LOUISM. ARIN AND PETER WARNECK
1514 the partitioning of the various deactivation and reaction processes. The kS9 for the molecules denoted by footnote c in Table I may represent minimum rate constants. Small concentrations of M caused large decreases in intensity of the OHt 9-7 emission band, but the slope of the log I vs. distance plot did not change as markedly (see Figure 1). This probably indicates that M is reacting with one of the initial reactants, probably 0 3 since it is always in excess relative to H. This would , were in effect cause the values of [MI and [ O x ] which used in deriving hQ, to be too large, and the resulting IcS9 to be too small. The k5" reported in Table I for NO is questionable. There was ea. 0.7% NO2 present in the NO as an impurity, and reaction 8 is of compar-
+ NO
+ NO2*
(8) able rate (2.1 X cm3molecule-' sec-')15 to those reported in Table I. Small amounts of NO2 cause large intensity decreases in the OH? 9-7 band because reaction 9 is very eEcient (4.8 f 0.5 X lo-" cm3 mole0 3
H
---+ 0 2
+ NOz-NO
+ OH (V
=
0)
(9)
cule-' sec-1)8 in competing with reaction 1. No ICb9 have been reported in Table I for CO and NHs. The reason for this omission is that the plots of log I vs. distance when M = CO or NHB had smaller slopes than
did the plots for the "blank" runs, and thus no value of kG9 could be determined. It is quite clear why CO is not suitable for study by this technique. The reaction of CO and OH (v = 0) as in eq 10 has been fol-
CO
+ OH +COz* + H
(10)
lowed by a fast-flow esr method a t 300°K by DixonLewis, et aZ.,16their rate constant being 1.9 X cm3 molecule-' sec-I. The H is thus being regenerated down the tube and can then further react with 0 3 t o produce additional OH? (v = 9), causing a lesser slope in the plot. The same process is probably occurring i.e. for "3, NH3
+ OH +NHzOH + H
(11)
It is concluded that the present technique can derive only minimum values for when R/i reacts significantly with OB,and it cannot be used for studying those M which react with OH? to regenerate H. However, the method does seem to be worthwhile for studying reactions of M and OHt which do not suffer from these limitations. (16) L. F. Phillips and H. I. Schiff, J. Chem. Phys., 36, 1609 (1962). (16) G. Dixon-Lewis, W. E. Wilson, and A. A. Westenberg, ibid., 44, 2877 (1966).
Reaction of Ozone with Carbon Monoxide by Louis M. Arin I o n k s , Inc., Watertoum, Massachusetts 02172
and Peter Warneck" Max-Planck-Institut f a r Chemie (Otto-Hahn-Institut), 66 Mainz, West Germany (Received September 27,1971) Publication costs assisted bu Max-Planck-Institut f u r Chemie
Ozone and carbon monoxide were found to react rapidly due to catalysis by a volatile impurity in the CO, but when the impurity was removed, the reaction was too slow for its rate to be measurable.
Recent observations by Seiler and Jungel have demonstrated that carbon monoxide, present in tropospheric air a t a level of 10-7 parts per volume, is rapidly consumed in the stratosphere. The stratospheric oxida~ ~account ~ tion of GO has been predicted t h e ~ r e t i c a l l yon of the reaction with OH radicals, but the observed decay of co concentration with altitude above the tropopause' is much faster than the calculated photoThe Journal of Physical Chemistry, Vol. 76,No. 11, 1972
chemical equilibrium concentrations of OH would allow. Hence, additional CO oxidation processes must be considered. One possibility is the reaction with ozone 0 3
+ co +coz +
w. Seiler and c, Junge,
0 2
Tellus, 21, 447 (196Q).
(2) J. Pressman and P. Warneck, J . Atm. Sci., 27, 155 (1970). (3) E.Hesstvedt, Nature (London), 225, 50 (1970).
