Reaction of O(3P) with the alkyl iodides: CF3I, CH3I, CH2I2, C2H5I, 1-C3H7I and 2-C3H7I

The kinetics of the reactions between O(3P) and various alkyl iodides (RI) was investigated using the pulsed laser photolysis–resonance fluorescence (PLP–RF) technique at temperatures between 223 and 363 K. The reactions studied were O(3P) + CF3I (1), O(3P) + CH3I (2), O(3P) + CH2I2 (3), O(3P) + C2H5I (4), O(3P) + 1-C3H7I (5), and O(3P) + 2-C3H7I (6).

The Arrhenius expressions (units of cm3 molecule−1 s−1) obtained were: k1(223–363 K) = (8.73 ± 1.29) × 10−12exp{−(216 ± 38)/T}, k2(223–363 K) = (9.88 ± 1.67) × 10−12exp{(183 ± 50)/T}, k3 (256–363 K) = (7.36 ± 0.31) × 10−11 cm3 molecule−1 s−1, k4(223–363 K) = (1.58 ± 0.22) × 10−11exp{(239 ± 39)/T} k5(223–363 K) = (1.10 ± 0.17) × 10−11exp{(367 ± 42)/T}, and k6(223–363 K) = (1.84 ± 0.31) × 10−11exp{(296 ± 46)/T}.

With the exception of k1 and k3, all the other rate coefficients display a negative temperature dependence, which highlights the importance of association complex formation in reactions of O(3P) + RI.


Interest in the reactions of alkyl iodides (RI) with oxygen atoms (O(3P)) stems from a desire to understand the mechanism of these processes, which have multi-channel product pathways on singlet and triplet potential energy surfaces and which involve the initial formation of a R–I–O complex.1–5

Whereas IO elimination is the most important product channel for the smallest iodoalkanes, CH3I and CF3I (e.g., Φ(IO) ≈ 0.8–0.9 for CF3I),6 the presence of a β-position C–H bond in the longer chain iodoalkanes enables the intramolecular abstraction of an H-atom via a five-membered transition state to eliminate HOI with a yield as high as ≈0.8 for e.g., O(3P) + C2H5I.7

The formation of HOI in a spin forbidden process has been used as a laboratory source of HOI, as has the spin allowed formation of IO and R′.

O(3P) + RI → IO + R′ O(3P) + RI → HOI + R′R′ These reactions have found application in experiments designed to provide basic physico-chemical data on these molecules,7,8 and also parameters relevant to their atmospheric chemistry9,10 and photochemistry.11,12

The rate coefficients of reactions which proceed via formation of an association complex that can stabilise and rearrange to form products often display negative temperature dependences, i.e. the rate coefficients increase with decreasing temperature.

This effect has been observed for CH3I, but there are no data describing the temperature dependences of the rate coefficients for larger RI.

In this paper, we present the temperature dependences of the overall rate coefficients for the reactions of O(3P) with CF3I, CH3I, C2H5I, 1-C3H7I, 2-C3H7I and CH2I2.

For CH2I2, C2H5I, and 1-C3H7I these data are the first to be reported at any temperature.


The experiments were carried out at temperatures ranging from 223 to 363 K and pressures of 100 Torr He (or N2) using the pulsed laser photolysis, resonance fluorescence technique (PLP–RF).

For CF3I a systematic investigation of the pressure dependence was carried out at 223 K. The experimental set up for PLP–RF is essentially that which has already been described in the literature13 and is given briefly below.

The reactions were conducted in a thermostatted quartz reaction vessel of ≈500 ml volume with Brewster angle entrance and exit windows which could be heated to prevent condensation of ambient water vapour when working at low temperatures.

Gas flows were regulated by calibrated mass flow controllers, and the pressure was monitored using 100 and 1000 Torr capacitance manometers.

In some experiments CH2I2 was passed through a Teflon needle valve instead of the flow controller, but no significant differences could be observed.

The temperature in the reaction zone was monitored with a J-type thermocouple, and regulated by circulating cryogenic fluid through a jacket.

