Mass Spectrometric Studies of Atom ReactionsPart 4.-Kinetics of 0 3 Formation in a Stream of Electrically Discharged 0 2BY A. MATHIAS AND H. I. SCHIFFUpper Atmosphere Chemistry Group, McGill University, Montreal, CanadaReceived 10th January, 1964A mass spectrometer was used to study the 0 3 concentration as a function of time after O2was partially dissociated by a microwave discharge. Reproducible results were obtained onlywhen the 0 2 was passed through traps cooled in liquid air, which suggested catalytic decompositionby hydrogenous impurities. Even with purified 0 2 , the 0 3 concentration reached maximum ratherthan steady-state values which would have been expected if only the reactions,were operative. The observation that more 0 3 could be destroyed than the number of 0 atomspresent suggested the occurrence of the reaction,(3)The 0 and 0: concentrations were determined by titration techniques. The ratio of [O;]/[O]was found to increase with pressure and the kinetics were consistent with the mechanism suggested.Values of the rate constant ratios were found to be k3lk2 = 1.7 and k2/k1 = 1 .2 ~ 1019 molecules,in agreement with the results of 0 3 decomposition studies. The inclusion of reaction (3) there-fore provides an explanation for the spread in the reported values of kl and permits reconciliationof the work on discharged 0 2 and on 0 3 decomposition.0; 4- 0 3 + 0 2 -k 0 2 4-0.Perhaps the most important chemical reactions which occur in the stratosphereare those which determine its ozone content.Aside from photolytic processes,these are primarilyO + ~ ~ + M - + O J + M (1)0+03+202. (2)The rate constants for these reactions are therefore of great importance in chemicalaeronomy. Unfortunately they are not known with any unanimity.Laboratory studies of these reactions are frequently performed with flow systemscontaining small percentages of atoms produced from electrically discharged 0 2 .The homogeneous recombination rates in such systems are usually equated to2k1[0][02][M]. However, such studies have produced a large spread in the reportedvalues of kl.1 These values are also appreciably lower than those which are deducedfrom studies of the thermal and photolytic decomposition of 03.2, 3A value for k:! has recently been obtained 4 from measurements of the rate ofdisappearance of O3 added to a stream of discharged 0 2 .Although it agreedreasonably well with the results of one study on the thermal decomposition of 0 3 , sit was considerably higher than the results of other such investigations.6 An evenhigher value 7 was obtained from the reaction of 0 3 with 0 atoms produced by thetitration reaction of " active " nitrogen with NO. However, the titration reactionalso produces 8 9 9 excited N2 molecules which are capable of decomposing 03.3A. MATHIAS AND H. I . SCHIFF 39The present paper reports an attempt to resolve these differences by an inde-pendent method. A mass spectrometric study was made of the 0 3 concentrationin a stream of discharged 02. If only reactions (1) and (2) were operative, a steadystate concentration would be established :from which the ratio of rate constants could be readily evaluated This relationshipwas, however, not found to be valid, which showed the occurrence of other processes.EXPERIMENTALThe mass spectrometer used was identical to the one described earlier4 except thatrhenium filaments were used throughout and the sampling hole was 15 p wide and 20 11deep.Linde U.S.P. grade 02, containing less than 0.1 % N2, was used directly since samplespurified by liquefaction gave identical results. The oxygen was passed through a 100 cmlength of quartz tubing 10 mm int. diam., over the mass spectrometer sampling " leak "and thence to the pump. The mass flow rate was controlled by a fine needle valve andmeasured with a calibrated capillary flow meter.The pressure in the system was variedat constant mass flow rate by throttling a valve located between the mass spectrometerleak and the pump. The pressure was