Rate Constant of the Reaction O + 2 O 2 + O 3 +O2 BY FREDERICK KAUFMAN AND JOHN R. KELSO Ballistic Research Laboratories, Aberdeen Proving Ground, Md., U.S.A. Received 1 3th January, 1964 The recombination of 0-atoms was studied in a discharge-flow system for purified oxygen whose mole fraction of total hydrogen was 10-6 or less. The absence of recombination below a pressure of about 3 mm Hg and the very slow recombination at 3-6 mm Hg indicates the presence of meta- stable, energetic species capable of dissociating 0 3 and of generating additional 0-atoms well down- stream of the discharge. Experiments in which 0 3 was added to discharged oxygen provide evidence for the presence of these metastable species. The concurrent but unrelated catalytic recombination of 0 by hydrogen impurities explains the wide range of rate constants and their disagreement with those calculated from the thermal dissociation of 0 3 .The recombination was then studied for 0-atoms produced by the rapid thermal decomposition of 0 3 at a few mm Hg. No indication for the presence of metastable species was observed, and a rate constant kl of 7.5 x 10-34 cm6 molecule-2 sec-1 was obtained, in fair agreement with the ozone data. Information on the homogeneous, three-body recombination of 0 and 0 2 is available from two sources : ozone decomposition studies and direct measurements of 0-atom recombination. The thermal decomposition of ozone was recently studied by conventional manometric methods 1 9 2 in the temperature range 7O-13O0C, and in a shock tube 3 from 420-640°C. One of the manometric studies 1 also in- cluded the re-evaluation of older experimental data.4 Moreover, all available data prior to 1960 were summarized and critically reviewed in an authoritative report 5 on the ozone+oxygen system.These studies provide values of the rate constant k!l for M+03-+0+02+M(-l), from which the desired rate constant k y for the recombination process can be calculated using the equilibrium constant.6 Where M is not 0 2 but 0 3 , Ar, or N2, conversion to k? also requires information on the relative efficiency of these gases in (- 1). Such information is available from the above papers and from a new study of the photolysis of ozone with red light.7 The second approach, i.e., the direct measurement of k y in a low-pressure discharge-flow system under steady-state condition, has given discordant results.The present work was thus prompted by the finding that increasingly rigorous purifi- cation of the oxygen led to decreasing values of k? and an increasing discrepancy with the results of ozone decomposition studies. Its experimental work consists of three parts: (i) development of methods for the quantitative determination of impurities, particularly those containing nitrogen and hydrogen ; (ii) study of 0-atom decays in a discharge-flow system using 0 2 of known and controlled purity ; (iii) study of 0-atom decays in the same flow system but without a discharge. EXPERIMENTAL DISCHARGE-FLOW SYSTEM The apparatus is similar to that described in earlier papers.8-10 The main flow tube is of precision-bore, 2.54f0.003 cm int.d i m . and about 1 m long. At its upstream end a discharge is excited by C.W. microwave radiation from a magnetron, type QK-390 26F. KAUFMAN AND J . R . KELSO 27 (Raytheon Mfg. Co.), at 2450 Mclsec, with up to 800 W power. Between the quartz discharge tube and the flow tube are two mixing inlets, one for diluent gases and the other for ozonized oxygen. Near the upstream end of the flow tube is a multi-perforated loop of Teflon tubing for the rapid admixture of NO or NO2 to the gas stream. A mechanical pump (4501./min) produces linear average flow velocities of up to 6m/sec. All flows are controlled by stainless steel needle valves and measured by the recorded pressure drop in one of various calibrated volumes by means of electrical pressure transducers. 0-atom concentrations are measured along the flow tube by the intensity of the O+NO (air afterglow) light emission at eight equidistant positions by means of a photomultiplier movable along the tube which receives the radiation through two collimating slits in a direction perpendicular to the tube. The amplified photomultiplier output is recorded by a strip-chart recorder.NO2 " titrations " of 0-atoms are carried out by measuring that flow of added NO2 which will reduce the O+ NO light intensity to a small fraction (- 1 %) of its maximum, at a point 30cm downstream of the point of addition. The surface of the flow tube is coated with aqueous H3P04 to decrease surface recombination, and the system is then evacuated to < 10-4 mm Hg with an oil diffusion pump. PURIFICATION OF 0 2 A N D MEASUREMENT OF IMPURITIES A few experiments were run with 0 2 vaporized from liquid oxygen collected from the thermal decomposition of purified KMn04.It was found possible, however, to remove hydrogenous impurities from cylinder oxygen to within the limits of the analytical methods, and cylinder oxygen was then used exclusively. Hydrogenous impurities were removed by passing the cylinder gas at 1 atni pressure through a quartz tube, 1 cm int. diam., 30 cm long, packed with quartz chips, and heated to 1100°C ; then through a column, 90 cm long, packed with zeolite molecular sieve, type 5A; and finally after expansion through the needle valve, at the lower pressure, through two glass traps filled with glass beads and glass wool and immersed in liquid nitrogen.