Reactions of O(2'0,) and O(2"P,) with Halogenomethanes BY MICHAEL C. ADDISON, ROBERT J. DONOVAN AND JOHN GARRAWAY Department of Chemistry, University of Edinburgh, West Mains Road, Edinburgh EH9 355 Received 20th December, 1978 Product branching ratios for the reaction of 0(2'D2) with the halogenomethanes CF3CI, CF3Br, CF31 and CFzHCl are presented. The dominant channel is shown to be abstraction yielding a halogen oxide. This contrasts with the behaviour observed with hydrocarbons, where insertion into C-H bonds dominates. Quenching of 0(2*D2) to the ground state is also observed with the halo- genomethanes and accounts for "N 30 % of the total removal cross-section. Reaction of 0(2'D2) with CF2HCl leads to the formation of CIO (55 %) and to the elimination of HCl (40 %).The latter process is accompanied by the formation of CF2 and OG13PJ). The reactions of 0(2'D2) are compared with those for O(23PJ), where these are known, and the absolute rate for reaction of O(23PJ) with CFJ is determined as (1.1 & 0.3) x lo-" cm3 molecule-' s-l at 300 K. The results are discussed in terms of the main topological features on the potential surfaces involved. Reactions of O(2l D2) with hydrocarbons have been studied e~tensively.l-~ The reaction cross-sections are large and the main reaction channel involves insertion into C-H bonds. Insertion has been shown to proceed indiscriminately and the total reaction cross-section found to be proportional to the number of C-H bonds in the m~lecule.~ A number of other reaction channels have also been recognised and may be summarised as follows, 0(2ID2) + R H - + ROHI ?-+ ROH (65 %) (1) (2) (3) (4) + R + OH (20-30 %) -+ R'O + H2( <lo %) -+ RH + 0(23~,)(<3 %).It is clear that quenching is negligible and that a direct abstraction reaction, leading to OH formation, plays an appreciable role. By comparison the reactions of 0(2lD2) with halogen-containing molecules have been little studied, although it is known that the reaction cross-sections are again The formation of halogen oxide products has been observed and lower limits for branching into this channel In the present work we have made a detailed study of the branching ratios into different reaction channels for a number of halogen-containing molecules. The domi- nant channel is shown to be abstraction of a halogen atom.Quenching to the ground state is also an important process. We also present data for the reaction of O(23PJ) with CFJ and compare these, together with data for the other halogenomethanes, with those for the analogous re- actions involving O(2lD2).M. C. ADDISON, R . J . DONOVAN AND J . GARRAWAY 287 EXPERIMENTAL Three separate experimental arrangements were employed for this work, all of them based on the flash photolysis technique. (i) FLASH SPECTROSCOPY A conventional arrangement, suitable for photographing transient spectra in the visible and ultraviolet regions, was used to obtain kinetic data on the halogen oxides and CF2. Spectra were dispersed on a Hilger-Watts medium quartz spectrograph and recorded on Kodak Panchro-Royal film. A more detailed description of this technique and the data processing has been given in ref.(6) and (7). (ii) TIME-RESOLVED PHOTOMETRY I N THE VACUUM ULTRAVIOLET This apparatus employed a conventional flash photolysis unit coupled to a vacuum ultra- violet monochromator and fast photometric recording system. It was used to monitor the formation and decay of O(23P.,) (via the resonance lines at A = 130 nm), following quenching of O(2lD2) by the halogenomethanes, and also to obtain absolute rate data for reaction of OQ3P,) with CF31. for work on S(33PJ); however, for the present work an EMR542 solar blind photomultiplier was used. The use of this photomultiplier eliminated the effect of scattered light from the flash lamp and allowed kinetic measurements to commence during the flash.A flow system was used for the atomic lamp and the best results were obtained when very