measured just beyond the mass spectrometer leakeither with a Consolidated micromanometer or with a McLeod gauge. Although therewas a considerable pressure drop down this long length of relatively narrow tubing, theresults of these experiments indicated that the 0 3 concentration adjusted to the local pressurein the system so that the pressure at the leak was the significant parameter.The oxygen was dissociated to the extent of a few % in a cylindrical microwave cavitypowered by a Microthem unit. This cavity could be slid along the quartz tubing to alterthe distance (and hence the time) between the discharge and the mass spectrometer leak.The mass spectrometer was calibrated for 0 3 by introducing part of the effluent gasfrom a laboratory ozonizer into the flow system and measuring the mass 32 and 48 peakheights.Simultaneously, a known volume of the same gas was diverted through anaqueous 2 % KI solution, and its 0 3 content determined by conventional thiosulphatetitration.RESULTSExperiments with this arrangement gave results for the dependence of 0 3 con-centration on time which are typified by the plots of the open circles shown in fig. 1.Curves of this sort were irreproducible but all had the general characteristic of arise to a relatively flat portion followed by a further rapid increase at longer reactiontimes.However, when traps cooled by liquid air were inserted directly before thequartz tube, reproducible curves similar to the one through the closed circles wereobtained .A family of such curves at different pressures is shown in fig. 2. It will be seenthat the 0 3 concentrations reach maximum rather than steady-state values and thatthe maxima become sharper as the pressure increases.The dependence of the maximum 0 3 concentration on pressure is shown in fig. 3.The lines are drawn through two sets of experiments performed with different flowtubes and microwave generators. The linear pressure dependence is also at variancewith the quadratic relationship for the steady-state concentration based on.reactions(1) and (2). These results strongly suggest the presence of other reactive species.To test for the existence of such species, the apparatus was modified in the follow-ing manner. A section of 10 mm int. diam. Pyrex tubing, containing an inletthrough which other reactants could be added to the gas stream, was inserted be-tween the mass spectrometer leak and the quartz tube. This inlet, similar to thos40 MASS SPECTROMETRIC STUDIES OF ATOM REACTIONSdescribed previously,4~ 10 was located 25 cm from the mass spectrometer leak.Excess NO2 was added through this inlet to determine the 0-atom concentration inthe discharged 0 2 stream.4 Alternately, a relatively large flow of pure 0 3 wasintroduced through the inlet to determine the maximum amount which could bedestroyed by the discharged 0 2 .The amount of 0 3 which was decomposed wasI I I II I I01 0 2 03 0t , secFIG. 1 . 4 3 concentration as a function of time : 0, values obtained without traps cooled to 77°K ;e, values obtained with traps cooled to 77°K.always found to be greater than the 0-atom concentration, thus providing directevidence for the presence of some other species capable of decomposing 0 3 , mostlikely excited molecules, 03 :o;+ o,-+o,+ o,+ 0. (3)The concentration of excited molecules can be derived from the relation[0~]=2.(A[03]-A[NO2]) where A[NO2] and A[O3] are the amounts of NO2 and O3respectively which can be destroyed by the discharged oxygen stream. The inclusionof 3 in the expression results from reaction (3) being followed by (1) which consumesan additional O3 molecule.A number of experiments were performed to determine the concentration of0 atoms and 0: molecules at reaction times which produced the maxima in the[03]-time curves.The distance between the discharge and mass spectrometer leakwhich gave the maximum O3 concentration was first determined and the value of[031max noted. The discharge was then moved back 25 cm so that the [031maxposition now occurred at the inlet. The values of [O] and [O:] were then deter-mined as described in the previous paragraph. The results of these experimentsare shown in table 1A. MATHIAS AND H. I . SCHIFF 410:I 0.2 0.3 0.4 0.5time (sec)FIG. 2.