- 8 .6 O1 -4 *2 2co 300 400 0 0 XH x lo6 FIG. 1 .-Calibration plots of 01 E (IoH, 41-4/10) against mole fraction of total added hydrogen. Curve 1 : FoZ = 3.6 cm3 S.T.P./sec ; P = 1.28 mm Hg ; H2 added ; curve 2 : FoZ = 3.6 cm3 S.T.P./sec ; P = 5.24 mm Hg ; H2 added ; curve 3 : Fo2 = 0.9 cm3 S.T.P./sec ; P = 0.58 mm Hg ; H2 (0) or CH4 (A) added. No attempt was made to remove nitrogeneous impurities, but selected cylinders were used whose discharged gas showed very little afterglow. Since N 2 is fairly efficiently con- verted to NO in such discharges,ll the intensity of the residual afterglow when no nitric oxide is added serves as a semiquantitative measure of N-impurities. The mole fraction of N 2 in selected cylinder 0 2 was thus found to range from 1 to 5 x 10-6 which is sufficiently low that the effect of the small amounts of NO on the decay of 0-atoms can be neglected.28 The amount of H-impurities was determined by focusing light from the microwave discharge on the entrance of a grating spectrometer (Leeds and Northrup, 0.75 m, Ebert, Photoelectric, Recording) and recording emitted intensities of the lower members of the Ql branch of the (0,O) band of OH, 2X++2ll, as well as of the atomic transition, at 6158.2 A of oxygen.The intensity ratio, (IOH,Q~-~/IO) = a was found to decrease from 0.28 for un- purified 0 2 to between 0 and 0.01 for purified gas. To relate a to the mole fraction of hydrogen XH, known mixtures were prepared by adding small flows of 1.2 % H 2 in purified 0 2 to larger flows of purified 0 2 .The results are shown in fig. 1 in which a is plotted against XH for several mixtures. Three conclusions can be drawn: (i) a varies linearly with XH for &<5x 10-5 as expected; (ii) the limiting slope for small XH increases some- what with pressure at constant flow ; (iii) a appears to be a single-valued function of the mole fraction of total hydrogen independent of its form. Curve 3 includes some points representing H2 additions and others representing CHq additions. Under the vigorous conditions of oxidation in the discharge, this equivalence is reasonable and suggests com- plete conversion of hydrogenous impurities to H, OH, and H 2 0 . The lower slope of curve 3 is probably due to the lower pressure and lower flow.Since most of the experi- ments reported below were carried out at pressures of 1-6 mm Hg and flows of 1-5-25 c11-13 S.T.P./sec, the observed a of 0.005-0.01 for purified 0 2 corresponds to an XH of about 0.5- 2 x 10-6 (or half that when expressed as X H ~ ) . FORMATION OF 0-ATOMS WITHOUT DISCHARGE The direct dissociation of 0 2 in a flow system requires very high temperatures. Though these could be attained either in special furnaces or on certain hot filaments, a simpler method proved to be successful. The rapid thermal decomposition of ozone at low pressure should be capable of producing enough atomic oxygen under steady-state flow conditions to permit the study of recombination reactions in the absence of interfering metastable discharge products.Therefore, a measured flow of purified 0 2 is passed thfough the annular space of a Siemens-type ozonizer across which an a.c. voltage of 10-12 kV produces 0.4-3 % 0 3 . The ozonized oxygen is then expanded into the low pressure region (1-9 mm Hg) thfough a stainless steel needle valve, and it traverses a quartz tube, 30 cm long, 2.5 cm int. diam., placed in a furnace at a temperature of 800-1 100°C. A Pyrex tube of 60 cm length connects the downstream end of the 0 3 decomposition region to the inlet of the flow tube where all measurements of 0-atoms are carried out. The connecting tube is internally coated with H3P04 in order to inhibit surface recombination. The gas stream enters downstream of the discharge tube to permit experiments in which 0 3 is added to discharged oxygen.In the " thermal " 0-atom experiments, the discharge is off and the rapid thermal decomposition of 0 3 is the only source of atomic oxygen. The slight, further pressure drop between the furnace and flow tube amounts to 0.2-3.5 mm Hg depending on pressure and total flow. At the temperature of the furnace, the residence time of the ozonized oxygen in the hot quartz tube ranged from 10 to 80 msec, and was usually equal to 10-20 half-lives of the thermal decomposition of 0 3 (in the