Typical flow rates were ≈500 cm3 (STP) min−1 (sccm), sufficient to replenish the reaction volume between laser pulses, and prevent build up of products which could result in secondary loss processes of O(3P).

Pulsed photolysis radiation at 351 nm was provided by an excimer laser at a repetition rate of between 2 and 10 Hz.

The photolysis beam intersected the light from an O-atom resonance lamp at the focus of a MgF2 lens that directed fluorescence to a VUV photomultiplier.

The resonance lamp was operated at a total pressure of 3 Torr He, and its emission was filtered by a 1 mm thick CaF2 window, that prevented transmission of shorter wavelength emission from N and H atoms.

Time dependent O(3P) signals were amplified, digitised and counted by a multichannel scaler usually operating at a resolution of 5 μs per channel, and averaging between 1000 and 3000 decay profiles.

Oxygen atoms were produced by the 351 nm photolysis of NO2: NO2 + hν → NO + O(3P) Typically, NO2 concentrations of (6 ± 1) × 1013 molecules cm−3 and laser fluences of 2–12 mJ cm−2 were used, resulting in the formation of approximately (8–50) × 1010 O(3P) cm−3 per pulse.

The approximate sensitivity (S/N = 1) for detection of O(3P) was ≈2 × 1010 atoms cm−3 for a total integration time of 5–10 ms.

An accurate measurement of the concentration of the alkyl iodides is necessary to derive rate coefficients (see below) and was carried out on-line by absorption spectroscopy using a 44 cm cell mounted at the exit of the reaction cell and maintained in a vacuum housing at room temperature.

The intensity of either the 185 or 254 nm line of a low-pressure Hg lamp, was monitored by two photodiodes (one transmission, one reference) to derive the optical density (OD) due to the alkyl iodide.

In separate experiments the dependence of OD185 or OD254 on the concentration of RI was determined by flowing RI diluted in He through an optical absorption cell equipped with a monochromator and a diode array spectrometer to obtain a simultaneous measurement of absorption between ≈210 and 360 nm.

This was converted to a concentration by least squares fitting to evaluated literature spectra,14 enabling effective cross sections at 254 and 185 nm to be derived as: σ254(CF3I) = 4.19 × 10−19, σ254(CH3I) = 1.14 × 10−18, σ185(CH2I2 = 3.33 × 10−17, σ254(C2H5I) = 1.14 × 10−18, σ254(1-C3H7I) = 1.35 × 10−18, σ254(2-C3H7I) = 1.23 × 10−18 cm2 molecule−1.


The commercial gases used in this study had the following stated minimum purity: NO2 (Aldrich 99.5%), He (Linde 99.999%), N2 (Linde 99.999%), CF3I (ABCR GmbH 99%), all other RI (Aldrich, 99%).

All RI samples were vacuum-distilled before use and the gas phase checked for I2 impurity by optical absorption at 450–610 nm.

Only for CH2I2 was I2 detected, at a level of 0.2% of the RI concentration.

For CF3I an upper limit of 0.01% could be determined, and for all other RI this value was <0.1 × 10−3.

Results and discussion

In all experiments the concentrations of O(3P) and RI were chosen so that pseudo first order conditions prevailed, i.e. [RI] ≫ [O(3P)] with a value of [RI]/[O(3P)] of between 400 and 8000.

In the absence of secondary loss processes for O(3P), the concentration profile of O(3P) is then given by: [O(3P)]t = [O(3P)]0 exp (−(k([RI] + d)t) where [O(3P)]t is the oxygen atom concentration at time t, and [O(3P)]0 is the initial oxygen atom concentration.

The term d represents the summed loss of O(3P) due to diffusion from the reaction zone and reaction with NO2 and was typically 500–700 s−1 (mainly due to reaction with NO2).

Plots of ln O(3P)-signal versust yield the first-order decay constant, k′, which is related to the bimolecular rate coefficient by: k′ = k[RI] + d The assumption of no significant secondary loss of O(3P) could be tested by demonstrating that variation of the laser fluence, and thus the radical density and degree of conversion of RI had no influence on measured values of k′.

The one exception to this was encountered in experiments with CH2I2, in which formation of CH2I and I occurs directly by photolysis.

