REACTIONS OF ATOMIC OXYGEN WITH MOLECULAR OXYGEN BY RICHARD A. OGG, JR. AND WILLIAM T. SUTPHEN Chemistry Department, Stanford University, Stanford, California Received 5th February, 1954 Oxygen gas highly enriched in 018 isotope has been used for rate studies of the exchange reaction The relative proportions were followed by mass spectrometer analysis. Addel i: otopically normal ozone is found to catalyze attainment of this equilibrium, the reaction being very rapid in comparison with 018 exchange between 0 2 and 0 3 . This precludes the direct 0 atom transfer reaction between 0 2 and O3 as being appreciably involved. The only alternative appears to be the ultra-rapid exchange reactions exemplified by 0 + 0018 -+ 00 -t 0 1 8 Vagaries in the dissociation of ozone make a precise determination of the rate constant very difficult, but an activation energy of practically zero and a " normal " frequency factor somewhat less than 1012 mole-1 cm3 sec-1 appear indicated.Some uses of the system as a reagent for detecting production of 0 atoms are discussed. The present study suggests that at low pressures the concentration of 0 atoms exceeds that cor- responding to equilibrium with 0 3 and 0 2 , an effect attributed to energy chains. 016016 -k 0 1 8 0 1 8 J 2016018. The term " very fast " or " ultrarapid " reaction is employed in the sense of being characterized by a very large rafe constant. An ultrarapid second-order reaction of suitable type may be reduced to a pseudo first-order reaction of con- venient time scale by the device of drastic reduction of one reactant concentration, employing an accurately known chemical equilibrium to control and measure this concentration.The classic gas phase example is the ultrarapid reaction H 1- H-H (ortho) H-H (para) + H, whose rate is best studied 1 by utilizing the extraordinarily small but very accurately known equilibrium concentration of atomic hydrogen in hydrogen gas at moderate pressures and temperatures. It appeared to the authors that an even more rapid reaction should be the analogous identical exchange between atomic and molecular oxygen. The moderate activation energy of the hydrogen exchange corresponds to a short-range potential barrier-there is no chemically stable H3 species. However, the well-known stability of the 0 3 species indicates that the short-range interactions between 0 and 0 2 should be in the sense of a potential well.One would thus expect a truly negligible activation energy for the oxygen identical exchange reaction-the rate constant for this process should approach the limit for ultrarapid reactions. Since 016 and 0 1 8 isotopes have nuclear spin quantum number zero, the exist- ence of both orfho and para species for 016016 and 018018 is precluded. Prac- tical study of the exchange reaction in question can be achieved only by its effect on approach to isotopic equilibrium between molecular oxygen species. The rarity of 017 makes the only usable equilibrium the following That the rate of attainment of this equilibrium is completely negligible under ordinary conditions 2 precludes interference by formation of so-called " 0 4 ".If 47 016016 4- 0 1 8 0 1 8 2016018.48 REACTIONS OF ATOMIC OXYGEN this latter species is really formed at low temperatures in liquid oxygen, it cannot have a ring structure. While the 0 - 0 2 exchange reaction in principle could be studied by the thcrmal establishment of the above isotopic equilibrium, the very great dissociation energy of 0 2 would demand inconveniently high temperatures. An equilibrium which a t room temperature offers oxygen atom concentration of suitable magnitude is that involving ozone, It was appreciated that there was possibility of interference from the conceivable direct exchange reaction between 0 2 and 0 3 , exemplified by ooo:k + 00 2- 00 -t- o*oo, where the asterisk indicates the isotope labelling.As will appear from the expcri- mental studies, this direct reaction is fortunately so slow in comparison with the atomic processes as to offer negligible interference. A preliminary account of these studies has already appeared.2 0 3 0 2 -1- 0. EXPERIMENTAL The samples of enriched gaseous oxygen containing some 25-30 % 018 isotope were made available through the courtesy of Prof. A. 0. Nier of the University of Minnesota. The enrichment had been accomplished by thermal diffusion, starting with approximately 1.5 % 