EVIDENCE RELATING TO THE COMBUSTION OF HYDROCARBONS BY R. G. w. NORRXSH Received z z n d March, 1951 It is generally agreed that the combustion of hydrocarbons occurs by way of a chain process showing delayed branching, but there have been differences of opinion about the identity of the intermediate compound essential to this kinetic conception. In the present memoir some new evidence bearing on the problem -not all of which has yet been published-is cited. The kinetics of the combustion of gaseous hydrocarbons exhibit certain common characteristics which are mechanistically explained by the conception of degenerate branching. According to this idea, which has been fully expounded by Semenoff,l the auto-catalytic character of the reactions and the long induction periods are associated with the build-up of a moderately stable intermediate compound by way of a chain reaction, and this intermediate is responsible for a delayed branching, leading to the slow multiplication of reaction centres.While there is not any serious difference of opinion about the general kinetic character of the reactions, there are more than one hypothesis as to the exact nature of the reactions involved and the relevant intermediate substances re- sponsible for the branching reactions. It is the purpose of this paper to cite some fresh evidence which nominates the intermediate aldehydes as the most generally important branching agents, and to show how the oxidation of hydrocarbons can be represented as an attack on the hydrocarbon molecules by hydroxpl radicals. It is thought that the mechanism proposed, which does not exclude the participation of intermediate peroxides at low temperatures, gives a logical interpretation of the phenomena with reasonable economy of hypothesis.The Concentration of 1ntermediates.-If AT is the average life of the molecules of the intermediate substance in seconds, and Y is the maximum rate of reaction of the hydrocarbon in mm. Hglsec., then the maximum pressure reached by the intermediate during the reaction is YAT mm., and the concentration of the intermediate must be greatest at the time of maximum rate. Now the rate of a branched chain reaction can be shown to conform to the relationship Y = A(e4t - I) and when 9 the net branching factor is greater than zero the reaction accelerates either towards ignition or degenerate explosion.In the latter case characteristic of hydrocarbon oxidation the time of development to maximum velocity is of the order of minutes and hours. And arguing on kinetic grounds, Semenoff has concluded that the average life of a chain centre before it enters into a branching reaction, is of the order of seconds or minutes. Thus for a value of AT of 60 sec., and a maximum reaction rate of 0-1 mm. of hydrocarbon oxidized per second, the maximum pressure of intermediate is YAT = 0.1 x 60 = 6 mm. oxidation indicates an intermediate responsible for delayed branching, In general it may be judged that the time scale of the rate of hydrocarbon sufficiently stable to reach pressures of the order millimetres when the reaction is proceeding isothermally at an ordinary measurable velocity.Semenoff, Chemical Kinetics and Chain Reactions (rg35), p. 68. 269270 COMBUSTION OF HYDROCARBONS With both methane and ethylene the pressure of formaldehyde pro- duced as intermediate reaches values of this order, but the concentration of hydroperoxide, if formed, is too small to be measured. Upon the basis of this calculation it is not present in sufficient concentration to allow it to function as the essential agent in the degenerate branching process. That formaldehyde IS a in these two cases is the functioning intermediate is shown also by the fact that its addition to the system reduces the period of acceleration (induction period) and in sufficient quantity completely eliminates it. Further, Dr.Harding' has found that if more formalde- hyde be added than is necessary completely to suppress the induction period, the reactions will start immediately at an enhanced rate, which will rapidly decline to the normal rate as the pressure of formaldehyde adjusts itself to the value corresponding to the stationary state proper to the reaction parameters. In such a case we mag speak of a negative induction period. The Effect of Light on the Oxidation of Methane and Ethylene.