low (<O.l %) oxygen/helium ratios were passed through the microwave discharge. An extensive series of experiments was carried out to establish that this new arrangement gave a linear photo- metric response with stable molecules such as 03. Curves of growth for O(23PJ) were then determined by photolysing O3 under optically thin conditions [in the presence of excess Nz to quench O(2lD2) to O(23PJ)] over a range of pressures (fig. 1). As a final check the rate of the reaction between O(z3PJ) and NO2 was determined * as k == (1.1 C 0.3) x lo-" cm3 molecule-' s-', in excellent agreement with the accepted result obtained by resonance fluorescence' [k -- (9.12 The experimental arrangement was similar to one described previously 0.44) x 10-l' cm3 molecule -'s-' at 295 K]. 0.4 O s 5 1 0.3 0.2 0.1 0 1 2 3 6 7 0 ( 23PJ ) / a r b .units FIG. 1.-Curve of growth for O(z3PJ) using the three resonance lines at 130.2, 130.5, 130.6 nm. The O(23PJ) concentration was taken to be proportional to that of 03, which was varied over the range 0.4-6.1 N m-'. * In these experiments O(23PJ) was formed by photolysis of NOz ( x 1 %) in the visible and near U.V. regions.In all of these experiments the output from the photomultiplier was fed to a fast analogue- to-digital converter (Datalab DL905) and data were processed in the standard way. (iii) The yield of OH from reaction of 0(2'D2) with CF2HCI was determined using an arrange- ment similar to that described by Morley and Smith." The intense OH emission produced by a microwave discharge through a flowing mixture of water vapour in argon carrier gas was focused through the reaction vessel and onto the slit of a McKee-Pederson (MP1018B) monochromator which selected lo the Q13 line at 308.15 nm.A chlorine gas filter surrounded the reaction vessel and reduced scattered light from the flash to a negligible level. The output from the photomultiplier was fed to a transient recorder (Datalab DL905) and data were pro- cessed as in section (ii) above. For all experiments 0(2'D2) was produced by the ultraviolet photolysis of 03(1 = 200- 300 nm) and, where required, O(23P.,) was formed by adding an excess of N2, to quench O(2lD2) to O(23P,). The experimental conditions used with the three different techniques varied significantly and will be described in the appropriate section dealing with results.TIME-RESOLVED PHOTOMETRY I N THE NEAR-ULTRAVIOLET RESULTS ABSOLUTE CONCENTRAT IONS OF o(2'02) PRODUCED BY THE FLASH Photolysis of 0, in the ultraviolet (200-300 nm) is known to produce almost exclusively O(2'0,) and thus the absolute yield of this atomic state can be determined by observing the amount of O3 removed by the flash. In pure 0, (or 0, + SF, and O3 + He mixtures) O(2l0,) reacts rapidly with a second O3 molecule, and under our conditions the amount of 0, removed immediately after the flash ?ill in fact be twice the amount photolysed in the primary photochemical step. However, by adding excess CO, to the ozone, the effect of the secondary reaction can be eliminated, as O(2'0,) is quenched to the ground state.We therefore carried out experiments to determine the amount of O3 removed after the flash (30 ,us) both in the presence and absence of CO,. The depletion in the presence of CO, was found to be 12 & 1 % (Po, = 26.6 N m-,) and in the absence of CO, 24 & 2 %, this gives a yield of O(2'0,) of 8 x loL4 atoms cm-3 per flash. These results confirm that the yield of O(23PJ) in the ultraviolet photolysis of 0, is negligible (< 10 %) and that O(2'0,) is removed en- tirely by reaction with 03, physical quenching being unimportant. The decay of 0, at times > 30 p s was observed to be very slow, as expected from the known slow rates for reactions involving O(23PJ) and O2(dAg) with 0,.REACTION OF O(2'0,) W I T H CF3C1 When O3 is photolysed in the presence of excess CF3Cl (PCFICI = 2.7 kN m-? a strong spectrum of C10 is observed via the (A2n t- X 2 n ) system, and its rate of formation closely follows the integrated form of the flash. No C10 is observed when CO, or N, is added to quench O(2'0,) and it is clear from these, as well as earlier experiment^,^*' that C10 results from a fast reaction between O(2l0,) and CF3C1. There are, however, three possible mechanisms for C10 formation. The first and most obvious is the direct formation of C10 in a primary abstraction step O(2'0,) + CF3C1+ CIO + CF3. ( 5 ) A second possibility is insertion into the C-C1 bond followed by fragmentation to yield C10 0(2lD,) + CF3Cl -+ CF30Cll -+ CF3 + C10.