018 content. Since at this level the abundance of 018018 is very low, and only moderate temperatures are encountered in the diffusion columns, the enrichment should be predominantly in the 016018 species. This proves to be the case, typical samples displaying the following approximate values of the abundance ratios as found by the mass spectrometer, 34/32 = 1, 34/36 = 100.Hence initially, (34/32) (34/36) = 100. Isotopic equilibrium was established in some samples to check the reliability of the mass spectrometer analysis. The methods employed were the passage of an elcctric discharge from a Tesla coil, or irradiation with a mercury resonance lamp of a sample confined in a silica glass vessel containing liquid mercury at room temperature. By either method the final experimental value of (34/32) (34/36) was found to be in accurate agreement with the theoretically expected value of 4.00. The samples were storcd for use in Pyrex glass containers provided with a hollow-bore high-vacuum stopcock.Connection to the vacuum system was through suitable ground joints. Ozone was prepared from isotopically ordinary oxygen (which was dried by slow dis- tillation from the liquid), by passage through an ozonizer incorporated into the vacuum system. The effiuent, containing approximately 2 % 0 3 , was condensed with liquid nitrogen directly into the reaction vessel. Since the ozonizer proved quite consistent in yield, the desired quantity of ozone could be obtained by metering the volume of oxygen used. This was achievcd by use of a special burette to measure the volume of liquid oxygen evaporated. The mixture of oxygen and ozone in thc vessel was kept cooled with liquid nitrogen, and pumped until no further oxygen was evaporated. (The volatility of 0 3 is negligible at this temperature.) The final determination of the exact amount of ozone in the sample was made by determination of the amount of oxygen produced by pyrolysis.In most experiments the extent of ozone decomposition during contact with the enriched oxygen was small, but it could be very accurately assayed by consideration of the 018/016 ratio in the oxygen after contact. Any reduction of this ratio below the initial figure was due to dilution with 016 isotope produced by ozone decomposition. As described below, the remaining ozone was quantitatively yyrolyzed at the termination of the experiment. The only inert gas so far employed extensively is carbon dioxide. Samples were dried by careful evaporation from the solid, and stored in vessels attached through stopcocks to the vacuum line.From manometric measurement of the pressure, and the known volume of the storage bulbs, an accurately measurcd quantity of carbon dioxide could be condensed in the reaction vessel by cooling with liquid nitrogen. The typical reaction vessel was constructed from a 100cm3 I<jcldahl flask of Pyrex glass. It was provided with a relatively long slender ncck terminating in a hollow-bore stopcock and ground joint. The volume of the flask, relative to that of the various storage bulbs, was carefully determined by a series of gas expansions and manometricRICHARD A . OGG, J R . AND WILLIAM T . SUTPHEN 49 measurements of pressure. Hence, the introduction of a measured gas sample into the calibrated flask allowed computation of the corresponding concentration.Description of a typical experiment follows. The approximately known ozone sample was introduced as described above. With the deoxygenated ozone still condensed at liquid nitrogen temperature, the measured sample of carbon dioxide was introduced, of course condensing. The vessel was then opened to the enriched oxygen reservoir, and a sample was introduced, the quantity being determined by initial and final mano- metric pressure ineasurements on the storage bulb of known volume. The nitrogen coolant was then replaced by an acetone 4- solid carbon dioxide bath, and the vessel was allowed to remain at the corresponding temperature for a period depending upon the size of the carbon dioxide sample in the flask. (For the highest pressures used this approached as much as an hour.) The object of this procedure was to vaporize the ozone and as much as possible of the carbon dioxide, allowing them to mix thoroughly with the oxygen by diffusion.At this low temperature the reaction was negligibly slow. After the mixing procedure the flask was rapidly prewarmed and