This is discussed in detail below.

O(3P) + CF3I

The O(3P) + CF3I reaction was studied at seven temperatures between 223 and 363 K and at four different pressures from 20 Torr to 310 Torr in He and at 100 Torr in N2 at 223 K. A representative data set displaying O(3P) decays in the presence of various excess concentrations of CF3I (in this case at 243 K) is given in Fig. 1.

A plot of kversus [CF3I], the slope of which defines the rate coefficient according to eqn. (ii), is given in Fig. 2 for 223 and 333 K. It is apparent from this figure that the reaction of O(3P) with CF3I has an activation barrier (rate coefficient increases with increasing temperature).

The complete data set, covering temperatures between 223 and 363 K is given in Table 1 and displayed in Arrhenius form in Fig. 3.

Data from previous experimental studies, obtained using both resonance fluorescence detection of O(3P) and laser induced fluorescence detection of IO (the major product, see below) are also displayed in Fig. 3 and listed in Table 2.

The fit line shown is to the Arrhenius expression, k(T) = Aexp(−E/RT), and is described by: k1(223–363 K) = (8.73 ± 1.29) × 10−12exp(−{216 ± 38}/T) The errors in E/R are 2σ random only, returned by the weighted (by 1/σ2) fit to the data presented in Table 1.

The errors in the pre-exponential factor include both 2σ random error (as above for E/R) and also an additional 5% systematic error due to estimated uncertainty in the absorption cross sections of CF3I.

Table 2 compares the rate constants at 298 K and the Arrhenius parameters of this work with those obtained in previous studies.

The results from the most recent studies are in excellent agreement throughout the common temperature range.

Although the temperature dependence of Hölscher et al15. agrees with the present data, the absolute values are somewhat higher, as are those of Atkinson et al16. and the older room temperature studies of Addison et al17. and Watson et al.18

Watson et al. monitored IF formation from the reaction sequence: O(3P) + CF3I → IO + CF3O(3P) + CF3 → F2CO + FF + CF3I → IF + CF3 Given the indirect nature of this determination, the result of (6.5 ± 1.5) × 10−12 cm3 molecule−1 s−1 can be considered in satisfactory agreement with the present determination.

The result of Addison et al., k3 = (1.1 ± 0.3) × 10−11 cm3 molecule−1 s−117 was obtained by direct observation of O(3P) by resonance absorption in the presence of excess CF3I and is a factor of ≈2 larger than the present result.

Atkinson used cavity ring down detection of the IO product to obtain (5.8 ± 1.5) × 10−12 cm3 molecule−1 s−1 at 295 K.

Possible causes for determining rate coefficients that are too high include (a) the presence of reactive impurities (e.g., I2), (b) the secondary loss of O(3P) via reaction with IO or (c) underestimated CF3I concentrations.

As discussed already, all of these potential sources of error could be minimised in the present PLP–RF study.

As mentioned above, IO is a known product of reaction (1).

The thermodynamically accessible product channels (from Gilles et al6.) are: O(3P) + CF3I → IO + CF3O(3P) + CF3I → IF + CF2OO(3P) + CF3I → I + F + CF2OO(3P) + CF3I → I + CF3OO(3P) + CF3I + M → M + CF3IO Product formation in this reaction has been extensively investigated by Gilles et al.,6 who determined that channel (1a) dominates with temperature independent yields of IO of ≈0.8–0.9, whereas no evidence has been found for the formation of CF3O6 or IF.16

Gilles et al6. speculated that an adduct (either CF3IO or CF3OI) could be formed.

They also provide an extended discussion of the earlier literature regarding product formation, which is not repeated here.

However, since the work of Gilles et al., 6 indirect evidence for the presence of the pressure dependent channel at low temperatures (1e) has been given by Bloss et al12. who, in contrast to Gilles et al.,6 observed a pressure dependence in the IO yield at 223 K. In addition, the work of Dillon and Heard19 reveals an increase in the overall rate coefficient at 223 K with increasing pressure.

In agreement with the present work, a pressure dependence of k1 was not observed at room temperature.