-03 concentrations as a function of time at different pressures.The gas has been passedthrough traps cooled to 77°K.I I 42 MASS SPECTROMETRIC STUDIES OF ATOM REACTIONSTABLE 1 .---CONCENTRATIONS OF REACTANTS CORRESPONDING TO [O3]mm1021 r033rnax 101 rot1torr 10-4 torr 10-4 tom 10-3 tom0.250.620.731 -041-201 -451.851-952.303.201.581.672.766-102-123.647.3 13.8 14.8 16.6219.419.435.712.420.814.512-49.335.527-890-062.1 14.405.404.6513.412.019.918.616.2DISCUSSIONThe behaviour shown by the lower curves in fig. 1 can be understood if it isassumed that water vapour is the impurity removed by the traps. It is well knownthat water vapour is largely dissociated in an electrical discharge. The reaction ofH atoms with 0 3 ,has a rate constant 11 which is at least 103 times larger than the rate constant ofreaction (1).Thus, if one H atom is formed per H20 molecule, then in a dischargedgas stream containing a few % of 0 atoms, (4) will dominate over (1) whenever theH20 content of the original gas exceeds one part in 105. Moreover, the OHradicals formed either in the discharge or as a result of (4) will undergo the rapidreaction, 12Reactions (4) and (5) therefore constitute an H-atom catalyzed equivalent to reaction(2). The overall kinetics will be indistinguishable from (1) and (2) alone, so thatthe rate of 0-atom recombination should remain unaltered.* On the other hand,the 0 3 concentration will be lowered to a pseudo steady-state value given approx-imately by kl[0][02]2/kq[H], which can be associated with the flat portions of thecurves in fig.1. The 0 3 concentration will increase again only after the H atomsare removed by relatively slow three-body processes. Other evidence for the catalyticeffect of H atoms in 0-atom recombination has recently been reported.7If (l), (2) and (3) are the predominant reactions in dried, discharged oxygen, asteady-state 0 3 concentration will not be produced. Instead, a maximum will bereached given byH4-03-)OH+02, (4)O+OH+02+H. (5)klC0lCO2l2C031max = k2[O] + k3[0;]’This expression can be rearranged to the form* A similar argument might be invoked for nitrogeneous impurities which could yield NO inthe discharge. The catalytic 0 3 decomposition could then occur by the counterpart reactionsNO+03 +N0;!+02 (4’) and NO2f-O +N0+02 (5’). However, the rate constant for (4’) is notsignificantly larger than for (2), so that this reaction cannot be significantA .MATHIAS AND H . I . SCHIFF 43Thus, if this mechanism is operative, a plot of the left-hand side of this expressionagainst [O%]/[O] should yield a straight line. Such a plot is shown in fig. 4 for thedata of table 1. Least-mean-square treatment of the data yields the equation[02]2/[03]max = 4-02 x 102+6.98 x 102[0J.,]/[O] (torr).Fig. 3 shows that [03]max is proportional to [02] and independent of [O] or[O:]. Inspection of the expression for [03Imax shows that this result will be con-sistent with the proposed mechanism only if [O*,]/[O] is a linear function of [OZ] ;i.e.- k2 + -- k3[0Y21 = a[O4, where a is some constant. Fig. 5 shows that withinadmittedly large experimental error such a relationship seems to apply fora set of data obtained under similar conditions. However, a probably dependskl kl [OlFIG. 4.-A plot of [0#/[03] against [O,*]/[O] obtained at reaction times corresponding to maximum0 3 concentrations.on the purity of the gas and other discharge parameters, and there is no reason whyit should remain unaltered if discharge conditions are changed. This is probablywhy the data of fig. 3, obtained with two different experimental arrangements, lieon curves with different slopes. The increase of [O?]