J. CHEM. SOC. FARADAY TRANS., 1994, 90(9), 1205-1210 1205 Investigation into the Kinetics and Mechanism of the Reaction of NO, with CH,O, at 298 K and 2.5 Torr: A Potential Source of OH in the Night-time Troposphere? Peter Biggs, Carlos E. Canosa-Mas, Jean-Marc Fracheboud, Dudley E. Shallcross and Richard P. Wayne* Physical Chemistry Laboratory, South Parks Road, Oxford, UK OX1 3QZ The kinetics of the reaction CH,O, + NO, +CH,O + NO, + 0, (1) have been studied at 298 K and at pressures between 2 and 3 Torr of helium using the discharge-flow technique combined with laser-induced fluorescence detection of the methoxyl radical and measurements of the NO, radical using visible absorption. Numerical modelling of the concentration-time profile of CH,O with or without NO, and NO as a titrant has allowed us to show that CH,O is a product of reaction (1) and to derive a rate constant k, = (1.0 f0.6) x cm3 molecule-' s-', at 95% confidence limits.A comparison of the reactivities of NO, and NO, towards the species R, RO and RO,, where R = H or CH,, is given. The implication of reaction (1) in the possible production of OH in the atmosphere at night is discussed. Peroxy radicals (e.g.CH302)are generated during the day in at T= 298 K and at 2.5 Torr pressure, with helium as the the troposphere by photo-oxidation of organic compounds.' carrier gas. Pseudo-first-order conditions were maintained The radicals may persist into the night and it has been sug- throughout, with NO, as the excess reactant. Methylperoxy gested by Platt et al.' that NO, can react with peroxy rad- radicals were generated by allowing oxygen and methane to icals flow through the inner sliding injector and fluorine atoms to CH,O, + NO, +CH,O + NO, + 0, (1) flow through the outer sliding injector.The methyl radicals, thus taking over the day-time r81e of NO which converts produced by the reaction of fluorine atoms with methane, peroxy radicals to alkoxy radicals (e.g. CH,O) by the reac- added to 0, in the reaction tion CH, + 0, + M -,CH302+ M (7) CH,O, + NO +CH,O + NO, (2) forming methyl peroxy radicals in the outer sliding injector. Where NO and NO, levels are low, peroxy radicals will react However, in this low-pressure system it was never possible to with HO, convert all the methyl radicals to CH302via reaction (7).As a result, some CH,O was produced from the reaction ofCH,O, + H02 +CH302H+ 02 (3) residual methyl radicals with CH,O, and will also undergo self reaction CH302+ CH, +2CH,O (8)CH,O, + CH302+products (4) Further, one channel of the self reaction of CH,02 albeit slowly. The alkoxy species (CH,O) formed in reaction (1) can lead to the production of HO, by its reaction with CH,O, + CH3024 2CH,O + 0, (9) 02 9 also contributed to the CH,O signal. The CH302 emerging CH,O + 0, +HO, + HCHO (5) from the tip of the injector therefore always contained some CH,O. In order to monitor the [CH,O,], NO was added to Hall et aL3 and Mellouki et aL4 have shown that the major the system just before the LIF cell (and necessarily in a fixed channel of the reaction between HO, and NO, position with respect to the detector), effecting the conversion HO, + NO, +OH + NO, + 02 (6) of CH302to CH,O is sufficiently fast to effect the conversion of H02 to OH.The CH302+ NO CH,O + NO2 (2)hydroxyl radical is the dominant day-time oxidant but has generally been assumed hitherto to have no sources at night; which was reflected in an increase in the LIF signal from the NO3-mediated oxidation process makes it possible for CH,O. The ratio of [CH,O,] :[CH,O] in the flow could be OH to be formed. determined in this way. Typical (initial) concentrations used In this paper, we describe a study employing a low-pressure in this system were: [F] = 5 x lo', molecule crn-,, discharge-flow apparatus to investigate the kinetics