In order to examine the presence of pressure effects at low temperatures, we carried out experiments at 223 K in the presence of between 10 and 310 Torr He, and at 100 Torr N2.

The results, shown in Fig. 4, indicate no significant pressure dependence at this temperature, in stark contrast to the result of Dillon and Heard.19

The reason for this discrepancy remains unclear.

Although the present data set weakens the arguments in favour of a termolecular channel, they do not rule it out, as the lack of pressure dependence could also be attributed to the reaction being at the high pressure limit at 10 Torr and 223 K.

O(3P) + CH3I

The PLP–RF data for CH3I (and the following RI) were obtained as described above, and the pseudo first-order decays and the plots of kversus [RI] were qualitatively similar to those for CF3I.

The complete data set obtained for reaction of O(3P) with CH3I is listed in Table 1, and plotted in Arrhenius form in Fig. 5, which also displays results of previous studies.

The temperature dependence of the rate coefficient is given by: k2(223–363 K) = (9.88 ± 1.67) × 10−12exp{(183 ± 50)/T} where the error limits were derived as for eqn. (iii).

The absolute values of the rate coefficient and its temperature dependence obtained in the present study are in excellent agreement with the measurements of Gilles et al6. which were also obtained using mainly O(3P) resonance fluorescence, but with some data obtained using LIF detection of IO.

The data of Hölscher et al.,15 obtained using only LIF detection of IO, agree at temperatures close to 300 K, but are systematically higher by ≈10% at lower temperatures and display a stronger temperature dependence.

The estimate of Kwong et al20. was based on an indirect flow tube study, and is orders of magnitude too low, suggesting that some of the assumptions made to describe the formation of excited I2 (the only product detected) were erroneous, as some of the now recognised product channels were not considered.

The products of this reaction have been examined in detail by Gilles et al.,6 and the possible exothermic channels are listed below: O(3P) + CH3I → IO + CH3O(3P) + CH3I → OH + CH2IO(3P) + CH3I → H + I + HCHOO(3P) + CH3I → I + CH3OO(3P) + CH3I → HI + HCHOO(3P) + CH3I → CH3IO (or CH3OI) The observation of multiple product channels, and a negative temperature dependence are indicative of a complex mechanism, though in this case rearrangement to form HOI via internal abstraction from a five-membered ring is not possible.

Gilles et al6. have determined product yields for IO (40–50%), OH (16%), H (7%), CH3O (<3%) and HI (<5%).

O(3P) + CH2I2

CH2I2 is the only di-iodoalkane investigated in this work, and hence the only one which has an absorption spectrum that extends out to 351 nm, the wavelength of the photolysis laser.

The absorption cross-section of CH2I2 at 351 nm is reported as ≈3.2 × 10−19 cm2 molecule−1,14 which is similar in magnitude to that of NO2, the precursor of O(3P) in these experiments.

As CH2I2 was present in concentrations similar to those of NO2, the photolytic generation of unwanted radicals (i.e. CH2I) at concentrations exceeding those of O(3P) is a potential problem.

CH2I2 + hν (351 nm) → CH2I + IFor experiments with CH2I2, two problems were encountered, which were not apparent for the other RI examined in this study.

Firstly, the signal due to O(3P) was modified by a post laser-pulse offset, which was also observed in the absence of NO2.

This offset signal was constant on the time scales of a typical experiment, suggesting photolytic generation of a non-reactive species that can scatter the light from the resonance lamp, with I atoms from reaction (10) being the obvious candidate.

The data were therefore treated by correcting for the measured offset, or by modifying eqn. (i) to account for time independent offset.

Both gave the same result, and the second option, which saves considerable experimental time, was adopted.

A more serious problem was the observation of a laser fluence dependence in the first-order decay constant of O(3P), with faster decay constants obtained at higher laser fluence.

This clearly indicated that O(3P) was being removed by reaction with CH2I or other reactive species formed in secondary processes.

At laser fluences ≤3 mJ cm−2 values of k′ were, within experimental uncertainty, independent of fluence.

All subsequent experiments were therefore conducted at a laser fluence of ≈3 mJ cm−2 and using higher NO2 concentrations than for the other RI in order to generate sufficient O(3P) for detection.