(6)M . C . ADDISON, R . J . DONOVAN AND J . GARRAWAY 289 The third possibility is that C1 atoms are produced by a displacement reaction, fol- lowed by the fast reaction of C1 with 03, i.e., O(2'0,) + CF3C1 -+ CF30 + C1 (7) (8) C1+ 0,- C10 + 0,. Under the conditions of our experiment it would be difficult to distinguish between these three mechanisms simply by observing the rate of formation of C10 as they are all very rapid. We can, however, use a chemical method to distinguish between the first two and the third mechanisms. By adding a small amount of ethane to mixtures of 0, and CF,Cl (PCZHs = 67 N mW2), any C1 atoms formed in the primary step can be removed before reacting with 0,; the yield of C10 will then be reduced by an amount which depends on the yield of free chlorine atoms in the primary step.As the pressure of C2H6 used is very much lower than that of CF,CI it will not interfere by reacting with O(2l0,) (>95 74 of the excited oxygen atoms react with CF,Cl). Our results show that the yield of CIO is only slightly reduced by addition of C2H6 and C1 atom formation accounts for <20 % of the total O(2'0,) removal by CF,Cl. Quantitative yields of C10 were determined oia the (5, 0) band of the A211 +- X211 system" [see ref. (7) for a detailed discussion], and when these are compared with the amount of O(2'0,) produced by the flash we find that 65 % of the excited oxygen gives rise to C10 formation in a primary step. Measuring the yield of OQ3PJ) by time resolved photometry in the vacuum ultra- violet proved more difficult than first envisaged.Absorption by CF3Cl reduced the intensity of the oxygen resonance line reaching the photomultiplier (the 130.6 nm line was used), as expected; however, we also observed a change in the sensitivity with which OQ3PJ) could be detected : as the extent of absorption by CF3Cl increased, the sensitivity for detecting O(23PJ) decreased. Thus in order to measure the yield of OQ3PJ) quantitatively, new curves of growth were determined over a range of condi- tions under which the oxygen resonance lines were attenuated by an absorbing gas such as CF,CI. For the present work relative O(z3PJ) concentrations were read directly from the appropriate curve of growth and the yield of O(23PJ) formed by the quenching of 0(2102) by CF3C1 was determined by comparing the concentrations produced in the absence and presence of excess N2 [the N, quenches a large and calculable fraction of the 0(2'D2) directly to the ground state].By this means the branching ratio for O(23PJ) formation was determined as 30 2:; %. Typical conditions in these ex- periments were Po, = 0.5 N m-2, PCFBCl = 4.0 N m-2, with a flash energy of 180 J. The branching ratios for all the channels determined in this work are summarised in table 1. We also include a recent estimate 12a for the elimination channel 12b O(2'0,) + CF3Cl--+ CF20 + FCl (9) which is seen to have a small branching ratio. REACTION OF 0(2'D2) W I T H CF3Br AND CF31 Reaction of 0(2'D2) with CF,Br results in the rapid formation of BrO; however the decay is also rapid and this makes the absolute determination of the BrO yield extremely diffic~lt.