and [CH,] = (2-5) x lo', molecule ern-,, [O,] = (2-5) x 10l6 mechanism of reaction (1).The reaction is of considerable molecule ern-,, [NO,], = 0.4-3.5 x lo', molecule ern-,; potential importance since CH,O, is the most abundant initial concentrations of organic radicals were (0.5-5) x lo', peroxy species in the atmosphere.' molecule ern-,. Oxygen (BOC, 99%) was passed through a This study represents the first direct discharge-flow experi- trap containing molecular sieve 4A (BDH) to remove water. ment in which radicals were generated separately and in Nitric oxide (Messer Griesheim, 99?40)was purified by repeat- which the CH,O product of reaction (1)was identified explic- ed freeze-pumpthaw cycles at 77 K. All other materials and itly.purification procedures used were identical with those described in the preceding paper.6 Experimental Results and Discussion The apparatus used was identical to that described in the Kinetic Data and Their Analysispreceding paper.6 A flow method was employed; CH,O was detected by laser-induced fluorescence (LIF) and NO, by In the study of reaction (l), four signals (identified as A, B, C multi-path optical absorption. Experiments were performed and D) were measured for the methoxyl radical at each J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 contact time. As previously explained (see Experimental section), we always observed a methoxyl signal when gener- ating CH302. Signal A is the total fluorescence from meth- oxyl produced from reaction (8) and (9) in the absence of any added NO or NO,.On addition of NO,, the methoxyl fluo- rescence always increased and the LIF response was termed signal B. This increase was due to formation of CH,O in -mCreaction (1). Addition of NO to the reaction system altered the LIF signals because of the conversion of CH302 to .-CI) v) CH,O in reaction (2). Signal A became signal C, and signal B became signal D. A typical set of signals A, B, C and D is shown in Fig. 1, and the caption specifies the chemical com- ponents used to generate each signal. In principle, of course, addition of NO increases the fluorescence, because CH302 is -converted to CH,O in reaction (2). However, since NO reacts chart motion with NO, and CH,O reacts with both NO and NO, Fig. 1 Chart trace showing an example of the signals recorded in the experiments on the CH,O, + NO, system.The full lines are the NO, + NO +2N0, (10) LIF signals from CH,O: A, for the system CH,O/CH,O, ; B, for the system CH,O/CH,O, + NO, ; C, for the system CH,O/CH,O, CH,O + NO +products (11) + NO; D, for the system CH,O/CH,O, + NO, + NO. The dashed CH,O + NO, -,products (12) line shows the concentrations of NO, for [NO,], = 3.3 x lo', mol-ecule cm-, and ACNO,] = 0.6 x lo1, molecule ern-,. the response of LIF intensity to added NO is complex. All four signals were used in the numerical modelling of these the reaction. The occurrence of reaction (1 3) makes analytical chemical systems (see later) to determine [CH,O,]/[CH,O] treatment of the experimental data impossible, even though for a range of contact times and NO, concentrations. One [NO,] 9 [CH,O,].Other competing and secondary reac- crucial feature should be noted about the typical signals tions, such as the second-order loss of CH,O,, further com- shown in Fig. 1: the [CH,O] increases from signal A to plicate the analysis. For these reasons, the rate constant k,signal B, i.e. on addition of NO, to CH302. This result was derived by numerical modelling using the programdemonstrates explicitly for the first time that NO, reacts with FACSIMILE7 to integrate the differential equations describ- CH302 to produce CH,O in reaction (1) ing the kinetics of the set of reactions displayed in Table 1.CH302+ NO, -,CH,O + NO, + 0, (1) Initial examination of the results of such modelling