presence of excess 02). The concentration of 0 3 was measured by light absorption measure- ments along the full length of the flow tube. The 0 3 analysis apparatus consists of a mercury lamp (Spectroline Quartz Pencil Lamp), quartz collimating and condensing lenses, light filters to isolate the 2537 A line as described by Kasha 12 or, later, a small grating monochromator (Farrand, Catalogue no.103420), quartz windows, and a photomultiplier, type 1 P-28, using the common amplifying and recording system of the grating spectrometer and the photomultipliers on the flow tube. The total optical path length is 115 cm and the minimum analyzable 0 3 pressure is about 0.5-1 x 10-4 mm Hg. In several experiments, it was shown that as the temperature of the 0 3 decomposition furnace was raised, the ozone concentration in the flow tube remained virtually unchanged until a temperature of about 500°C was reached, then decreased upon further temperature increase so that at and above 750°C, the 0 3 was completely decomposed. At the largest flow rates and at pressures above 5 mm Hg, a small amount of O3(" 5 x 10-4 mm Hg) was found to be present in the flow tube even at the highest furnace temperature. This was undoubtedly due to the recombination of 0 in the flow system.F .KAUFMAN A N D J . R . KELSO 29 When NO was added to such thermally decomposed, ozonized oxygen, a normal emission of the air afterglow was observed whose intensity was proportional to the NO concentra- tion. Small NO2 additions also brought on the O+NO glow, but larger amounts then extinguished the glow in a normal atom " titration ". Only very small concentrations of atomic oxygen could be produced in this manner, usually less than 0.5 p, but quite sufficient for the study of its decay down the tube. The small yield of atomic oxygen from ozone (usually t 2 %) is probably due to its fast surface recombination in the heated quartz tube.RESULTS DECAY OF O-ATOMS FROM DISCHARGED 0 2 Experiments were performed at several pressures and flowrates for purified 0 2 ; for unpurified 0 2 whose total mole fraction XH of hydrogen was determined simul- taneously; for purified 0 2 in the presence of a plug of glass wool between the dis- charge and flow tube and with known additions of H2; and for purified 0 2 with or without addition of 0 3 downstream. In each case, the emitted light intensity was measured at eight equidistant places along a section of the flow tube or 53 cm length. In most cases, the very small amount of NO produced from the N2 impurity in the discharge gave sufficient light intensity.Traces of additional NO were often introduced upstream of the observation section in order to show that the decay was independent of the NO concentration. With that (experimentally verified) as- sumption, the decay should be expressible as d l n I dln[O] k* dx dx 2 ) ' --=--- -- where k* is the effective first-order rate constant for the O-atom decay, u is the average flow velocity, and I the measured light intensity. The dependence of I on [O] by I = 10 [O] [NO] and the constancy of [NO] due to its rapid regeneration from NO2 have been abundantly verified in this pressure range, and together they make dl/dx equal to d[O]/dx. The disappearance of 0 should be attributable to 0 +wall (w) which makes k* = k , + 2k1[02]2 assuming (a) first-order wall recombination ; (b) ozone steady state; (c) neglect of 0 + 0 + 0 2 because of small fraction of 0 present; and (d) neglect of or correction for effects of viscous pressure drop or axial diffusion along the tube.Of these, (a) and ( d ) were shown to be satisfied, and (c) is justified because [02]/[0]>200 in most experiments. The question (b) of the 0 3 steady state is a complicated one, but it can at most lead to a range of a factor of two in the second term of k*. If a steady state is established, [ 0 3 l S s = (kl/k2)[02]2. The ratio kl/k2 is known approximately from work on the thermal decomposition and photolysis of 03. A qualitative comparison of the initial 0- atom concentration, [Ole, the expected [O&, and the change of [O], A[O] = [O]O- [O], will show whether or not steady state was approached. If either [O]O or A[O] are much smaller than the expected [O&, steady state could not obtain, and k" would be given by k* = k,+ k1[02]2.In an intermediate case, k* might increase along the tube as [Oj] builds up and approaches [O3lSs. A typical example of experimental results for highly purified 02(XH< 1 x 10-6) is shown in fig. 2. Logarithmic decays are shown for 5 experiments using a constant flow of about 3.6 cm3 s.t.p./sec, of 0 2 , but varying the pressure in 5 steps from 1.2530 REACTION 0 + 2 0 2 + 0 3 + 0 2 to 5.08 mm Hg by bleeding air directly into the pump. The corresponding flow velocities ranged from 430 cm/sec at the lowest pressure to 100 cm/sec at the highest. The most surprising result in fig.2 and confirmed many times is the total absence of any 0-atom decay at the lowest pressures. Curve 1 (1.25 mm Hg) thus shows a slight rise of I with x, curve 2 is flat at first and then drops slightly, and succeeding curves show a continuing trend of increasing initial rates of decay followed by ac- celerating rates downstream. Thus the initial k* would be slightly negative at 1.25 mrn Hg, zero at 2-18, about 0.3 sec-1 at 2.91, 0.8 sec-1 at 4.11, and 0.9 at 5.08. I I I- ---0 - 0 1 O-- I I I I I I 0 10 20 30 40 50 distance, cm FIG. 2.