(1)
REACTION OF OZONEWITH CARBON MONOXIDE The oxidation of CO by ozone had been studied previously by Harteck and Dondes4 in the temperature range 35-258", where the reaction is most probably stimulated by the thermal decomposition of ozone. We have re-investigated this reaction a t 23" to facilitate the direct observation of reaction 1. The results reported below show that the reaction is catalyzed by small amounts of impurities, but when these are removed, the rate of reaction 1 is very small. The reaction was studied in a 12-1. thermostated Pyrex flask. Two Teflon-stoppercd taps connected the flask to a mercury-free vacuum and gas-handling manifold on onc side, and to an analytical train on the other. Samples takcn for analysis were of about 3 cm3 volume. Ozone was determined by swecping the sample with an inert carrier gas through an aqueous solution of 1% sodium diphenylamine sulfonate acidified with perchloric acid. This solution absorbs ozone com~ l e t e l y . The ~ resulting optical density of the solution is proportional to ozone concentration and was measured at the wavelength 590 mm with a double-beam spectrophotometer. The concentrations of the other gases wore determined by gas chromatography using a thermal conductivity detector and a molecular sieve 5A column. Ozone was generated from dry oxygen in a Siemens ozonizer, stored on silica gel in a trap subjected to Dry Ice temperature, freed from oxygen by pumping, and admitted to the flask through a line containing only Teflon-stoppered taps. Carbon monoxide was the best research gradc available commerically (Matheson). The impurity content was given as Hz < 5 ppm, O2 < 10 ppm, A < 10 ppm, N2 < 250 ppm, COZ < 10 ppm, total hydrocarbon content as methane < 2 ppm. An independent analysis indicated a methane impurity of about 1 ppm. The remaining gases used, hydrogen, oxygen, carbon dioxide and mcthane, were of research grade quality. Initial pressures of ozone were between 3 and 2.5 Torr, the range of CO partial pressures was similar. The behavior of the reaction was unusual in that it occurred rapidly during the process of mixing the reactants, whereas thereafter the gas mixture was stable and the reaction was imperceptible except for the slow decay of ozone due to its thermal decomposition. Typical results are shown in Figure 1 for a run starting with the admixture of 15 Torr of CO to 22 Torr of ozone in the flask. Most experiments were performed in this manner, Le., by adding CO to ozone. The initial fast reaction was often accompanied by a bluish flash when mixing was rapid or by a luminous zone in the neck of the flask when the rate of mixing was kept slow. As expected, the reaction of ozone with carbon monoxide resulted in the production of carbon dioxide and oxygen. The ratio of COz produced vs. CO consumed was, on the average, 0.98 f 0.21 as determined from 22 runs. The oxygen product concentration could not be measured with sufficient
1515
RUN No. 1-101 A Partial Reuw of 02mr Eoforo the Addition of Q) 0 Partiil Pressure of OZOM Attu the Addition of CO
w
PARTIAL PRESSURE
OF CO
5i OO
10
PARTIAL 1000 PRESSURE OF COI moo0
100
TIME I minuter )
Figure 1. Change of reactant and product partial pressures with time after initial rapid reaction period. I
"
'
~
'
'
'
I
'
-
'
'
0
8 FLASH OCCURReO
ADOITION 0
UPON
OF W
NO R A S H
'VESSEL
I
I
I
,
I
1
,
I
.
OCCURRED UPON
ADOITON of W TR
I
I
Y H Y
I
,
-
1
accuracy because it was falsified by the decomposition of ozone in the gas chromatograph inlet line. The consumption of ozone sometimes exceeded that of CO, namely, when the reactants mere allowed to mix rapidly, and when the ratio of initial reactant concentrations, (CO)o/(O&, was greater than unity. Under these conditions the consumption of ozone was essentially complete, as Figure 2 indicates. This behavior may be due in part to the thermal decomposition of ozone induced by the release of heat from the reaction and/or to the dynamical heating effect upon gas entry discussed by Gray and Disregarding those runs in which the consumption of ozone approaches loo%, one finds as the average of the ratio of ozone consumed to COz produced: 1.01 -I: 0.20, as determined from 15 (4) P. Harteck and 8. Dondes, J . Chem. Phys., 26, 1734 (1957). (5) H. H. Bovee and L. J. Robinson, Anal. Chem., 33, 1115 (1961). (6) D. H. Fine, P. Gray, and R. Mackinven, 12th International Symposium on Combustion, Combustion Institute, Pittsburgh, Pa., 1969, p 545. (7) H. Goodman and P. Gray, Trans. Faraday Soe., 66, 2772 (1970). The Journal of Physical Chemistry, Vol. 76, No. 11, 1978