was then immersed in a water bath maintained at constant temperature. After a measured time interval the vessel was withdrawn from the water bath and rapidly cooled with liquid nitrogen, thus “stopping” the progress of the reaction. The oxygen gas was then transferred from the flask to the sample tube used for the mass Spectrometer analysis. In order to conserve the invaluable enriched oxygen supply a special pumping piocedure was em- ployed to facilitate as nearly complete transfer as possible.The portion of the vacuum line to which the reaction vessel was attached was of very small volume, and was provided with a side arm filled with silica gel. After opening the stopcock of the nitrogen cooled reaction flask this side arm was cooled with liquid nitrogen. This had the result of ad- sorbing the oxygen gas practically quantitatively on the silica gel. The reaction vessel stopcock was then closed, and that of the sample tube (attached to the same section of the line) was opcned. By warming the silica gel the oxygen was desorbed and allowed to expand into the relatively large sample tube. Repeated mass spectrometer analyses demonstrated conclusively that repeated sorption and desorption with the silica gel did not alter the 36, 34 and 32 proportions in the enriched oxygen sample.The trace of oxygen gas remaining in the reaction vessel was removed by direct exhaustion with the vacuum pump. Complete removal of oxygen was necessary to avoid errors in the estimation of the 018/016 ratio in the ozone. Repeated outgassing of the condensed ozone and carbon dioxide assured iiegligible retention of oxygen. The reaction flask was then warmed to a temperature of several hunhed degrees Centigrade and was maintained at this condition until pyrolysis of the ozone proved complete. The vessel was again cooled with liquid nitrogen and the quantity of oxygen gas produced by the pyrolysis was determined by manometric measurement of pressure. (The coolant nitrogen was maintained at a rixcd level on the slender neck to minimize uncertainty to the temperature of the gas sample.) This oxygen gas was then transferred to another sample tube for mass spectrometer analysis.As an additional precaution a sample of the remaining carbon dioxide was also taken for mass spectrometer analysis. However, in no case was the 018/016 ratio in the carbon dioxide found to be greater than the natural value. This is proof that the role played by the carbon dioxide is purely physical. Such possible reactions as 0 1 8 -I- c02 y2 0 -I- OCOlS are slow in comparison with the other steps. For the few photochemical experiments reported, the reaction vessel was fashioned from a 100 cm3 flask of fused pure silica, with a silica to Pyrex graded seal. This vessel was also used for some of the thermal runs.Such intercomparisons as have been made indicate that results from the Pyrex and silica vessels are indistinguishable. This constitutes the principal experimental indication of the essentially homogeneous character of the reaction. As indicated in the previous communication 2 the earlier analyses were carried out with mass spectrometers at the Uiiiversity of California. A few analyses were also ob- tained at the laboratories of the Consolidated Engineering Corporation, Pasadena, Cali- fornia. The Stanford Research Institute has recently acquired a mas sspectrometcr manufactured by the above corporation, and this instrument has been used for all recent work. In the analysis, emphasis is placed on the 34/32 and 34/36 ratios. The abundance of the 0 1 7 species is too low to necessitate any corrections.It was found desirable to determine the relative magnitude of mass number 28, as this gave a measure of possible nitrogen contamination, indicative of any air leaks.50 REACTIONS OF ATOMIC OXYGEN RESULTS To minimize the wordiness of presenting the data, the approach toward equilibrium in the reaction will be referred to by the admittcdly rather crude term “scrambling ”. The term “ isotope exchange ” will (unless specifically defined otherwise) refer to approach toward equilibrium in the set of reactions exemplified by 016016 3- 018018 zz 2016018 016018 -/- 016016016 .