- These considerations lead to the conclusion that any influence favouring the dissociation of the formaldehyde into free radicals should have the effect of increasing the net branching factor with a consequent reduction of the time of development of the reaction and an acceleration of the maximum rate.This effect was achieved by Dr. D. Patnaik6 and the author, using a very powerful beam of ultra-violet light of wavelength A 3800-2400 A which is known to dissociate formaldehyde into H atoms and carbon monoxide. The reactions (methane at 484-5' C and ethylene at 431.5' C) responded immediately to the effect of the light, accelerating when the light was admitted and decelerating when it was excluded, and the induction period was reduced. In cases near the thermal ignition limit the slow reaction could indeed be converted to ignition by irradiation. A further exam- ination of the kinetics both in the light and the dark showed that the effect of the light was merely to augment the thermal process without altering the course of its kinetics, and that while the rate RD of the dark reactions is represented by the expression the rate in the light R, is given by where (Hy) is the pressure of the hydrocarbon, I the intensity of the light, and P the total pressure in the system, the diameter and surface being kept constant. If, therefore, we subtract the rate in the dark from that in the light we see that the augmented reaction due to irradiation follows a similar law to the dark reaction, with the intensity of light re- placing the oxygen pressure.The kinetic results are explained in detail if the delayed branching due to the formaldehyde originates from its thermal oxidation by the reaction * RD = K W Y ) (0,) + KAHy) * ( O P B, = K,(HY)(02) + K,'(HY)I + K2(HY)2(OIP + KawYT)ZIP R, - RD = KI'(Hy)I + K,'(Hy)'IP /OOH 0 HCHO+O,= [ HC ,o ] =HCCoH+ 0, AH-0, Bone and Gardner, Pvoc.Roy. SOC. A , 1936, 154, 297. 9 Foord and Norrish, Proc. Roy. Soc. A , 1936, 157, 503. 4 Harding (not yet published). 5 Norrish and Patnaik, Nature, 1949, 163, 883. * In what follows, the reaction of an aldehyde with oxygen t o form oxygen atoms will frequently be postulated. It is believed that the process is associated with the formation of a peracid which is known to occur a t low temperatures, possibly by a chain reaction involving RC radicals, as formulated by Ubbelohde 80 and by McDowell and Thomasa1 /oo- N OR. G. W. NORRISH and in the photochemical reaction from H2C0 + hv = H + HCO \ H + CO. The effect also persists in the spectral region h 3500-3800 A where the formaldehyde does not yield H atoms directly and must be ascribed to some form of induced predissociation, such as H2C0 + h u = H2C09 ‘€€,COX + X = H + HCO + X which has been already postulated by Henri to set in at elevated tem- peratures. Very similar results have been obtained by Dr.Garbatski and the author for the photochemical oxidation of ethane. It must be stressed that we have, in all the above cases, influenced thermal gas reactions photochemically in spite of the fact that none of the primary reactants show the slightest absorption of the light used. The photochemical effect is exerted through the intermediate product, a result which in itself provides confirmation of the kinetic conception of delayed branching. It is further to be noted that hydroperoxide and hydrogen peroxide do not absorb light of wavelength longer than 3000 while the photochemical effect persists up to 3800 A The kinetic requirements of both the thermal and photochemical oxidation of methane and ethylene are satisfied by a simple extension of the scheme advanced for these reactions by the author.6 For methane this takes the form Initiation and degenerate branching CH4 + O2 = CH20 + H,O H2CO + 0, = H,CO, + 0 0 + CH4 = CH, + OH OH + CH4 = CH, + H,O CH, + 0% = H2C0 + OH propagation 1 coupled formalde- H + 0, + H,CO = CO + H,O + OH CH@ + OH = CO + H,O + H H + 0, + X = HO, + X’ OH + surface + termination termination * 1 photochemical H2C0 + X + hu = H + H + CO + X } branching- H&O + hv = H + H + CO It may be said to be an academic question whether the CH,00 radical takes part.and the radical may have a transitory existence, but at the temperatures At higher temperatures of hydrocarbon oxidation the transient peracid is assumed to eliminate an oxygen atom Reaction (5) might be written : CH, + 0, --+ CH,COO H,CO + OH RC/ooH) -+ \O The net overall thermal value of the reaction RCHO -!