The data thus obtained are listed in Table 1 and displayed in an Arrhenius plot in Fig. 6.

The somewhat larger experimental uncertainty compared to those of the other RI reflects the problems mentioned above.

Within the experimental uncertainty and the temperature range covered the rate coefficient can be considered independent of temperature, with a value of k3 = (7.36 ± 0.47) × 10−11 cm3 molecule−1 s−1 obtained from a weighted average of the data in Table 1.

The error limits contain both random error (see footnote to Table 1) plus 5% systematic error related to uncertainty in the cross-section of CH2I2.

O(3P) + C2H5I

The present data yield a temperature dependent rate coefficient described by: k4 = (1.58 ± 0.22) × 10−11 exp{(239 ± 39)/T} The room temperature rate coefficient is k4(298 K) = (3.51 ± 0.24) × 10−11 cm3 molecule−1 s−1 whereby the errors contain both random error (see footnote to Table 1) and an estimated 5% systematic error arising from uncertainty in the absorption cross section of C2H5I. This value can be compared with unpublished data of Gilles ((2.38 ± 0.73) × 10−11 cm3 molecule−1 s−1) which was cited by Loomis et al.4

This value is significantly lower than the present result, and thus in contrast to the other data common to this work and the published data of Gilles et al6. which are in excellent agreement.

Previous studies of the dynamics of this reaction have shown the presence of at least two reaction pathways, generating HOI and IO:5,8 O(3P) + C2H5I → HOI + C2H4O(3P) + C2H5I → IO + C2H5 with ≈80–90% of the reaction proceeding via the HOI forming channel at 298 K.7

The negative temperature dependence of k4 is thus consistent with formation of a long lived association complex (i.e. R–I–O) which can rearrange via a five-membered ring to form the products of channel (4a), or dissociate to form R + IO as in channel (4b).

O(3P) + 1-C3H7I and O(3P) + 2-C3H7I

Our data on these two reactions give Arrhenius expressions of: k5 = (1.10 ± 0.17) × 10−11 exp{(367 ± 42)/T}k6 = (1.84 ± 0.31) × 10−11 exp{(296 ± 46)/T} The errors are derived as in eqn. (iii).

The room temperature rate coefficients are: k5(298 K) = (3.79 ± 0.31) × 10−11 cm3 molecule−1 s−1 and k6(298 K) = (4.97 ± 0.38) × 10−11 cm3 molecule−1 s−1, whereby the errors contain both random error (see footnote to Table 1) and an estimated 5% systematic error arising from uncertainty in the absorption cross-section of C3H7I. The room temperature result for 2-C3H7I is in excellent agreement with the value of (5.18 ± 0.44) × 10−11 cm3 molecule−1 s−1 presented by Gilles et al.6

As for C2H5I, the negative temperature dependence of both reactions indicates the formation of an association complex that can rearrange to form both IO and HOI:8 O(3P) + C3H7I → HOI + C3H6O(3P) + C3H7I → IO + C3H7 The yield of IO in reaction (6b) has been determined as 0.21 at room temperature,6 those of HOI or other product channels have not been determined.

The larger rate coefficient for 2-C3H7I compared to 1-C3H7I can be attributed to the six β-position H atoms available for formation of the five-membered ring for C3H7I compared to just two for 1-C3H7I, and to the greater electron density around the C–I bond due to two adjacent electron donating CH3 groups.

Steric effects presumably reduce these effects so that the rate coefficients differ by a factor of only 1.35.

Reactivity trends

The 298 K rate coefficients from this work are listed in Table 3 and reveal a strong trend in reactivity of various RI towards O(3P) with the slowest rate coefficient provided by CF3I, and the largest by the doubly substituted CH2I2.

The rate coefficients are between 3 and 7 orders of magnitude larger than those observed for reaction of O(3P) with the equivalent non-iodine substituted hydrocarbon, with k(O(3P) + CH3I)/k(O(3P) + CH4) = 3.7 × 106, k(O(3P) + CH2I2)/k(O(3P) + CH4) = 1.50 × 107, k(O(3P) + C2H5I)/k(O(3P) + C2H6) = 7.3 × 104 and k(O(3P) + C3H7I)/k(O(3P) + C3H8) = 4.9 × 103.