~ Nevertheless we can obtain a useful lower limit for the yield of BrO and using the extinction coefficient given by Clyne et al.13 for the (4,O) band of the A211-XZll system we find a branching ratio for BrO formation of >25 %. It should be noted that reaction between Br and O3 is much slower than the corre-290 REACTIONS OF O(2'0,) AND 0(33P,) WITH HALOGENOMETHANES sponding reaction for C1 atoms, and that we can therefore distinguish between BrO formed in a primary step and that formed by secondary reaction of Br with 03.Experiments to determine the yield of I 0 from reaction of O(2'0,) with CF31 are considerably more difficult than the corresponding experiments with CF3Cl and CF3Br. A strong spectrum of I 0 is observed; however, some photolysis of CF31 occurs, (it absorbs in the same region as 0,) and it is known that iodine atoms react rapidly with 0, to yield 10.Preliminary results from our laboratory show that spin-orbit excited TABLE BRANCHING RATIOS FOR PRODUCT CHANNELS IN THE REMOVAL OF 0(2'D2) BY HALOGENOMETHANES ~~ ~ ~ reactant products/% quenching halogen halogen other to 0(23~.,) oxide atom products CF3Cl 30 2 % 65 i 10% <20 % FCl( z 10 %) CF2HCl 28 ? ::%* 55 10% <lo% OH(5%) CF3Br >25 % ~~~ * Yield of OQ3PJ) based on CF, formation (i.e., uia the dissociative excitation channel yielding CF2 + HCl + 0) is 45 f 10 %. iodine atoms, I(S2Pli2), react more rapidly with 0, than ground state I(52P3/2) atoms; the rate constant for the ground state iodine atom reaction has been determined14 as k = 0.8 x Thus I 0 can be formed by more than one reaction and detailed experiments are required to distinguish between the various possibilities.Our present results indicate that the yield of I 0 from reaction of O(2'0,) with CF31 is substantial and we hope to report a quantitative measure for the branching ratio into this channel at the Discussion. cm3 molecule-' s-l. REACTION OF O(2'0,) WITH CF2HCl Strong transient spectra of C10 and CF2 were observed following the photolysis of 0, or N20 in the presence of CF2HCl (Po, = 13.3 N m-2; PCFzHC1 = 2.7 kN m-,). Both spectra were completely suppressed by addition of excess N2, showing that they resulted from reaction of O(2'0,) with CF,HCl.Photolysis of CF,HC1 in the far- ultraviolet is known to produce CF,; however, this region is not transmitted by our equipment and CF, was not observed when CF,HCl (PCFaHC, = 2.7 kN m-2) alone was flashed in the reaction vessel. The yield of C10 was measured as described for CF,Cl and found to be 55 & 10 % of the initial 0(2'D2) yield. Addition of small amounts of ethane to the system had no significant effect on the C10 yield, showing that C1 atom formation is of little importance (< 10 %) in the removal of O(2'0,) by CF2HC1. The yield of CF, was determined using the known extinction coefficient for the v; = 6 band (249 nm) of the A'B, +- X'A, system, given by Tyerman." The branching ratio into this channel was determined as 45 zt 10 %. The branching ratio for O(2,P,) formation was measured using the method de- scribed above for CF3C1 and found to be 28 Z i2 %, which suggests that both CF2 and O(23PJ) must be formed in the same process. The formation of OH radicals was not observed using plate photometry, however we would expect OH to react rapidly with CF,HCl under the conditions employed. Using the more sensitive technique of time resolved spectrophotometry at 308 nmM .C . ADDISON, R . J . DONOVAN AND J . GARRAWAY 29 1 (PCFIHCI = 2.0 kN m-2, POa = 40 N m-2) formation of OH was detected but in very low yield. A careful calibration of the system was achieved using the reactions of O(2'0,) with H20 and CH,. Assuming that H20 gives two OH radicals for each O(2'DJ atom reacting, the yield of OH from CH, was found to be 80 %, in good agreement with previous work.I6 The yield of OH from CF2HCI, based on the same method, was found to be only 5 %.REACTION OF O(2,PJ) WITH CF31 The kinetics of O(2,PJ) removal by CF,I were investigated using time-resolved spectrophotometry at 130 nm (the slit width used was 800 pn and thus the three atomic lines at 130.2, 130.5 and 130.6 nm were transmitted by the monochromator). By photolysing 03(P0, = 1.33 N m-2) in the