- 0, = RCOOH + 0 is zero, and in all cases where the latter reaction is written an intermediate tiansition process of the above type is to be undeistood. 6 Norrish, Colloquium on Reactions of Inflammation and Combustion in Gases (C.N.R.S.) (Pans), 1948, 16, or Revue de L’Inst. France de Petrole, 1949, 4, 288. * Reactions (7) and (8) are only formally to be regarded as ternary reactions. The transition complex HO, formed with ca. 50 kcal. is unstable unless deactivated (reaction (8)) and reactive only if it encounters a formaldehyde molecule (re- action (7)) before it decomposes or is deactivated.272 COMBUSTION OF HYDROCARBONS of the oxidation of methane there is no evidence of the reaction for no trace of methyl hydroperoxide can be detected. Argument from Effect of Temperature on Velocity of Reaction.- Further kinetic analysis of the process of degenerate branching shows that the overall temperature dependence of the reaction, when self heating is excluded, is mainly dependent on the energy of activation of the branch- ing reaction furnished by the intermediate.Dr. Harding' and the author have shown that for ethylene the value of the energy of activation varies in a parallel way with that of formaldehyde from 2 6 kcal.at 400' C to 53 kcal. at 550" C. They conclude that two processes of degenerate branching are involved, the reaction predominant at lower temperatures giving place to H2C0 + O2 = HO, + HCO, AH =,- 50. It may be, however, that traces of olefinic peroxide are formed with ethylene at the lower temperatures (though none was ever detected in our work in spite of careful search) for there sometimes occurred a small drop in pressure during the induction period. C2HI + 0, = C2HI02 = &H,O is not excluded. The Oxidation of Higher Params.-The ignition phenomena of higher paraffins, from propane upwards and possibly ethane, fall into two groups : the higher temperature ignitions in the region above 450' C, and the cool flame phenomena in the temperature range between ca.260-400' C. This is illustrated in Fig. I, due to Newitt and Thornes.* The Higher Temperature Ignition.-The kinetics associated with the higher temperature ignition appear to follow closely the pattern of those of methane, both as regards the time scale and intermediates. They have all the characteristics of delayed branching reactions and with propane at pressures below atmospheric outside the cool flame region, Mr. Galvin finds similar dependence of the reaction rate at 395" C on partial pressures, surface and total pressure, as is found for methane, and a similar effect of irradiation. Aldehydes are easily detected at concentrations consistent with their functioning as the essential intermediate while peroxides, Q at these higher temperatures, are unstable and, if formed, their concentration would appear to be too low to satisfy the kinetic requirement of degenerate branching.Thus the high temperature oxidation of higher paraffins can be represented by a similar mechanism to that of methane, modified by reactions to account for the observed degradative oxidation of the alde- hydes to formaldehyde and the final products, carbon monoxide and water. We may write : CH,COO + CHI = CH,OOH + CH, H,CO + 0, = H2C02 + 0, AH = 0, A reaction such a s RCH,CHO + 0, = RCH,COOH + 0 Initiation and branching RCH2CH3 + 0, = RCH,CHO + H20 RCH,CH, + 0 = RCH,CH, + OH propagation RCH,CH, + 0, = RCH,CHO + OH RCH,CH3 + OH = RCH,CH, + H,O RCH,CHO, + O H = RCH, + CO + H,O , c o ~ ~ ~ d ~ RCH, + 0, = RCHO + OH gradation of aldehyde.Termination may take place a t the surface or in the gas phase by the recombination of propagating radicals. In addition, alcohols may be 7 Harding and Nomsh, Nutwe, 1949, 163, 797. 8 Newitt and Thornes, J . C!;enz. SOC., 1937, 1656. Q Neumann, A d a Physzcochzm., 1938, 9, 527.R. G. W. NORRISH 27 3 expected to result from an attack on the secondary hydrogen atoms of the hydrocarbon by hydroxyl radicals, e.g., RCH,CH3 + OH = RCHCH, + H,O RCHCH, + 0, = RCHO + CH30 CH30 + RCH,CH3 = CH,OH + RCH,CH, and Mr. Knox, in agreement with Newitt and Thornes,8 has recently found measurable quantities of alcohols in the oxidation of propane at 390° C. In addition, Mr. Galvin has been able strongly to accelerate the oxidation of propane at 3 8 1 O C (in regions of pressure and temperature outside the cool flame zone) by irradiation with light of wavelength 3000-3800 A, a part of the spectrum which is not absorbed by peroxides.CURVE POR ICNlTlON OF CJ Ha + 0, MIXTURES. AFTER NEWlTT AM THORNES. 