( 016016 + 018016016. The earlier studies on the ozone catalysis of “ scrambling ” were carried out without addition of foreign gas, and constituted the principal basis of the previous report.2 As will be noted in the discussion, these studies are not satisfactory for quantitative evaluation of the rate constants for the important reaction steps.They did, however, serve to demon- strate the feasibility of the extended study, in that they strikingly demonstrated the fact that the scrambling was rapid in comparison with the isotope exchange. ‘Table 1 gives in considerable detail the data relevant to a typical example of these earlier experi- ments. In presenting this and the remainder of the analytical results, the symbol 0i* under “ sample analyzed ” refers to the enriched oxygen gas sample introduced initially into the vessel, and finally separated by fractional distillation from the residual ozone, as described above.The symbol 02(03) refers to the oxygen gas sample produced by pyrolysis of the ozone, as described above. The “ initial analysis ” given for this refers to a sampIe of oxygen gas used for the preparation of the ozone. TABLE 1 No added foreign gas ; temp. 0” C ; initial partial pressure 0 3 , 50 mm Hg ; time interval, 15 min ; initial partial pressure 0 2 , 15 mm Hg initial final sample analyzed 34/32 36/34 yo 01s 34/32 36/34 % 0 1 8 0,9002 0.0115 31.0 0.4145 0.0685 17-2 0 2 ( 0 3 ) 000399 - 0.20 04065 040071 0.33 0 2 * The analytical results in table 1 were obtained with the Consolidated Engineering Nier Isotope Ratio instrument, and are regarded as especially trustworthy. Consideration of the above data exemplifies the conclusion indicated in the previous report -namely that the isotope exchange is slow as compared with the scrambling.It will be observed that the final 0 1 8 abundance in 0 3 has been but slightly increased above the initial value. Had isotopic equilibrium between 0 2 and 0 3 been reached, the abundance of 0 1 8 would have been approximately 5.2 atomic percent. On the other hand, in the 0 2 * the initial value of (34/32) (34/36) is 78, whereas the uncorrected final value is 6.0. It must be noted that some decomposition of the 0 3 has occurred, as indicated by the decrease of 0 1 8 from 31.0 to 17.2 %. Since the ozone is practically pure 016, this means dilution with the species 016016, i.e. 32. When decomposition of this is extensive the detailed correction is rather complicated.However, a first approximation is simply to subtract the corre- sponding jitzal excess 32, and to estimate the (34/32) (34/36) product is the hypothetical residual oxygen. This corrected product is 10.9. Comparison with the limiting value of 4.0 indicates the extensive scrambling. A further comment on the evaluation of the analytical results may be made. Simple dilution with isotopically normal oxygen, i.e. “ 32 ”, unaccompanied by true scrambling, can of couise reduce the product (34/32) (34/36) to any value, including the range less than 4.0. However, this would leave the ratio 34/36 unaffected. Since the enriched oxygen used is deficient in the species 36, a qualitatively reliable index of scrambling is the increase in the ratio 36/34. Such an increase is seen in table 1.A Considerable body of such experiments as that in table 1 indicated always the same sense of result, i.e. that ozone caused scrambling, accompanied by a relatively slow isotope exchange. In the attcmpt to permit more nearly quantitative evaluation of the rate phenomena, these experiments were modified by addition of carbon dioxide at moderate pressures and by change of temperature. A selection of results fiom such experiments appears in table 2. The analytical results in this case were obtained with a WestinghouseRICHARD A. OGG, J R . A N D WILLIAM T. SUTPHEN 51 mass spectrometcr, in which the isotope ratios are found from peak heights. There was a considerable background at mass 36, and in consequence the vitally important ratio 36/34 is solnewhat uncertain, especially for small values.The enriched oxygen used in these experiments had initial 34/32 and 36/34 of 0.925 and 0.0234 respectively. The initial product (34/32) (34/36) was thus 39.6. The final value of this product quoted in table 2 has been corrected for oxygen dilution as discussed before. In several cases the ozone decomposition was so slight that this correction is of no significance. TABLE 2 final 0 2 * final Oz(O3) cxpt. O" temp time C O ~ P Of,'' O/n' g$ ___- mi' mm HE! mm Hg mmHg mm Hg 34/32 36/34 ( ~ ~ & ~ 34/32 36/34 %Ols 30 0 15 0 19.6 523 51.2 28 0 15 200 18.5 38.2 35.0 29 0 15 500 20.2 