showed clearly that varying k, and k13 had a complementary effect It is evident that CH302 is regenerated in the step on the predicted [CH,O] concentration. This being the case, it would thus be impossible to extract the two rate constants CH,O + NO, +CH,02 + NO, (13) independently from our experimental data for the CH,O con- As discussed in the preceding paper,6 the efficiency of regen- centrations. However, altering kl, in the model affected k, in eration of CH,O in steps (13) and (1) is 0.6to 0.7. This result such a way that the ratio r = k,/k13 remained constant. The argues against a major channel for reaction (1) leading to approach we adopted for determining k, was therefore to use products such as CH,O, especially as wall losses also remove the value determined in other independent experiments6 for radicals.The effects of adding NO when NO, is present (the k13 in combination with the ratio r obtained here. D signals) are also compatible with virtually all reactions of We always observed some consumption of NO, (ANO,,CH,O, with NO, yielding CH,O. However, we recognise typically 10-20%) on addition of CH,O,, which was obvi- that competing channels might make a minor contribution to ously due to reaction (1) and, to a lesser extent, reaction (13). Table 1 Rate parameters used in the numerical model (see text) reaction k/cm3 molecule-' s-' reaction ref. ~~ CH,O, + NO, -+ CH,O + NO, + 0, CH,O, + NO -+ CH,O + NO, CH,O, + CH,O, +products CH, + 0,+ M-+CH,O, + M CH, + CH,O, -+ CH,O + CH,O CH,O, + CH,O, +CH,O + CH,O + 0, NO, + NO +2N0, CH,O + NO -+ products CH,O + NO, -+ products CH,O + NO, -+ CH,O, + NO, F + CH,+CH, + HF CH, + NO, -+ CH,O + NO, CH, + CH, -+ C,H, (0.6-1.6) x lo-', 7.6 x lo-', (0.9-2.0) x lo-', (3.0-6.0) x lo-" 3.0 x lo-" 4.0 x lo-', 2.3 x lo-" (1.6-3.6) x lo-', 8.0 x lo-" 4.0 x lo-'' 4.0 x lo-" 3.0 x 10-13 1.1 x 10-13 1 2 3 7 8 9 10 11 12 13 14 15 16 this work 9 11 10" 13 11 9 16 8 6 9 6 12 CH, -+ products CH, + CH,O -+ products CH,O + CH,O, +products CH,O + CH,O +products CH,O -+ products CH,O, +products 1 s-l 4.0 x lo-" 3.0 x 1.0 x lo-" 15-50 s-' 3-5 s-l 17 18 19 20 21 22 this work 14 15 this work this work b ~~ ~~_____~~____ a The bath gas in the sliding injector, where the CH,O, was generated, was an approximate 1 : 1 He-0, mixture; the exact rate constant used to model a particular experiment depended on this bath-gas ratio.The geometric mean value was used to estimate this rate constant. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 The values of k, and k13 are therefore related to the mea- sured A[N03], so that measurements of this quantity might be used, in conjunction with the values of r already deter- mined, to fix independently the values of the two rate con-stants within the set of experiments described here. However, we regarded the precision of the measurement for ACNO,] to be insufficient for this purpose. Instead, the determinations of ACNO,] were used as a further check that the model was correct.Details of the Numerical Modelling A standardised protocol was adopted for the numerical evaluation of the signals. The procedure for fitting A and B signals is conventional and straightforward, and can be run to give a good match, as judged by the squared deviations, to points at all the contact times simultaneously. For the C and D signals, the integrations must be run for each point indi- vidually, because of the chemical processes initiated by the injection of NO. Our analysis starts with the A signal, which is determined by the chemistry of the organic radicals alone. The LIF calibration factors (checked against the initial F-atom concentration) and the rate coefficients for homo- geneous and heterogeneous reactions of CH,O and CH,O, are adjusted, always within limits imposed by published data.An important