5 Cool Flames LOW Temperature 2 Codflamas Ignltion ReTim. I I I I 200 400 600 PRESSURE rn m. 9. FIG. I . There is thus evidence that the kinetics of the high temperature oxidation of the higher hydrocarbons conform to those of the oxidation of methane and it is in accordance with the arguments advanced here to conclude that formaldehyde which is the final product of the oxidative degradation of aldehydes and which, owing to its greater stability accumu- lates to intermediate concentrations much greater than the higher alde- hydes, begins to function as a branching agent only at the higher tem- peratures.Thus, as the temperature rises toward the upper ignition limit, ,$ will increase due to the participation of the formaldehyde in the branching In this way we can explain the fact that when formaldehyde is added to a mixture of propane and oxygen it sensitizes the high temperature ignition, but does not affect the low temperature oxidation below 350° C. There does not appear to be any reason for looking further than the alde- hydes, and in particular formaldehyde, as the essential intermediates leading to branching in the reaction zone approaching the high temperature ignition. The Lower Temperature Ignition Phenomena.-The cool flame phe- nomena are illustrated by the curve for propane air mixtures, due to Newitt and Thornes (Fig. I).Between 400° and 270° C the temperature- pressure curve of ignition is associated with a region extending on the low reaction : RCHO + 0, = RC0,H + 0.274 COMBUSTION OF HYDROCARBONS pressure side of ignition within the bounds of which intense reaction associated with cool flames is observed. In closed vessels the cool flame may be present during the whole reaction, which follows the normal sinusoidal curve of pressure against time, or under other conditions of partial or total pressure, one or more flames may successively traverse the reacting mixture. Under such conditions the passage of each flame is associated with a sharp pressure pulse of very short duration imposed upon the normal pressure time curve.The pressure and temperature limits within which Newitt and Thornes observed one, two or more flames are shown in Fig. (I). Outside the cold flame region the reaction is much slower both at high and low temperature but it still follows the normal sinusoidal pressure-time curve. The temperature limits within which the cool flames occur are much the same for all the higher hydrocarbons studied, namely, about z60° to 400° C. Similar observations are also characteristic of aliphatic ethers, ketones and aldehydes. This in itself would suggest a common origin of the phenomenon and the conclusion is borne out by the fact that the spectra of all cool flames which have been observed lo, 11-acetaldehyde, diethylether, propionaldehyde and hexane- are identical with the fluorescence spectrum of formaldehyde.The fact that cool flames of this specific nature are characteristic of the oxidation of aldehydes and that aldehydes are formed in profusion as common inter- mediates in the oxidation of the paraffins and other ketones in the cool flame zone suggests that their origin is to be sought in some reaction derived from the aldehyde itself. That this reaction is a side reaction and not part of the main course of oxidation is shown by the fact that only one quantum of radiation per 1oS molecules of aldehyde reacted is emitted ; but whatever it is it must be a reaction capable of yielding sufficient energy to provide excited molecules of formaldehyde. The spectrum emitted required a maximum excitation of the order of 95 kcal. It has been suggested by the author that this is to be found in the reaction but it is now clear that there is serious objection to this.In a static system of oxidizing acetaldehyde, Newitt, Baxt and Kelkar la showed that multiple cool flames can follow one another through the mixture, and it is not easy to see how the above