51-7 51-7 31 0 16 207 18.4 41-1 39.7 34 0 10 203 5.4 26.5 24.8 35 0 10 211 5.3 26.4 25.3 36 0 10 213 5.4 14.4 14.6 39 $27 2 204 19.7 35.1 33.3 38** -22 4 217 5.3 34.6 33.5 40** -17 8 158 26.0 17.4 17.3 0382 0.521 0.68 1 0.567 0.408 0.459 0.665 0.450 0.634 0.805 0.1 15 0.0640 0.043 0.063 1 0.0523 0.060 1 0.0678 0.0565 0.0408 0.0360 6.7 1 13.7 23.1 15.3 16.9 15.7 13.0 17.3 21.2 23.6 0.00896 0.01 10 0.02 10 00226 0.008 19 0.009 10 0.00985 0.0096 000986 0.0176 - 0.44 - 0.54 - 1.02 - 1.10 - 0.40 - 0.44 - 0.48 - 047 - 0.48 - 087 ** bath, chilled acetone.While a more dctailed appraisal is given in the later discussion, it may be mentioned here that the data in table 2 present unexpected features. The increase of COz pressure, while causing the expected acceleration of isotope exchange, actually inhibits the scrambling reaction. Also, the temperature coefficient of the reactions is surprisingly low. As a working hypothesis it was proposed that the equilibrium between O3,02 and 0 is seriously disturbed at low pressures, possibly due to an energy chain in the ozone pyrolysis.It was expected that sufficiently large quantities of foreign gas would effectively suppress TABLE 3 44 4-21 15 2.0 46 -150 5 2.0 49 $35 5 2.0 52 $21 15 2.0 53 1-35 10 2.0 54 -1-50 3 2.0 47 0 375 2.8 51 -k97 5 2.0 451 -1-17 15 2.0 482 1-22 15 2.0 503 +21 15 1.0 19-7 25.6 23-7 20.3 33-3 26.5 20.3 27.7 24-6 10.5 29.6 23.3 10.0 24.0 19.7 9.6 25-3 19.8 19.2 29.6 22-1 20.8 30.7 11.1 19.8 30.0 28.1 20.2 24.0 7.3 20.0 34.8 5.4 0-600 0.399 0.7 13 0.386 0.432 0.309 0.387 0.174 0.55 0.15 0.1 3 0.0602 0-0427 0.0207 0,0275 0.0303 0.025 1 0.022 1 0.0461 0.0405 0.0377 0.0341 13.4 20.0 43.0 34.0 3 1.3 32.3 42.0 22.9 21.7 0.00940 - 0.47 0.00853 - 0.43 0.00962 - 0.48 0.00783 - 0.39 0,00798 - 0.40 0.0111 - - 0.0872 0.0222 - 0.0141 - - 0.115 0.0294 - 0.106 0.0280 - - - - 1 Pyrex flask, irradiated with intense red light ; 2 silica flask, irradiated with ultra-violet light ; 3 silica flask, irradiated with ultra-violet light.these vagaries-a continuation of the effect demonstrated in 30, 28 and 29. In conse- quence, a further series of experiments was undertaken, in which the partial pressure of carbon dioxide was made as large as feasible. The corresponding details are given in table 3. The analyses of this set were performed with the new mass spectrometer of the Stanford Research Institute. The enriched oxygen used in expt. 41 to 52 inclusive had initial 34/32 and 36/34 of 0.915 and 0.0185 respectively, and thus an initial product (34/32) (36/32) of 49.5.The oxygen used in expt. 53 and 54 had initial 34/32 and 36/34 of 0.964 and 000978 respectively, and hence an initial product (34/32) (34/36) of 98.6. Included in table 3 are also the52 REACTIONS OF ATOMIC OXYGEN results of three photochemical experiments. The light source in expt, 45 was a high- pressure mercury arc, from which spectroscopic tests showed a high flux in the red region absorbed by ozone. The Pyrex flask served to filter any ultra-violet light absorbable by ozone. The source in expt. 48 and 50 was a low-power mercury resonance lamp, delivering most of its radiant energy at 2537A. This is strongly absorbed by ozone, and it will be noted that considerable photolysis resulted. The product (34/32) (34/36) has been corrected for isotope dilution as before.It will be observed that this number has not been set down for some of the experiments -the inferences drawn from these are essentially qualitative. An explosion caused the loss of the pyrolyzed ozone in expt. 54, but the scrambling results are still significant. As in the previous tables, when no entry appears under 36/34 for the O2(O3), this means that the ratio was too small to measure accurately. Detailcd appraisal of the data in tablc 3 is deferred to the discussion, but it may be noted that the expected " quenching " effect on the scrambling of adding carbon dioxide at high pressure has been observed. This is most strikingly shown by expt. 47 at 0" C.In spite of its protracted duration, the scrambling is negligible as compared to the experiinents (at corresponding temperature) whose results appear in table 2. Expt. 51 at 97" C shows that sufficient temperature elevation makes both scrambling and isotope exchange quite rapid. However, the isotope exchange still lags behind. For the enriched 0 2 the final (34/32) (34/36) is 3.78, indicating practical scrambling