check at this point is that the simulated C signal also fits the experimental data well, because a good match indicates that [CH ,O2]/[CH30] has been correctly determined. From this stage, the parameters adjusted so far are regarded as fixed. Next, NO,-initiated chemistry is intro- duced. The only adjustable parameters are k, and k,, in fitting to the experimental points of curve B. Finally, the D signals are calculated as a check, although poorer matches with the experimental data have been regarded as acceptable because of the considerable complexity, and critical behav- iour, of the chemistry when CH,O, CH302, NO, NO, and NO, are all present together.Fig. 2 shows a sample set of experimental data (squares and triangles) and the associated modelled CH,O concentration (solid curves). Fig. 2(a)shows signals A and C, while Fig. 2(b) shows signals B and D. For clarity of presentation, points for signals C and D are given for only three positions of the sliding injector; the modelled curves for these signals show the anticipated changes in [CH,O] from the point of injection of NO. It can be seen, for the values of k, and k,, chosen, that all four signals behave as expected, both as a function of contact time and with respect to the addition of NO3 and NO. The model required as inputs the initial [O,] in the sliding injector and the initial [NO,] in the main flow.Allowance was made for the dilution of the radicals that emerged from the sliding injector. In our implementation of the kinetic modelling, we also started with [Fl0 and [CH,], and regard- ed the various rate coefficients for the initial reactions leading to CH30 and CH,O, as fitting parameters. This procedure really serves to define [CH,O] and [CH30,] as needed to match the A and C signals. We regard this method as rather more satisfactory than the simpler method of varying [CH,O], and [CH3O2IO, because we are forced to adopt realistic values at the outset for the concentration of F atoms and for the rate coefficients. Table 1 shows which rate con- stants, other than k, and k13, were varied, and the limits between which they were altered in the fitting procedures.It should be emphasised again that, once these rate coefficients had been used to fit the A and C signals, they remained unal- tered while fitting the B and D signals. Table 2 shows the relevant experimental conditions and the ratio I (= k1/k13) obtained for each run. In total, model- ling was performed on 18 experiments, from which r was 1207 I I II time/s I I I N ‘j 05 c--. a I 0, I 1 I I 0.02 0.04 0.06 c time/s Fig. 2 An example of experimental data and the associated model- led CH,O concentrations (a)(0)experimental signal A; (A) experi-mental signal C; (b) (m) experimental signal B; (A)experimental signal D. The solid lines are the corresponding modelled values.determined to be (0.43 & 0.09); errors are quoted as one stan- dard deviation. With this ratio r and the value determined6 for k13 = (2.3 & 0.7) x lo-’, cm3 molecule-’ s-l, a value for k, = (1.0 0.6)x 10-l2 cm3 molecule- s-was calculated; the quoted error, at the 95% confidence limits, is the com- bined error in the ratio r and the rate constant kI3. Using these values for k, and k13, it was possible to reproduce the observed consumption in [NO,] (ACNO,]), within the preci- sion of the measurements, for all experiments. Calibration Factor Fcp, We consider here the calibration factor Fca, relating the observed CH,O signal on the chart recorder to an absolute concentration. Although Fca,was one of the parameters gen- erated by the numerical fitting procedures, we also performed direct calibrations from time to time.To generate CH,O quantitatively,’ the reaction of methyl nitrite with fluorine atoms CH,ONO + F +CH,O + FNO (23) was employed. The LIF signal was a linear