ exp( - 3 18 kJ mol-’ /RT) s-l . This rate constant iscompared with the results of Tsang,14 Halstead et aZ.15 and Baronnet et aLz4 in fig. 4and it will be observed that the agreement is excellent.This is very reassuring sincethere have been unexplained discrepancies between rate constants obtained with th18 PYROLYSIS OF NEOPENTANEsingle-pulse shock tube in earlier work. The temperature range is too limited toprovide an accurate breakdown into the separate Arrhenius parameters and as RRKMcalculations show that the reaction must be close to its high pressure limit it is moresatisfactory to select an activation energy equal to the bond dissociation energy.The rate constant expression then becomes 3.3 x 10l6 exp( - 336 kJ mol-l/RT) s-l.As the rate constants for reactions (2) and (10) were constrained by the relationk l o = 0.5 k2, it is more satisfactory to discuss the results in terms of klo for whichother measurements are available.At 1200 K, Clark, Izod and Kistiakowsky ’’estimated a value of 2 x lo8 dm3 mol-1 s-l while Pacey and Purnell 26 obtained6 x lo7 dm3 mol-1 s-l. The present work gives a value of 4 x lo7 dm3 mol-’ s-l whichseems to support the lower of the two earlier measurements. The rate constantexpressions for k2 and klo seem entirely justified and add further weight to theobservation of non-Arrhenius behaviour for such reactions.The value for k14 is one-half that reported previously lo from measurements madeat lower temperatures and the agreement must be considered very satisfactory. It isdoubtful whether any rate constants reported for hydrogen atom addition to olefinscan justifiably be extrapolated to temperatures above 1000 K and there is considerableneed for additional experimentation in this area.The value obtained for the ratio kS2/k6 cannot be considered accurate as the productconcentrations are not particularly sensitive to this ratio.However there does seemto be a significant discrepancy between the present value and the value of 1.007 x lo6exp( - 173.7 kJ mol-l/RT) quoted previous1y.l This may be associated with the factthat, in the present system, C4H9 radicals are formed by H atom addition to olefinsso that both iso- and t-isomers occur, initially with excess energy. The reactionpressures also differed substantially between the two investigations.The value estimated for k19 must be regarded as suspect, first because it is a second-ary reaction involving one of the products, and secondly because other reactionssuch as (20) could be involved. The rates of the latter cannot be satisfactorily assessedbecause they are in competition with other hydrogen atom reactions to which thedecomposition is insensitive since they fall in class III.The authors wish to express their gratitude to the Hydrocarbon Research Panelof the Institute of Petroleum for the award of a studentship to one of them (K.0. W.).J. N. Bradley, Proc. Roy. SOC. A, 1974, 337, 199.J. N. Bradley and M. A. Frend, Trans. Faruhy Soc., 1971, 67, 1.J. N. Bradley and M. A. Frend, J. Phys. Chem., 1971,75,1492.J. N. Bradley and K. 0. West, J.C.S. Faruhy I, 1975, 71,967.C. G. Broyden, Mathematics of Computation, 1965, 19, 577 ; Computer J., 1969, 12,406.J. N. Bradley and K. 0. West, unpublished results.S. W. Benson and H. E. O”ea1, Kinetic Data on Gus Phase Unitnolecular Reactions (NationalBureau of Standards, NSRDS-NBS21, 1970).P. D. Pacey and J. H. Purnell, J.C.S. Furudzy I, 1972, 68, 1462.(Butterworth, London, 1972).’ K. H. Anderson and S. W. Bcnson, J. Chem. Phys., 1964, 40, 3747.lo J. A. Kerr and M. J. Parsonage, Evaluated Kinetic Data on Gas Phase Addition Reactionsl 1 A. F. Trotman-Dickenson and E. W. R. Steacie, J. Chern. Phys., 1951, 19, 169.l 2 B. de B. Darwent and R. Roberts, Disc. Faruhy Soc., 1953, 14,55.l 3 P. J. Boddy and E. W. R. Steacie, Canad. J. Chem., 1960,38, 1576.l4 W. Tsang, J. Chetn. Phys., 1966, 44, 4283.M. P. Halstead, R. S . Konar, D. A. Leathard, R. M. Marshall and J. H. Purnell, Proc. Roy.Soc. A, 1969,310,525.IG W. R. Trost and E. W. R. Steacie, J. Chem. Phys., 1948, 16, 361J . N. BRADLEY AND K. 0. WEST 19l7 A. F. Trotman-Dickenson and G. S. Milne, Tables of Bimolecular Gas Reactions (NationalBureau of Standards, NSRDS-NBS9, 1967).K. Schofield, Planet. Space Sci., 1967, 15,643.l9 R. R. Baidwin and A. Melvin, J. Chem. SOC., 1964,1785.2o K. 0. West, Ph.D. Thesis (University of Essex, 1975).21 V. V. Voevodsky and V. N. Kondratiev, Progr. Reaction Kinetics, 1961,1,41.22 R. S. Konar, R. M. Marshall and J. H. Purnell, Trans. Farahy Soc., 1968,64,405.23 R. S. Konar, R. M. Marshall and J. H. Purnell, Int. J. Chem. Kinetics, 1973, 5,1007.24 F. Baronnet, M. Dzierzynski, R. Martin and M. Niclause, Compr. rend. C, 1968, 267, 937.2 5 T. C. Clark, I. P. J. Izod and G. B. Kistiakowsky, J. Chem.Phys., 1971, 54, 1295.26 P. D. Pacey and J. H. Purnell, J.C.S. F4ruday I, 1972, 68,1462.(PAPER 5/196