presence of excess N2(PN2 = 800 N m-2), suitable concentrations of O(23PJ) could be generated ( ~ 3 % photolysis of 0, oc- curred). The decay of the ground state oxygen atom under these conditions was found to be very slow, as expected. Addition of small partial pressures of CF31 (0.13-0.6 N m-2) resulted in a marked increase in the rate of decay and by measuring the pseudo first-order rate coefficients for removal of O(2,PJ) over a range of CF,I pressures (data are shown in fig.2) the second-order rate constant was determined as, ko(23p,) + CFsI = (1.1 & 0.3) x cm3 molecule-' s-'. I I I I I I I 0 0.1 0.2 0.3 0.4 0.5 0.6 PCF3, / N m-2 FIG. 2.-Plot of the first-order rate coefficients for removal of O(23PJ) against partial pressure of CF31. (Pas = 1.3 N m-'; PNa = 800 N m-'). A small correction'' was made for departure from Beer-Lambert behaviour and the slope of fig. 2 should be multiplied by 1.3 ( y = 0.76 based on the data in fig. 1) to obtain the rate constant given above. Some photolysis of CF31 will inevitably occur under the conditions used; however, the percentage photolysis will be much less than that for 0, (ie., <3%), due to the lower extinction coefficient for CF,I, and should have no effect on the kinetics of the oxygen atom decay.As a check, further experiments were carried out over a range of flash energies (180-320 J). No significant difference in the decay rate for O(2,PJ)could be detected and we conclude that radical-radical reactions do not influence the observed kinetics and that photolysis of CF31 is unimportant. Some slow regeneration of O(z3PJ) will occur via the reaction of 02(a1Ag) with 03, but this is entirely negligible on the time scale used here. DISCUSSION REACTION OF o(2'02) WITH CF3C1, CF,Br A N D CF31 A major channel in the reactions of 0(2'D2) with halogenomethanes (excepting attack on C-F bonds)," is clearly the formation of a halogen oxide molecule. We shall concentrate our discussion on the reaction with CF3Cl, as the data for this molecule are most complete, but we expect the same general points to apply for CF3Br and CF31.Formation of C10 from CF3Cl can in principle occur by two mechanisms, the more direct being abstraction of a chlorine atom. The second possible mechanism involves insertion of 0(2ID2) into the C-Cl bond, to form a vibrationally excited hypochlorite CF30Cl '+, followed by fragmentation. CF30C1 is a stable molecular species and its thermal and photochemical reactions have been examined. The results suggest that the favoured primary dissociation channel is formation of CF30 and C1 (thermochemically this is the most favourable dissociation process).Thus if insertion of 0(2'D2) into C-Cl bonds was important, we would expect a high yield of Cl and not C10, contrary to observations. Our results therefore suggest that C10 formation occurs by a direct abstraction mechanism. Similar behaviour has been reported previously for reactions of singlet methylene (CH,), which is isoelectronic with O(2lD2), with ha loge no me thane^.^^-^^ Thus, while both singlet methylene and 0(2'D2) undergo fast insertion reactions into C-H bonds, the main reaction channel with halogenomethanes involves direct a b s t r a c t i ~ n . ' ~ - ~ ~ The above behaviour can be understood when we consider the strong interaction that will occur between the vacant p-orbital of 0(2'D2) (or CH,) and the lone pairs on the halogen atom.Thus the potential surface contains an attractive basin which surrounds the halogen atom and facilitates attack at this point in the molecule. A further attractive region must exist on the potential surface, corresponding to insertion of O(2'0,) into the C-C1 bond (the minimum corresponding to the ground state configuration for CF30Cl); however, it appears that this region is less accessible, possibly due to inertial effects; both Cl and CF3 are relatively heavy and need to move a substantial distance for insertion to occur (contrast this with the