reaction could give rise to such periodic phenomena. It would appear that some substance derived from the alde- hyde is alternatively built up to a limiting concentration and exploded, and Neumann has offered evidence, in the case of higher hydrocarbons, that this may be a hydroperoxide or dialkyl peroxide. Recently Bardwell and Hinshelwood,13 studying the cool flame of butanone oxidation, have obtained similar periodic cool flames and observed a fluctuation of per- oxide in the medium corresponding to the passage of each flame.They further found that the induction period preceding the cool flames was greatly reduced by the addition of a small quantity of acetaldehyde, while it is unaffected by formaldehyde. In the scheme of oxidation described above for the high pressure oxidation the aldehyde has been indicated as the source of alkyl radicals in hydrocarbon combustion. Similar processes could occur in acetal- dehyde oxidation itself, e.g. the scheme : RCH2CH0 + 0 = RCHO + H,CO, AH = - 74, 3 CH, lo + o /OOH CH,CHO + 0, --f CH,C 1 0 \OH CH,CHO + 0 -+ CH,CO + OH CH,CHO + OH -+ CH, + CO + H,O CH,CO + CH, + CO lo Emeleus, J . Chem. SOL, 1926, 2948 ; 1929, 1733. l1 Kondrat’ev, 2. Physik, 1930, 63, 322. lZNewitt, Baxt and Kelkar, J Chem.SOC., 1939, 1703. 13 Bardwell and IIinshelwood, Proc. Roy. SOC. A , 1951, 205, 375R. G. FV. NORRISH 275 leads to methyl radicals. radical can have but a transient existence, owing to its instability, i.e., as is indicated by its non-appearance in the methane oxidation, there can be little doubt about its existence at low temperatures, for-hydroperoxides are built up in measurable quantities, in the cold flame region. Thus the aldehyde may be the source of peroxide formation through the alkyl radicals it generates in hydrocarbon combustion, in its own combustion, and in the combustion of ethers and other bodies in which it is found, as an intermediate. We have the possibility of the reaction of the hydrocarbon radical with oxygen taking two courses Now while at high temperature the peroxide CH, + 0, = [CHaOJ + HCHO + OH RCHO + OH RCH,OO RCH, + 0 2 < the peroxide radical coming more into the picture at lower temperatures.Two courses of propagation in hydrocarbon systems are then opened up, namely, OH + RCH, = RCHa + Ha0 ROO + RCH, = ROOH + RCH,. If, therefore, some reaction of the accumulating peroxide, subject to ignition limits, could give rise to excited formaldehyde, as suggested by Harris and Egerton,14 the periodic cool flames in aldehyde hydrocarbon and other similar oxidizing media find ready explanation, but such a conclusion must be subject to the result of experiments designed to measure the spectra emitted by the explosive, decomposition and oxida- tion of hydroperoxides. The facts require that one or both should be identical with the fluorescence of formaldehyde.It is significant that the conclusions of Bardwell and Hinshelwood, that the cool flame of butanone oxidation is only a secondary process and unconnected with the main course of oxidation, are in agreement with the conclusions to be drawn from Townend’s intensity measurements for acetaldehyde.16 It is apparent from the results of Newitt and Thornes, for propane, and Bardwell and Hinshelwood, for butanone, that intermittent or periodic cool flames are associated with a parallel periodicity in the self-heating of the oxidizing medium. Each interval between cool flames is a period of exponential acceleration of the main oxidation reaction, with consequent self-heating. Towards the end of the process the peroxide which has been accumulating reaches its ignition limit and explodes with the emission of light.Thereafter the main oxidation process, if it does not pass to com- plete thermal ignition, suddenly decelerates, the self-heating is arrested and the temperature falls, The whole process of acceleration of the reaction to the next phase of self-heating and associated cool flame is then repeated. This periodic self-heating of the medium may be explained by a fluctu- ation in the branching factor associated with a small periodic variation in the concentration of the higher aldehydes, and peroxides, in accordance with the general principle suggested by Salnikov.16 By hypothesis the aldehyde is subject to two processes of oxidation-ne leading to branching the other not, and mechanisms for these two processes have been postulated above.If the latter, which is to be identified with the oxidative degrada- tion of the higher aldehydes to formaldehyde, increases more rapidly than the former with rise of temperature, the stationary concentration of higher 1* Harris and Egerton, Nature, 1938,412, S30 ; Pvoc. Roy. Sac. A , 1938, 168, I. ’5 Topps and Townend, Trans. Favaduy Soc., 1948.42, 345. 