1516 runs. These results show that the stoichiometry of the reaction is essentially as written in reaction 1. However, the temporal behavior is inconsistent with that of a simple bimolecular rate law. The unusual course of the reaction as demonstrated in Figure 1 may result from two mechanisms: (a) the products oxygen and/or carbon dioxide inhibit the reaction so that it is stopped after a certain amount of product is formed; or (b) the reaction is catalyzed by an impurity and ceases when the impurity is used up. A third possibility, the attainmcrit of thermodynamic equilibrium, can be eliminated from the discussion, because the equilibrium values of reaction 1 lie far to the right and either ozone, or CO, or both should have been consumed completely. I n most experiments, however, the reaction was incomplete. To determine the effects of oxygen and COz upon the reaction, mixtures of these gases with ozone were prepared and CO was added. No inhibition of the reaction was observed a t partial pressures of oxygen up to 10 Torr and COz up the 15 Torr. When carbon monoxide was added to the stable gas mixture resulting from the incomplete reaction of ozone with CO, the reaction proceeded further, consuming most of the remaining 0 3 . When more ozone was added to such a stable reaction mixture, no further reaction occurred. These rcsults clearly indicate (a) that the reaction is not inhibited by 0 2 or COz, (b) that it is caused by an impurity in the carbon monoxide, and (c) that conceivable impurities introduced with the ozone have no significant effect. Since, as a precaution, carbon monoxide had always been introduced to the system via a liquid nitrogencooled spiral trap, iron carbonyl was not considered a likely impurity. This was verified when CO was taken from a glass container sealed by the manufacturer; the reaction proceeded as usual. Attempts to purify the carbon monoxide by low-temperature adsorption-desorption cycles using silica gel, charcoal, or molecular sieve as the sorbant were unsuccessful. However carbon monoxide subjected t o the reaction with ozone so that ozone was consumed entirely could be freed from most of the product oxygen and COZ by adsorption on charcoal. Carbon monoxide purified in this manner, when admixed to ozone, produced no reaction. These observations therefore indicate that the impurity responsible for the reaction has a volatility similar to that of CO, precluding an effective separation by the applied trapping and sorption techniques; that the impurity present in the original CO is essentially consumed in the reaction; and that the CO thus purified is not again contaminated in the gas-handling manifold. The volatile character of the impurity pointed to either hydrogen or methane. The effect of these gases was tested by adding them in amounts of about 1 Torr to the stable gas mixture produced by reacting incomThe Journal of Physical Chemistry, Vol. 76, No. 11, 297.9
LOUISM. ARINAND PETERWARNECK pletely CO with ozone. Hydrogen was found to have no effect, but methane caused a slow further oxidation of CO. After 15 h r the consumption of ozone was essentially complete, but the amount of COz formation from CO was only 15% of the ozone destroyed. The results presented here show that the initial rapid reaction of ozone with CO is caused by an impurity in the carbon monoxide, that the impurity is consumed in the reaction, and that the carbon monoxide thus purified does not react noticeably with ozone. An upper limit for the rate of this uncatalyzed reaction 1 can be obtained from the decrease of ozone concentration in Figure 1 which being due mainly to the thermal 0 8 decomposition is about twice the upper limit change of COZ. The value of the rate coefficient thus derived is ICl 5 4 X cma/molecule sec. Clearly, this rate is so low that reaction 1 in its uncatalyzed mode can play no role in the oxidation of CO in the atmosphere. The catalyzed reaction, on the other hand, is so fast under the employed experimental conditions that a chain reaction is indicated. Morris and Kiki,* in commenting on our results, have suggested iron carbonyl as the impurity responsible for the catalyzed reaction. However, this impurity is not expected to survive the low-temperature adsorption-desorption cycle to which CO was subjected without success of purification. The impurity implicated by the described experiments is methane. Since CO is generated commercially from methyl formate, with distillation at liquid nitrogen temperature being the final purification step, methane is an expected impurity. Both methane and carbonyl can, in principle, initiate a fast reaction chain by their interaction with o z ~ n eor~ atomic ~ ~ oxygen generated from the thermal decomposition of 0 3 . The present knowledge about the mechanisms of such reactions is largely speculative, and their detailed discussion is not warranted here. However, the present results do make apparent the importance of minor impurities in the reaction system under discussion. Quite possibly, also the experiments by Harteck and Dondes4 were affected in this way.
Acknowledgment. This work was carried out while the authors were with GCA Corporation, Bedford, Mass. It was supported jointly by the Kational Air Pollution Control Administration, Department of Health, Education and Welfare, and the Coordinating Research Council, Inc. We are grateful to Dr. J. Bufalini for supplying us with a hydrocarbon analysis of the employed carbon monoxide. (8) E. D. Morris and H. Niki, J . Amer. Chem. Soc., 92, 5741 (1970). (9) F. J. Dillemuth, D. R. Skidmore, and C. C. Schubert, J . Phys. Chem., 64, 1496 (1960), in investigating the methane-ozone reaction have observed severe explosions when ozone was in excess, thus demonstrating that chain reactions can be triggered in this system.