equilibrium. However the ratio 34/32 in the O2(O3) is only 0.0872 as compared with 0.174 for the 0 2 * . Exchange equi- librium has not been attained. A point of interest is that the Oz(03) from expt. 51, 48 and 50 all show a value very close to 4.0 for the product (34/32) (34/36). DISCUSSION There is admittedly some question as to whether the reported results corre- spond to a truly homogeneous reaction.Exhaustive tests with packed flasks are planned for future work. However, the exchange of silica and Pyrex vessels caused no noticeable effect on the thermal reaction. In the high-pressure carbon dioxide experiments the gaseous diffusion should be sufficiently slow to ensure a minimum effect of surface reaction, and most emphasis is placed on these. Finally, the photoreaction is certainly essentially homogeneous, and apparently must involve the same significant step effecting scrambling as does the thermal reaction. In subsequent discussion it is provisionally assumed that no surface reactions need be considered. The general state of the experiniental data is still far from satisfactory in allow- ing accurate evaluation of rate constants, but a t least certain qualitative features of the mechanism seem reasonably established.The very considerable body of data shows invariably that scrambling is much more rapid than exchange, and hence that the possible reaction 000 + 0 0 1 8 -+ 00 + 00018, etc. (1) cannot be significantly involved in the scrambling. Rather crude estimates indicate that this calls for the activation energy of (1) to be a t least some 30 kcal/ mole. It appears that the scrambling can be reasonably attributed only to the expected atomic oxygen reaction 0 + 0 0 1 8 --+ 00 + 018, etc. k2 (2) However, the factors determining the stationary concentration of 0 are appar- ently not as simple as was inferred in the previous communication.2 For purposes of discussion, it is proposed that the high-pressure carbon dioxide experiments listed in table 3 correspond most nearly to the establishment of true equilibrium between 0 3 , 0 2 and 0.From the accurately known thermodynamic properties of these three species 3 the equilibrium constantR I C H A R D A . OGG, JR. AND WILLIAM T. SUTPHEN 53 has the value at 25" C of 2.29 x 10-12 atm, or 9.35 x 10-17 molelcm.3 If thermo- dynamic equilibrium is maintained, the reaction is a pseudo first-order process whose rate constant k, = Kk2[03]/[02]. At 25" C, ka = 9.35 x 10-17362[03]/[02] sec-1. The reverse reaction at corresponding equilibrium between O 3 , 0 2 and 0 is a pseudo first-order reaction whose rate constant kb = 4Kk2[03]/[02]. The time for half approach of initially pure 016018 to the equilibrium mixture with 016016 and 018018 is In [(2/ka + kb)].At 25" C this would have the value 1.5 x lOls[02]/k2[03] sec. If the ratio [03]/[02] were chosen as 1.5, which is representative for the data in table 3, the time would be 1015/k2. For this to be of the order of magnitude of 103 sec, the corresponding value of k2 would be 1012 mole-1 cm3 sec-1. Were the value of k 2 to be as large as indicated in the previous paragraph, the corresponding activation energy would be negligible. The change in the rate constant k, with temperature would thus be determined only by the change in K. This in turn would be conditioned by the dissociation energy of 0 3 into 0 2 and 0, namely 25 kcal/mole. With other conditions constant, an increase from 25" C to 50" C should increase k, by a factor of 101.4.From consideration of the rather erratic data in table 3, the increase does not seem to be as great as this. However, the data in table 3 shows a considerably greater temperature coefficient than that in table 2. The most plausible conclusion to be drawn from the above is that the partial pressure of carbon dioxide used in table 3 has improved the approach toward equilibrium between 0 3 , 0 2 and 0, but still not "quenched" completely the strange behaviour shown in the data of table 2. In light of this fact, it can scarcely be said that a very accurate estimate of k2 and its activation energy can be made. However, the most plausible figure indicated by the present data would be a prac- tically temperature independent value somewhat smaller than 1012 mole-1 cm3 sec-1.Such a figure seems reasonable for the exchange reaction anticipated in the discussion. If a figure of the above magnitude be accepted for