16 Salnikov, Compt. vend. U.R.S.S., 1948. 60, 405.276 COIMBUSTION O F HYDROCARBONS aldehyde will be diminished and the branching factor may fall.* It is known that in the cool flame zone the stationary concentration of aldehyde is higher, and the reaction much more intense than in the region of slow reaction outside.Any process of self-heating can therefore, if of sufficient magnitude, carry the system outside the cool flame region to the region of slow reaction and no self-heating where the branching factor is very small. With the suppression of self-heating the temperature of the system will very rapidly fall to that characteristic of the cool flame region. There will be a short period of induction and acceleration while the aldehyde and peroxide concentration is re-established, followed by self-heating and the next cool flame. The fluctuation of aldehyde pressure during the periodic process need not be great for the branching factor is very sensitive to small changes in the concentration of intermediate. It is understandable that such fluctuations could be within the experimental error of Bardwell and Hinshelwood, though with the slower reaction of Newitt and Thornes they were observable. That the periodic fluctuations in aldehyde concentration are indeed small is in accord with the long period of induction before the first flame, and the short interval between successive flames.The first may be said to represent the time for the branching process to build up the critical aldehyde concentration from zero, while the second represents the time required to make good the small deficiency of aldehyde due to the thermal fluctuation. In cases where there is one continuous cool flame, it is to be concluded that the self-heating is insufficient to take the reaction system outside the cool flame zone. We therefore have a continuous glow emitted by the peroxide which is all the time maintained at its ignition limit.The facts are further consistent with the conclusion that final ignition of the system in the region of cool flames is conditioned by a thermal ignition of the accumulated aldehyde. A striking observation made by Mr. Galvin and the author may be mentioned here : if a mixture consisting of 160 mm. of propane and 160 mm. of oxygen is reacted at 281O C-just inside the cool flame region-the in- duction period of 5 hr. is progressively cut down to zero by admixture of acetaldehyde up to 0.86 yo. The same induction period, of 5 hr., is also reduced to 20 min. by strong irradiation with ultra-violet light. Similar results were obtained by Townend and Chamberlain 17 who were able to induce a lower ignition peninsula in the oxidation of ethane, and to reduce the time lags to ignition in the same region from hours to seconds.The conclusion would seem to be reasonable, that it is the aldehyde whose build-up is accelerated by irradiation and which functions as the essential intermediate. Thus we conclude that the formation of peroxide is secondary to the production of aldehyde. There is nothing, however, in the above conception which would ex- clude the accumulated peroxide as a source of branching at these low tem- peratures. Such a reaction as could be superimposed on the aldehyde mechanism without any material alteration of the kinetics, but since the stability of the hydroperoxide rapidly decreases with rise of temperature the contribution of this reaction to branching will rapidly wane in the same direction. The above reaction may also, through the HO,, be the source of the hydrogen peroxide which Dr.Bailey and the author have found to be * The branching factor will in fact pass through a minimum as the temperature rises, since the formaldehyde, which all observations show to be comparatively stable and not contributing to the branching proceys a t low temperatures of the cool flame zone, begins to function