function of added [CH,ONO] so long as F atoms remained in substantial excess. In the linear regime, [CH,O] produced is proportion- al to [CH30NO) added,” and could be used to deduce Fcal after taking into account the experimentally measured losses of CH,O in the processes CH,O 4 products (21) CH,O + F -+ products (24) J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 2 Summary of modelling results pressure/Torr 2.54 2.54 2.54 2.73 2.56 2.98 3.19 2.81 2.41 2.87 2.37 2.35 3.51 3.41 3.46 3.42 2.58 2.61 [N0,]o/10'3 molecule cm-, 1.67 0.97 2.07 1.5 1.12 1.26 0.6 1.78 1.8 1.61 0.85 0.41 2.71 1.44 1.34 0.82 2.78 3.42 A[N03]/10'3 molecule cm-, 0.12 0.14 0.2 0.27 0.08 0.2 0.1 0.27 0.39 0.15 0.09 0.03 0.43 0.32 0.31 0.19 0.27 0.36 ratio, r 0.44 0.56 0.58 0.44 0.4 0.35 0.5 0.39 0.35 0.35 0.35 0.30 0.50 0.50 0.30 0.35 0.50 0.50 k,/10-'2 cm3 molecule-' s-' 1.01 1.29 1.33 1.01 0.92 0.81 1.15 0.90 0.81 0.81 0.81 0.69 1.15 1.15 0.69 0.81 1.15 1.15 r = 0.43 f0.09 and k, = (0.98 f0.20) x lo-', cm3 molecule-' s-', taking6 k,, = 2.3 x lo-', standard deviation. The values of ACNO,] are those determined experimentally.an3molecule-' s-'. Quoted errors are one Comparison with Results of Other Studies Only limited comparisons with data obtained by other workers are possible. Crowley et ~1.'~used a modulation technique to infer a rate constant for the reaction of CH,O, with NO, of 2.3 x lo-'' cm3 molecule-' s-', which is rather higher than, but not inconsistent with, the value we are suggesting here. However, after it was discovered that there was some non-photochemical production of NO, from HNO, , this figure was revised downwards by a factor of two to three.lg More recent experiments by the same group," in which CD302 and NO, were followed mass spectrometri- cally in a discharge-flow experiment, have suggested a rate coefficient of 2 x lo-', cm3 molecule-' s-', some ten times lower than the earlier published value.This value, obtained in a way that parallels (but does not exactly copy) our tech- nique, appears to us to be incompatible with the results that we have obtained by following the CH,O radical. We suggest that the key to the discrepancy may originate in the regener- ation of CH302 from CH,O, a process whose occurrence we have substantiated.6 Moortgat et d2'were unable to find evidence for this reaction. Our belief is that this failure to observe the reaction was a consequence of the manner in which the CD,O was generated (an upstream reaction between CD, and NO,). It seems quite likely that all the alkoxy radical could be consumed by reaction with NO,, so that none would be available for reaction with NO,.Since the alkoxy radical itself cannot be detected in the experiments of Moortgat et ~l.,~'these experiments cannot directly show, either, that CD30 is a product of the reaction of CD,O, with NO,. Regeneration of the peroxy radical could then make it appear that the reaction is much slower than it really is. We understand' ' that reanalysis using the scheme suggested by us here renders the earlier data of Moortgat et al." entirely compatible with our own measurements. Butkovskaya et ~1.~'have recently carried out discharge-flow experiments with LIF detection of CH,O. These experiments are almost exactly of the same type as our own, but seem to be limited to the CH,O + NO3 system.As far as we can judge, the data are very similar to ours.6 Analysis of the data using our sug- gested mechanism has enabled these workers to extract a rate coefficient for reaction (24in the same way that we did from the system in which CH,O was the initial reactant.6 The pr-- liminary value they obtain is 7 x lo-', cm3 molecule-' s-*. Our estimated value of k,, obtained in the same way is 1 x cm3 molecule-' s-', which is entirely compatible with the new result. However, as