Rate coefficients for reaction of O(3P) with CH4 and C2H6 were taken from ref. 21 and for reaction of O(3P) with C3H8 from ref. 22 Such large effects are not observed with substitution by e.g., Cl.

For example, the ratio k(O(3P) + CH3Cl)/k(O(3P) + CH4)) = 34 (rate coefficient for O(3P) + CH3Cl from a compilation23).

The great enhancement in rate coefficient together with the change in sign of its temperature dependence (O(3P) + RH or O(3P) + RBr have positive temperature dependences) indicate a change in mechanism, as does the existence of multiple reaction pathways for O(3P) + RI.

These observations are further indicators of the participation of an association complex, the formation of which may involve charge donation from the electron rich iodine atom to the oxygen atom.

Indeed, as discussed by Gilles et al.,6 the room temperature rate coefficient for reaction of O(3P) with several fluorinated alkyl iodides is correlated with the ionisation potential.

In Fig. 7, we plot the natural logarithm of the 298 K rate coefficients obtained in this work along with those for other selected iodine containing alkanes as a function of ionisation potential.

The rate coefficients for reaction with O(3P) of the mono-substituted iodo-alkanes display a clear negative dependence on ionisation potential.

We note that the trend line in Fig. 7 was derived by fitting only to those data points in which a negative temperature dependence has been established.

When the pre-exponential factor is used in place of the room temperature rate coefficient, the correlation is barely discernible.

The observation of a positive temperature dependence for O(3P) + CF3I may imply that the rate coefficient is influenced by barriers to rearrangement of the association complex, and the dependence of k on ionisation potential may be modified.

The temperature dependence of the reaction between O(3P) and the other fluorinated species is presently unknown.

The data for O(3P) + CH2I2 does not fit well with this trend, possibly reflecting stabilisation of the association complex via involvement of two iodine atoms.1

The rate coefficients obtained at 298 K can be compared to data obtained for the reactions of the same RI with both the OH radical and Cl atoms, which are also listed in Table 3.

With the exception of CH2I2 (for which no rate coefficients for reaction with Cl or OH are available) some interesting trends are observed.

For both OH and Cl the rate coefficient increases with the number of C–H bonds from C1 to C3, with a lower rate coefficient for 2-C3H7I than for 1-C3H7I, reflecting the reduced number of weaker secondary C–H bonds for 2-C3H7I compared to 1-C3H7I. For the reasons discussed above, the relative reactivity of O(3P) with 1-C3H7I and 2-C3H7I is reversed due to the larger number of β H-atoms that can be internally extracted as the association complex rearranges via a five-membered ring transition state to form HOI.

We note that the general trend for the reactivity of O(3P) and Cl towards hydrocarbons, wherein Cl abstraction is usually orders of magnitude faster than O(3P) is reversed for most of the iodine substituted hydrocarbons investigated, again indicating the relatively minor role of abstraction compared to addition–elimination.

Summary and conclusions

Rate coefficients were obtained at different temperatures for the reaction of O(3P) with a series of iodine-substituted alkanes (CF3I, CH3I, CH2I2 C2H5I, 1-C3H7I and 2-C3H7I).

For reaction of CH2I2, C2H5I and 1-C3H7I with O(3P) atoms this work represents the first rate coefficient determination reported at any temperature and for CH2I2, C2H5I, 1-C3H7I, 2-C3H7I with O(3P) atoms the first study of the temperature dependence.

With the exception of CF3I and CH2I2, all the reactions display an increase in rate coefficient with decreasing temperature, and are several orders of magnitude faster than their non-iodine-substituted hydrocarbon analogues.

In the case of CF3I, the pressure dependence was investigated, and, in contrast to previous work, none was found at any temperature down to 223 K. These data substantiate the concept that, at low temperatures, the reactions between O(3P) and RI proceed mainly via formation of an association complex in which O(3P) adds to the I atom, and which can then rearrange to form a number of products.