situation for C-H insertion where the much lighter H atom can move rapidly to accommodate the insertion process). Our data also provide information on another aspect of the singlet potential surface discussed above.Thus the singlet surface must be sufficiently attractive to be crossed by one or more triplet surfaces correlating with O(23PJ) + CF3C1 and non- adiabatic transitions at these crossings must be favourable, as evidenced by the rela- tively high branching ratio for O(23PJ) formation. For O(2'0,) interacting with CFJ the singlet surface may pass below the asymp- tote for O(23PJ) + CF31 (fig. 3) and could therefore influence the dynamics of the reaction between O(23PJ) with CF,I (see below). Stable compounds with the structure RIO can be prepared (e.g., iodosobenzene, C6H510) showing that the singlet surface has a very deep minimum in the region occupied by the lone pair electrons of iodine. * Removal of 0(2'D2) is much slower by CF, groups4*' and appears to proceed entirely by quench- ing.I8M .C . ADDISON, R . J . DONOVAN AND J . GARRAWAY 293 0(210,) + R I O ( ~ ~ P J -L t RI /- - - - R + I O FIG. 3.-Section through the proposed potential surfaces for O(Z3PJ) and 0(2'D2) interacting with an iodide. The lowest singlet surface is shown by the continuous line and the triplet surfaces by dashed lines. R I O REACTION OF 0(2'D2) WITH CF2HCI Lin23 has studied the photolysis of 0, in the presence of a number of hydrogen containing halogenomethanes, including CF,HCI, and observed stimulated emission from vibrationally excited hydrogen halide molecules formed in these systems. He proposed that this resulted from the insertion of 0(2'D2) into C-H bonds followed by the elimination of a vibrationally excited hydrogen halide molecule from the hot intermediate, e.g., 0(2'D2) + CF2HCl -+ CF,ClOHI --+ CFClO + HFI.(10) With CF2HCl, only HF emission was observed, although the formation of HCl is more exothermic. The present results clearly show that HF elimination cannot account for more than 10-20 %* of the total reaction cross-section and that elimina- tion of ground state HC1 is a more important process. It seems unlikely that chemical laser emission would result from a minor reaction channel and an alternative explanation for Lin's result is that excited HF is produced by secondary radical reactions. In a separate series of studies, Lin24 has suggested that the reaction, O(23PJ) + CF,H+ CFO + HF (AH = -433 kJ mo1-I) can give rise to H F laser emission. Our results show that both O(z3PJ) and CF,H are major products of the interaction of O(2'0,) with CF2HCl and we therefore sug- gest that reaction (11) could account for Lin's observations in the O3 + CF,HCl photochemical laser system.,, The dominant channel in the interaction of 0(2'D2) with CF2HC1 is clearly that leading to the formation of ClO(55 %) and, as the branching ratio is similar to that for CF,CI, we infer that the mechanism is the same.The second most important channel involves dissociative excitation viz., 0(2'D2) + CF,HCl+ CF, + HC1 + O(23PJ). (12) This channel is thermoneutral within the bounds of current thermodynamic data (AH = 17 -& 19 kJ mol-') and it is surprising that it competes so effectively with the other highly exothermic channels. However, the observed rapid formation of CF, and O(2,PJ) cannot be accounted for by any other process.We have shown that C1 atom formation is unimportant ( e l 0 % of the total cross-section) which rules out reactions such as O(2'0,) + CF2HC1+ CF2 + OH + C1. (13) * This is an upper limit based on the error bounds for the products which are directly observed.This is further confirmed by the very low yield of OH(5 %) observed. to be formed by the disproportionation reaction CF, is known 2CF,H+ CF, + CF,H2 (14) however, this could only account at most for 10 of the CF, observed as the dominant removal channel for two CF,H radicals is dimerisation. We, therefore, conclude that dissociative excitation [reaction (12)] accounts for z 40 % of the total cross-section. It is interesting to note that both the thermal and infrared multiphoton d i s s ~ c i a t i o n ~ ~ of CF,HCI lead to the formation of CF, and HCI.The other surprising feature is that OQ3P,) escapes from the force field of CF,, as CF,O is a very strongly bound molecule. This can, however, be understood when it is realised that CF2(X1A1) and O(23P,) do not correlate directly with the ground state of CF,O, but with an excited triplet state which may not allow efficient combination. From the bond additivity relationships suggested by Cvetanovic et al.3 and by Davidson et al.,5 we would expect OH formation to account for z 30 % of the total cross-section. This is clearly not the case and it appears that the distribution in the product channels does not follow the simple additivity relationship suggested for the total removal rates. The low yield of OH is in fact similar to the situation previously encountered with singlet methylene reactions, where it was found that attack at C-H bonds was reduced to a very low level when a chlorine atom is present on the same, or adjacent, carbon at om.26 The branching ratio for OH formation ( 5 "/o) is surprisingly low. REACTION OF O(23P,) WITH CF,Br AND CF31 Reaction of O(23PJ) with CF,Br, to yield BrO, is strongly endothermic (AH = +65 & 5 kJ mol-') and negligibly slow at 300 K. However, the reaction has been studied at elevated temperatures (800- 1200 K) and Arrhenius parameters determined 27 as A = (1.5 -& 0.5) x lo-" cm3 molecule s-' and E, = 57 4 kJ mo1-'. The activation energy for reaction is thus close to the endothermicity and the pre-expo- nential factor ( A ) is low when compared with reactions involving O(2'0,).We shall return to the latter point after discussing the corresponding reaction with CF,T. The reaction of OQ3PJ) with CF31 has been studied in some detail by Gorry et aL2* using the molecular beam technique and has been shown to involve the formation of a weakly bound collision complex. The product scattering (10) changes from a mainly backward, to a near isotropic distribution as the kinetic energy of the incident OQ3PJ) is increased. It was suggested" that at low collision energies the lifetime of the complex is shorter than its rotational period (as it is probably formed in low impact parameter collisions with low angular momentum).At higher collision energies the rotational period is reduced (higher angular momentum) and this leads to an increase in the forward scattering. The total cross-section for reaction was not determined in the molecular beam work but a thermally averaged (300 K) cross-section can be obtained from the present data as CJ 21 2 A'. It is clear that the reaction must be close to thermoneutral and our results provide an upper limit for the activation energy of E, < 6 kJ mol-I. When this is combined with the bond strength of CF31,29 D(CF,-I) = 221 +- 5 kJ mol-', we obtain a lower limit for the bond strength of I 0 as, D,(IO) 3 210 kJ mol-', which is consistent with the value given earlier by Radlein et al.30 We might expect the Arrhenius pre-exponential factor for reaction of O(23PJ) with CF31 to be similar to that for the analogous reaction with CF,Br, and the fact that the rate constant (at 300 K) for O(23P,) + CF,I is close to the pre-exponential factor forM .C . ADDISON, R . J . DONOVAN AND J . GARRAWAY 295 OQ3P,) + CF,Br, suggests that this is probably the case. These values are sur- prisingly low when compared with the analogous reactions for 0(2'D2) (where any kinematic constraints should be the same), but appear to be characteristic of reactions involving O(23P,) with halogen, or halogen-containing molecules. As these reactions involve attractive potential surfaces and a bound collision complex we would normally expect a substantial reaction cross-section or large pre-exponential factor. It has been suggested that the low values observed