we explain in the previous paper,6 we believe that the rate coefficient is better measured in the way described here, using CH,O, as the starting radical. As it happens, the final result is numerically identical, but we regard both the interpretation and the calculation as being more firmly based in the second method adopted by us.Our conclusion is, then, that there is no real conflict in the experimental data obtained in a variety of systems, but only in the interpretation of what the results mean. Our systematic study of all the radical species involved (CH,, CH,O, CH,O, and NO,) in these complex chemical systems has given us a new understanding of the processes involved and allowed us to advance our interpretation of the results. It is interesting next to attempt a comparison of the reacti- vities of NO, and NO, towards the species H and CH,, OH and CH,O, and HO, and CH302. Table 3 summarises the rate parameters for these systems. Hydrogen atoms appear to react four times faster than methyl radicals do with either NO, and NO,.It would seem then that the mechanism is the same for the reaction of the species H and CH, with NO, and NO,, i.e. straightforward abstraction forming the alkoxy species, and it is tempting to invoke a simple steric argument to explain the difference in rate coefficient between H and Table 3 Summary of rate constants at room temperature for the reactions of R, RO and RO, (R reaction H + NO, -+ OH + NO, CH, + NO, -+ CH,O + NO, H + NO, + OH + NO CH, + NO, -,CH,O + NO OH + NO, -+ products CH,O + NO, -P products OH + NO, +products CH,O + NO, + products HO, + NO, -,products CH,02+ NO, + products HO, + NO, -,products CH,O, + NO, -+ products = H and CH,) with NO, and NO, k(298 K) /cm3 molecule -' s -' ref.1.1 x 10-'O 23 3.5 x lo-" 6 1.25 x lo-'' 24 2.3 x lo-'' 8 2.6 x lo-'' 4 2.3 x lo-', 6 2.4 x lo-'' lo" 2.1 x lo-" 8" 3.6 x lo-', 4 1.0 x this work 4.7 x lo-', 9" 6.5 x lo-', 9" " Denotes the high-pressure limit. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 CH, . These systems also suggest that there is little difference between the reactivity of NO, and NO, towards alkyl species. The similarity between NO, and NO, is less apparent when we consider their reactions with OH and CH,O. For the NO, case, formation of an adduct (HONO, and CH,ONO,) is the dominant reaction pathway, whereas for NO,, formation of a peroxy species (H02 and CH,O,) is the major pathway.Analogous 0-atom transfer with NO, would be endothermic. It may well be the case that the NO, reac-tions proceed through some energised complex,6 but that the stability of the peroxynitrate formed (HOONO, or CH,00N02) is insufficient for this channel to compete with decomposition. The reactions of CH302 and HO, with NO, lead to rela- tively unstable peroxy addition compounds. For reaction with NO,, the main channels yield three products, one of which is the oxy radical (CH,O or OH). One observation that can be offered about the chemical reactions of NO, with respect to the peroxy radicals, CH302 and HO, ,is that NO, appears almost to behave like the adduct NO-0,, and the processes lead to the same oxy radicals as are obtained with NO, although the rate constants for the processes involving NO, are somewhat smaller.Our preliminary indirect experiments,' with acylperoxy radicals, R * CO * 0, , suggest exactly the same type of behaviour. With the oxy radicals, CH,O and OH, the NO, radical acts, instead, as an oxidant, in a manner completely different from NO or NO,. Atmospberic Implications A conclusion that might be drawn from the considerations of the last paragraph is that NO, can behave in the night-time atmosphere simultaneously like OH and like NO behave during the day. The radical can abstract hydrogen atoms from many reactants and initiate oxidation processes, in