-Spatial decay of 0 in discharged, purified 02. Fo2 = 3-5 cm3/sec ; X,<2 x 10-6. curve 1 : P = 1-25 mm Hg; o = 430 cmlsec 2: 2-18 245 3 : 2-91 190 4 : 4-11 125 5 : 5.08 100 Moreover, the contribution of the surface recombination must be subtracted to obtain an estimate of kl, since that process is unaffected by the concurrent homo- geneous reactions.The work with thermally produced 0-atoms gave k, = 0.5 to 0.6 sec-1 as shown below. When such a kw is subtracted, kl would be negative below 4 mm Hg and very small at 4 and 5 mm Hg. The simple mechanism for the homogeneous recombination is thus inapplicable. These and other experiments described below suggest the presence in discharged oxygen of energetic, metastable species capable of re-dissociating 0 3 formed in the recombination and of producing additional, small amounts of 0 downstream.F . KAUFMAN A N D J . R. KELSO 31 Typical examples of 0-atom decays using discharged unpurified 0 2 are shown in fig. 3. The conditions of flow and pressure closely correspond to those in fig.2, but hydrogen-containing impurities were not removed. a was measured for all runs and XH can be estimated (using fig. 1) to be about 3 x 10-5. At the lower pressures, the curves are similar to those of purified 02, i.e., there is still hardly any 0-atom decay. At higher pressures, the small amount of hydrogen impurities accelerates the recombination, but the initial k* is still far too small. This oxygen traversed the long column of zeolite, but the 0 2 purification furnace was off and the 0 10 20 30 40 50 distance, cm FIG. 3.-Spatial decay of 0 in discharged, unpurified 0 2 . Fo2 = 3.5 cm3/sec ; XH-30 x 10-6. curve 1 : P = 1.26 mm Hg ; D = 440 cm/sec 2 : 2.27 240 3 : 2.83 200 4: 4.11 130 5 : 5-00 108 two traps just before the discharge were not cooled.It seems probable that meta- stable species are again responsible for the abnormally slow decay, but that hydrogen impurities recombine 0-atoms by a concurrent and independent mechanism. The recombination is then characterized by plots of In [O] against distance which are concave duwnwards or S-shaped. Morgan, Phillips and Schiff 13 successfully used Pyrex glass wool between their discharge and flow tube to quench’vibrationally excited N2 in active nitrogen. Since the metastable species in this work may be vibrationally excited 02, the same remedy was tried here. Fig. 4 shows two 0-atom decays for purified 0 2 at pressures of 2.86 and 4-92mmHg and two corresponding decays in which a mole fraction of 7 x 10-5 of H2 was added upstream of the discharge (XH = 1.4 x 10-4).Curves 1 and 3 (without H2) show that if vibrationally excited 0 2 is efficiently deactivated on the H3P04 coated glass surface then it is not the species responsible since the32 REACTION 0 + 2 0 2 + 0 3 + 0 2 0-atom decay is again abnormally slow. Curves 2 and 4 show a similar accelerating effect to that in fig. 3, but of larger magnitude, since more hydrogen is present. Direct evidence for the decomposition of 0 3 by metastable species other than 0-atoms was obtained by mixing ozonized oxygen with discharged oxygen, measur- ing the 0 +NO light intensity and titrating with N O 2 in the presence and absence of 0 3 , and measuring the total 0 3 in the flow tube in the presence and absence of the discharge. Fig. 5 shows the effect of 0 3 on the spatial variation of 0. The flat I I I I distance, cm FIG. 4.-Spatial decay of 0 in discharged, purified 0 2 with glass wool plug between discharge and flow tube. Fo2 = 3.5 cm3/sec. curve 1 : P = 2.86 mm Hg ; X,< 1 x 10-6 2: same XH = 140 x 10-6 (added H2) 3 : P=4*92mrnHg; xH<1x10-6 4: same X , = 140 x 10-6 (added H2). profile (curve 1) is a normal one for discharged, purified 0 2 , except that 1.4 cm3 S.T.P./sec of the total 0 2 flow of 4.0 cm3/sec traverse the ozonizer which is turned off and thereby bypass the discharge. When the ozonizer is turned on, curve 2 is obtained. The 0-atom concentration now increases along the flow tube and is everywhere larger than in the absence of 0 3 . NO2 titrations without 0 3 gave 3-2 p of 0, but with added 0 3 they showed 8.7 p of 0 to be present. The partial pressure of O3 along the tube with the discharge off was 18-9 p, but it decreased to 8.0 p when the discharge was turned on. This experiment clearly shows not only that the reaction o;+ 0 ~ - ) 0 ~ + 0 ~ + O (3)