result from a very restrictive reaction geometry and that a near collinear collision is required before reaction can This was rationalised in terms of the molecular orbital structure for the collision inter- mediate which favours a linear 0-X-Y structure for lowest energy on the triplet potential surface.However, the above discussion on the quenching of O(2lD2) by halogenomethanes leads us to suggest an alternative explanation. We have seen that crossings between triplet and singlet surfaces must occur and that for iodoso com- pounds one of these may be close to the dissociation asymptote for O(23P,) +- RI (fig. 3). Thus the low reaction cross-section could result from a " low " triplet- singlet transition probability, while the scattering dynamics would be determined by the potential minimum in the singlet surface. CONCLUSIONS Reactions of O(2lD2) with halogenomethanes proceed with a large total cross- section, the dominant channel being abstraction to yield a halogen oxide.The singlet potential surface, on which these reactions occur, is strongly attractive and is crossed by lower lying triplet surfaces correlating to O(23PJ). This provides an efficient mechanism by which 0(21D2) is quenched to the ground state. The reactions of 0(2'D2) closely parallel those of singlet methylene. Reactions of O(z3PJ) with halogenomethanes have relatively low total cross-sections (and Arrhenius pre-exponential factors) and may involve a triplet-singlet surface crossing. We thank Drs H. Gillespie and G.Black for help in initiating this work and I.C.I. Ltd for the gift of samples of CF2HCl. H. Yamazaki and R. J. Cvetanovic, J. Chem. Phys., 1964, 41, 3703. A. J. Colussi and R. J. Cvetanovic, J. Phys. Chem., 1975, 79, 1891. P. Michaud, G. Paraskevopoulos and R. J. Cvetanovic, J . Phys. Chenz., 1974, 78, 1457. I. S. Fletcher and D. Husain, J. Phys. Chern., 1976, 80, 1837. J. A. Davidson and H. I. Schiff, J. Chem. Phys., 1978, 69, 4277. H. M. Gillespie and R. J. Donovan, Chem. Phys. Letters, 1976, 37, 468. H. M. Gillespie, J. Garraway and R. J. Donovan, J . Photochetn., 1977, 7, 29. R. J. Donovan and D. J. Little, Chern. Phys. Letters, 1972, 13, 488. D. D. Davis, J. T. Herron and R. E. Huie, J. Chem. Phys., 1973, 58, 530. M. A. A. Clyne and J. A. Coxon, Proc. Roy. SOC. A , 1968, 303, 207. and J. Wolfrum, Nature, 1976, 262, 204. lo C. Morley and I. W. M. Smith, J.C.S. Faraday ZZ, 1972, 68, 1016. l2 (a) J. Wolfrum and K. Kaufmann, personal communication; (b) R. J. Donovan, K. Kaufmarm l 3 M. A. A. Clyne and H. W. Cruse, Trans. Faraday SOC., 1970, 66, 2214. l4 M. A. A. Clyne and H. W. Cruse, Trans. Faraday Soc., 1970, 66, 2227. l5 W. J. R. Tyerman, Trans. Faraday SOC., 1969, 65, 1188. l6 C-L Lin and W. B. DeMore, J . Phys. Chem., 1973, 77, 863. l7 R. J. Donovan and H. M. Gillespie, Reaction Kinetics (Specialist Periodical Report, Chemical Society, London, 1975), vol. 1, p. 14. R. G. Green and R. P. Wayne, J. Photochem., 1977, 6, 371. l9 D. W. Setser, R. Littrelhand J. C. Hassler, J. Amer. Chern. SOC., 1965, 87, 2062. " C. H. Bamford, J. E. Casson and R. P. Wayne, Proc. Roy. SOC. A, 1966,289,287. 21 C . H. Bamford, J. E. Casson and A. N. Hughes, Proc. Roy. SOC. A, 1968,306, 135.296 REACTIONS OF 0(210~) AND 0(23~,) WITH HALOGENOMETHANES 22 R. L. Johnson and D. W. Setser, J. Phys. Chem., 1967, 71, 4366. 23 M. C. Lin, J. Phys. Chem., 1972, 76, 1425. 24 M. C. Lin, Int. J. Chem. Kinetics, 1973, 5, 173. 25 J. C. Stephenson and D. S. King, J. Chem. Phys., 1978, 69, 1485. 26 C. H. Bamford and J. E. Casson, Proc. Roy. SOC. A, 1969, 312, 163. 27 T. C. Frankiewicz, F. W. Williams and R. G. Gann, J . Chem. Phys., 1974, 61, 402. 28 P. A. Gorry, C . V. Nowikow and R. Grice, Chem. Phys. Letters, 1978, 55, 19. 29 E. N. Okafo and E. Whittle, Int. J. Chem. Kinetics, 1975, 7 , 273. 30 D. St. A. G. Radlein, J. C . Whitehead and R. Grice, Nature, 1975, 253, 37. 31 D. D. Parrish and D. R. Herschbach, J. Amer. Chem. Soc., 1973, 95, 6133.