the same manner as OH. The resulting organic radicals then add oxygen to form peroxy radicals, with which NO, interacts again, this time as though it were NO.In a quantitative discussion of night-time peroxy-radical chemistry, Platt et a/., point out the importance of NO,- initiated oxidation. These workers invoke the reaction between CH302 and NO, as a step in a route that could yield significant concentrations of OH radicals in the tropo- sphere at night. The central reactions of the scheme as it involves CH,O and CH302 are CH302+ CH302--* CH,O + CH,O + 0,; 30% (9) CH,O + 0, -+ HO, + HCHO (5) HO, + NO, + OH + NO, + 0,; 80% (6) CH,O, + NO, +CH,O + NO, + 0, (1) Reactions (9) and (6) are particular channels in which more than one branch is known, and the percentages indicated are fairly well established for these channels.In particular, the remaining 70% of the self reaction of CH,O, does not lead to any radicals, so that the reaction is generally terminating. If the reaction between CH,O, and NO, is both rapid and leads virtually exclusively to the products shown in channel (l),then the possibility arises that CH302 can be intercepted by NO, and the subsequent steps generate OH through process (6), again involving NO,. Platt et al., calculate that [OH] could approach 10' molecule ern-,, and that the rate of its production could be only one order of magnitude less by night than it is by day. This conclusion depends on the rate coeficient for reaction (1). The rate of channel (1) as a 1209 source of HO, ,and hence as a source of OH, will exceed that of channel (9) when ~NO3l/[cH302la 2kdk1 For" k, = 1.1 x lop1, cm3 molecule-' s-' and'8 k, = 2.3 x lo-', cm3 molecule-' s-', the inequality becomes [NO,]/[CH,O,] 2 0.1.This condition appears to have been met during the measurement campaigns in Brittany in 1989 but not in 1988.26With the lowest values of k, compatible with our measurements (say 0.5 x lo-', cm3 molecule-' s-'j, the corresponding limiting ratio of concentrations would be about 0.4, a condition possibly met only at the highest NO, encountered. However, exceptionally high [CH,O,] concentrations were inferred from the measure-ments. Concentrations of NO, also seemed abnormally low. Similar observations have been made recently at Schauins- land in the Black Forest.27 In these latter measurements, there is a clear anti-correlation between concentrations of NO, and RO, radicals, a result strongly suggestive of an interaction between the two radical species.Taking a 'normal' night-time concentration of NO, of lo9 molecule an-,, the concentration of CH302 would only have to be less than about 3 x lo9 molecule cm-, for reaction (1) to dominate over reaction (9), even using the lower limit of our rate constant. This limitation seems hardly likely to be restrictive. The essential condition for reaction (1) to be the major source of OH therefore is likely to be met in continen- tal and especially in urban areas. For reaction (1) to be a significant source of OH at night is another matter, because it is now necessary that [CH,02] is relatively high also.The maximum rate of OH production through reactions (1) and (6), taking k, = 1.5 x lo-', cm3 molecule-' s-', and a branching ratio4 of 80% for channel (6), is 0.8 x k, x [CH,O,][NO,]. For [NO,] = lo9 molecule cmP3 and [CH,O,] = lo7 molecule cm-,, our results suggest that this rate would be less than 1.2 x lo4 molecule cmP3 s-l, which is more than 100 times lower than the day-time rate of OH production. The recent measurements from the Black Forest,' give [NO,] = 8 x lo7 molecule cm-, when [RO,] = 9 x lo* cm3 molecule-' s-l (at 0300 CET). For these conditions, the maximum production rate of OH is thus about 9 x lo4 molecule an-,s-l if all RO, radicals possess a reactivity similar to CH,O, .Although