in this capacity as the high temperature ig- nition region is approached (methane oxidation). l8 Bailey and Norrish (not yet published). ROOH + 0, = RO, + HO, Townend and Chamberlain, Proc. Roy. SOG. A , 1936, 154, 95.R. G. W. NORRISH 277 the main peroxide isolated in the products of the combustion of hexane in the cool flame region.In this work we have made an extensive ana- lytical study of the products formed in the cold flame of hexane by a flow method in which the intensity of the cold ffame could be measured. The results are in accord with the " aldehyde " scheme outlined above, modified by the production of derivatives of hydroperoxide or peroxide radical (cyclic ethers, etc.) and consistent with attack on the hexane chain at secondary as well as primary hydrogen atoms. Thus, while all the alde- hydes from caproic aldehyde to formaldehyde have been identified as required by the aldehyde scheme, there is evidence, in addition, of such reactions as CH,CHCH,CH,CH, -+ CH,O + CH,(CH,) ,CHO CH ,OH 00 I '\ which lead to the production of alcohols, and of reactions of cracking, and disproportionation of hydrocarbon radicals which yield considerable quan- tities of olefines.The cold flame itself, at atmospheric pressures, may, under certain circumstances, become associated with a second more intense blue flame. This second flame establishes itself in the products of the first flame as the mixture is made richer in oxygen. Similar two-stage cold flames have been observed by Townend and his co-workers 18 in the oxidation of acetaldehyde and ethers, and by Newitt and Thornes 8 with propane, and it has been shown that both parts emit the spectrum of formaldehyde. We have been able to show that the intensity of emission of the first cold flame increases as the mixture becomes richer in oxygen and that simultaneously the production of aldehydes and peroxide in- creases also.But when the second flame forms the concentrations of aldehydes, peroxides and other condensable products decrease rapidly with increasing intensity of light emission, giving place to carbon monoxide and water and, in addition, a much increased yield of olefines. There seems to be little doubt that the second flame accompanies the combustion of aldehydes and other intermediates formed in the main reactions that proceed parallel with that of the first flame. Its temperature is about 430°C but it gets higher as the oxygen content is increased and may finally give place to the white flame of ignition. It is possible that at these higher temperatures its origin may be found in the direct oxidation of aldehydes by the reaction previously proposed in any case it is a stage which precedes the thermal ignition of the whole system. The results of our work on the products from the cool flame of hexane have led us to the conclusion that both the aldehyde and peroxide mechan- isms are functioning at the same time in this region. Relative Oxidizability of Hydrocarbons .-Finally we may consider the interesting facts concerning the relative rates of oxidation of hydro- carbons as we ascend the homologous series, to which attention was first drawn by Cullis and Hinshelwood.22 The fact that other things being made equal, the rate of oxidation of hydrocarbons increases towards a maximum as we ascend the homologous series can be readily understood when i t is realized that the secondary hydrogen atoms are more reactive than the primary. Thus as we ascend the series from ethane, the ratio of " primary " to I ' secondary " characteristics progressively falls towards RCH,CHO + 0 = RCHO + H,CO, AH = - 74 ; Maccormac and Townend, J . Ckem. Soc., 1940, 143. 2 O Ubbelohde, Proc. Roy. Soc. A , 1936, 152, 355. *1 McDowell and Thomas, J . Chem. Soc., 1949, 2205, 2217. 22 Cullis and Hinshelwood, Faraduy Soc. Discussions, 1947, 2, 117.27s ORGANIC PEROXIDES zero and since the primary hydrogen is less easily attackable than the secondary, we shall pass from a rate solely defined by primary hydro- gens in ethane, to one solely defined by secondary hydrogens in the case of an infinitely long hydrocarbon. Thus we should expect the oxidiza- bility per carbon atom to increase to a maximum as the length of the molecule increases. Department of Physical Chemistry, University of Cambridge, Free School Lane, Cambridge.