/[O] with 1021 also explainswhy the maxima in fig. 2 get narrower with increasing pressure, since the rate ofreaction (3) will then increase relative to the rate of reaction (2).The value of k2/kl = 1.2 x 1019 molecules cm-3 is in good agreement with theresults of a careful re-investigation 3 of the photolysis of 0 3 , and in satisfactoryagreement with the revised data from thermal 0 3 decomposition studies.Kaufmanand Kelso 13 have recently reported a value of 7-5 x 10-34 em6 molecule-2 sec-1 forkl from studies in flow systems uncomplicated by the presence of excited 0 2 . Com-bination of this value with the result of the present study yields a value of k2 = 9.0 44 MASS SPECTROMETRIC STUDIES OF ATOM REACTIONS10-15 cm3 molecule-1 sec-1 which is appreciably lower than the value of 2-5 x 10-14reported by Phillips and Schiff.4 However, their result may have been too high,since no particular precautions were taken to dry their oxygen.Also no correc-tions were made for possible reaction with 0:. This may not have been too seriousin their case because the same authors found 14 that the maximum 0 3 decompositionnever exceeded the 0-atom concentration by more than 10 %. This is probablybecause most of their experiments were at lower pressures where the present resultsshow that the [Oj.,]/[O] ratios are small, and probably because the ratio may alsodepend on gas purity.3020- *N - o_!o10I I I1.0 2.0 30[021 ton-FIG. 5.-The dependence on pressure of the [O,*]/[O] ratio obtained at reaction times correspondingto maximum 0 3 concentrations.From fig.4 it may be concluded that k3 = 1.7 k2. Reaction (3) may, however,be composite, since some measurements of the rates of 02 disappearance indicatedthe presence of more than one excited species. Arnold and Ogryzlo 15 have alsoreported that the concentration of 02(lAg) decreases slowly and O.#Zi) more rapidlywhen 0 3 is added to a stream of discharged 0 2 . A similar decrease in 02(1Ci)molecules was reported by Clyne et aZ.,7 although they estimate their concentrationto be less than 1 %. In the present work 0% concentrations amounting to several% were observed. Mass spectrometric studies 16, 17 of discharged 0 2 have indicatedup to 20 % of excited molecules which were tentatively identified as O@Ag). Theexcitation energy of this state is some 2 kcal less than the dissociation energy of0 3 .This is not, however, inconsistent with the deduction that k3 is of the sameorder of magnitude as k2 which is known to have an activation energy between4 and 6 kcal mole-1. Positive identification of the different species and their relativeimportance must await further study. However, the inclusion of reaction (3) doesprovide an explanation for the spread in the reported values of kl and permitsreconciliation of the work on discharged 0 2 and on 0 3 decompositionA . MATHIAS AND H . I . SCHIFF 45Acknowledgements are gratefully made for financial assistance to the U.S.A.F.Cambridge Research Laboratories and to the Defence Research Board of Canada.1 Kaufman and Kelso, Chemical Reactions in the Lower and Upper Atmosphere (Interscience,2 Campbell and Nudelman, AFOSR TN-60-502, 1960.3 Castellano and Schumacher, 2. physik. Chem., 1962, 34, 198.4 Phillips and Schiff, J. Chem. Physics, 1962, 36, 1509.5 Zaslowsky, Urbach, Leighton, Wnuk and Wojtowicz, J. Amer. Chem. SOC., 1960, 82, 2682.6 Benson and Axworthy, Ozone Chemistry and Technology (American Chemical Society, Wash-7 Clyne, Thrush and Wayne, Nature, 1963, 199, 1057.8 Phillips and Schiff, J. Chem. Physics, 1962, 36, 3283.9 Morgan, Phillips and Schiff, Disc. Firaday SOC., 1962, 33, 11 8.10 Morgan and Schiff, Can. J. Chem., 1963,41, 903.11 Phillips and Schiff, J. Chem. Physics, 1962, 37, 1233.12 Del Greco and Kaufman, Disc. Farday SOC. , 1962,33, 128.13 Kaufman and Kelso, Disc. Faraday SOC., this Discussion.14 Phillips and Schiff, J. Chem. Physics, 1962, 37, 924.15 Arnold and Ogryzlo, Abstr. 46th Con$ Chem. Inst. Canada ; Chem. in Can., 1963, 15, 54.16 Foner and Hudson, J. Chem. Physics, 1956,25,602.17 Herron and Schiff, Can. J. Chem., 1958, 36, 1159.New York, 1961), p. 255.ington, 1959), series 21, p. 388