this production rate is still less than 10% of that during the day, it seems that the possibility of the reaction playing some r61e in 'normal' atmospheric chemistry at night should not be discounted, and it clearly could be important when anomalous concen- tration conditions are present such as those found in the Brit- tany campaign. We would like to express our gratitude to the NERC (grant GR3/7359) for support for this project, and to the CEC and the Institute Frangais du Petrole, under whose auspices various parts of this and related work were carried out. We thank Richenda Connell for preparing the methyl nitrite samples used in this work. D.E.S. would like to thank the SERC for a research studentship during the tenure of which this work was conducted.References 1 P. D. Lightfoot, R. A. Cox, J. N. Crowley, M. Destriau, G. D. Hayman, M. E. Jenkin, G. K. Moortgat and F. Zabel, Atmos. Enuiron., A, 1992,26, 1805. 2 U. Platt, G. Le Bras, G. Poulet, J. P. Burrows and G. K. Moor-tgat, Nature (London),1990,348, 147. 3 I. W. Hall, R. P. Wayne, R. A. Cox, M. E. Jenkin and G. D. Hayman, J. Phys. Chem., 1988,92,5049. 1210 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 4 A. Mellouki, G. Le Bras and G. Poulet, J. Phys. Chem., 1988,92, 2229. 18 J. N. Crowley, J. P. Burrows, G. K. Moortgat, G. Poulet and G. Le Bras, Znt. J. Chem. Kinet., 1990,22,673. 5 6 7 8 9 10 A. R. Ravishankara, Annu. Rev. Phys. Chem., 1988,39,367. P.Biggs, C. E. Canosa-Mas, J-M. Fracheboud, D. E. Shallcross and R. P. Wayne, J. Chem. SOC., Faraday Trans,, 1994,90,1197. A. R. Curtis and W. P. Sweetenham, FACSIMILE, AERE- R12805, Harwell Laboratory, Oxfordshire, UK, 1988. P. Biggs, C. E. Canosa-Mas, J-M. Fracheboud, A. D. Parr, D. E. Shallcross, R. P. Wayne and F. Caralp, J. Chem. SOC., Faraday Trans., 1993,89,4163. R. Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr and J. Troe, J. Phys. Chem. Ref: Data, 1989,18,881. W. B. DeMore, S. P. Sander, D. M. Golden, M. J. Molina, R. F. Hampson, M. J. Kurylo, C. J. Howard and A. R. Ravishankara, Chemical Kinetics and Photochemical Data for use in Strato- spheric Modeling. Evaluation Number 10, JPL Publication 92-20, 1992. 19 20 21 22 23 24 J. P. Burrows, J.Crowley and D. Maric, EUROTRAC Annual Report 1991, Part 8 (LACTOZ),EUROTRAC ISS, Garmisch- Partenkirchen, 1992, p. 81. G. K. Moortgat, J. Crowley, F. Helleis, 0. Hone, W. Raber and C. Zahn, EUROTRAC Annual Report 1992, Part 8 (LACTOZ), EUROTRAC ISS, Garmisch-Partenkirchen, 1993, p. 149. J. N. Crowley, persoKa1 communication N. Butkovskaya, V. Daele, D. Johnstone, G. Laverdet, G. Poulet and G. Le Bras, EUROTRAC Annual Report 1992, Part 8 (LACTOZ), EUROTRAC ISS, Garmisch-Partenkirchen, 1993, p. 127. R. B. Boodaghians, C. E. Canosa-Mas, P. J. Carpenter and R. P. Wayne, J. Chem. SOC., Faraday Trans., 1988,84,931. M. A. A. Clyne and P. B. Monkhouse, J. Chem. Soc., Faraday Trans. 2, 1977,73,298. 11 12 13 F. G. Simon, W. Schneider and G. K. Moortgat, Znt. J. Chem. Kinet., 1990,22, 791. N. L. Arthur, J. Chem. SOC., Faraday Trans., 1986,82,331. W. Tsang and R. F. Hampson, J. Phys. Chem. Ref: Data, 1986, 15, 1087. 25 26 27 P. S. Owen, D. Phil Thesis, University of Oxford, 1994. T. Brauers, H. P. Dorn and U. Platt, in Physico-chemical Behav- iour of Atmospheric Pollutants, ed. G. Restelli and G. Angelletti, Kluwer, Dordrecht, 1990. D. Mihelcic, D. Klemp, P. Musgen, H. W. Patz and A. Volz- 14 J. Heicklen, Adv. Photochem., 1988, 14, 177. Thomas, J. Atmos. Chem., 1993,16,313. 15 U. Meier, H. H. Grootheer, G. Riekert and Th. Just, Ber. Bun- 16 senges. Phys. Chem., 1985,89,325. M. J. Frost and I. W. M. Smith, J. Chem. Soc., Faraday Trans., 1990,86 1757. 17 P. Biggs, C. E. Canosa-Mas, J-M. Fracheboud, D. E. Shallcross and R. P. Wayne, in preparation. Paper 3/06980J; Received 23rd November, 1993