k2, one is forced to the conclusion that at the low pressures corresponding to table 2 the stationary concen- tration of atomic oxygen is in very considerable excess over that corresponding to equilibrium with 0 2 and 0. A suggested cause of such an cffect is related to the irreversible pyrolysis of ozone. The mechanism in agreement with experimental studies 4 at high oxygen pressures calls for practical equilibrium between 03, 0 2 and 0, and the relatively slow step This reaction is exothermic by some 92 kcal/mole. Equipartition would produce " hot " molecules of 46 kcal/mole. In collision with 0 3 these could cause dissocia- tion of the latter into 0 2 and 0.One is thus faced with the likelihood of a branch- ing energy chain. The concentration of 0 atoms maintained by such a process could well exceed by a large factor the equilibrium value, even for relatively slow pyrolysis. Foreign gas molecules should have the effect of collisionally deactivat- ing the " hot " 0 2 molecules, and hence partly suppressing the abnormal production of 0 atoms. That this suppression is incomplete even at 2 atm of carbon dioxide is suggested by the data in table 3. Some remarks may be made on the photoreaction. The experiment with intense red light gave no significant increase in scrambling rate over the thermal value. The ultra-violet experiments on the other hand, gave a relatively enormous increase.There seems to be no reasonable doubt that the primary photoprocess at 2537A" involves dissociation of 0 3 into 0 2 and 0. On the other hand, it seems most probable that with red light an electronically activated molecule is 2 016018 --j 016018 + 018018 ka 016016 + 018018 -+ 2016018 kb 0 3 + O+ 2 0 2 .54 MECHANICAL METHOD FOR ACTIVATION formed, which may be collisionally deactivated, or may undergo metathetic re- action with 0 3 . Such mechanistic decisions are probably typical of the most useful application of the above studies in their present state, inasmuch as they are not sensitive to the rather considerable numerical uncertainty in the experi- mental value of the rate constant for the 0 atom exchange reaction. The above studies do not yet give very extensive information about the quasi- unimolecular dissociation of 0 3 into 0 2 and 0, except to indicate that the inert gas concentration so far used is still extremely far from producing approach to the asymptotic high pressure limit.It appears that a really precise determination of the rate constant k2 will demand production of 0 atoms by the thermal dissociation of 0 2 itself, i.e., by the experi- ment analogous to the ortho-para hydrogen interconversion.1 This study is con- templated for the future. Its completion should allow more nearly precise evalua- tion of the factors leading to abnormal 0 atom concentration in ozone pyrolysis. 1 Farkas, Orthohydrogen, Parahydrogen and Heuvy Hydrogen (Cambridge University 2 Ogg, Jr. and Sutphen, J. Chem. Physics, 1953, 21, 2078. 3 Latimer, Oxidation States of the Elements (Prentice-Hall, New York, 1952), p. 36. 4 Wulf, J. Amer. Chem. SOC., 1932, 54, 156. Press, London, 1935), p. 66. ADDENDUM Since the submission of our paper to the Discussion, the work in question has been pursued actively. The quantity of new data considerably exceeds that given in the present communication. It is intended to incorporate this, together with a more detailed dis- cussion, in a future separate communication. For thc present, the impact of the new work may be summarized in the statement that it supports the conclusions arrived at in the present paper. Some specific points may be emphasized. It has now been clearly demonstrated that experimental variation of the surface to volume ratio in the reaction vessels is without effect, and hence that the reactions in question are truly homogeneous. The effect of added carbon dioxide up to pressures as high as 16 atm has been followed systematically. The presently indicated inhibitory effect on the " scrambling " reaction has been found to approach an asymptotic limit. At this limit the experimental activation energy agrees satisfactorily with the dissociation energy of 0 3 into 0 2 and 0. The conclusions regarding energy chain quenching appear thoroughly substantiated. The effective rate constant for isotope exchange shows the monotonic dependence on carbon dioxide pressure expected from quasi-unimolecular reaction rate theory.