Vibrational deactivation studies of OH X 2Π (v = 1–5) by N2 and O2

The deactivation kinetics of vibrationally excited OH X 2Π (v = 1–5) were studied using a pulsed laser photolysis-pulsed laser induced fluorescence technique.

Temporal profiles of OH (v) were obtained by exciting off-diagonal (Δv = −1,−3) transitions in the A–X band of OH and monitoring the diagonal, blue shifted fluorescence.

Photolysis of O3 at 266 nm was used to produce O1D which reacted rapidly with H2, CH4 and H2O to produce OH (v).

Deactivation rate coefficients for OH (v = 1–5) with O2 and N2 and for OH (v = 2,1) with O3 were obtained.

The deactivation rate coefficients show an exponential dependence on vibrational level for both O2 and N2, however O2 is much more efficient.


An understanding of the kinetics of relaxation of vibrational levels of the OH radical is a topic of considerable interest both from fundamental and practical standpoints.1–3

In the troposphere and stratosphere the reaction of O(1D) with water produces OH in the first three vibrationally excited levels and with considerable rotational excitation.O(1D) + H2O → 2 OH (v ≤ 3)Reaction (1) is the main source of the hydroxyl radical in the lower atmosphere.

The reaction has been studied extensively4 and it is clear that the OH radical containing the old hydrogen-oxygen bond is produced vibrationally “cold” with 90% of the molecules in the v = 0 level, while the newly formed OH radical is produced with up to three quanta of vibrational excitation, the thermodynamic limit.

In a study of the reaction of O1D with H218O, Saunder et al.

(1992)4 reported relative OH vibrational populations of 0.39(v = 0):0.29(v = 1):0.3(v = 2) and we have observed population in v = 3.

Under tropospheric and stratospheric conditions N2, O2, and H2O are the potential quenchers for the vibrationally excited OH.

The rate of relaxation determines the extent of any deviation from local thermodynamic equilibrium.

In addition collisions with H2O are potentially reactive with implications for the isotopic distribution of tropospheric H2O. Reaction (1) is also the main source of interference in laser based sensors which utilize laser induced fluorescence (LIF) for the measurement of the OH radical.3

A detailed knowledge of OH relaxation rates in the lower atmosphere requires state specific relaxation rate coefficients for OH (v = 1–3) with N2, O2, and H2O. These rate coefficients are also required to model, and devise strategies to minimize, OH interference effects in laser based OH sensors.

The reaction of H atoms with ozone produces OH with up to nine quanta of vibrational excitation and an inverted vibrational population distribution.H + O3 → OH (v ≤ 9) + O2The vibrationally excited ground state OH radicals emit light in the infrared and red spectral region, hence reaction (2) is responsible for the intense Meinel emission from OH X 2A which occurs in the 80–100 km region and is one of the dominant features of the atmospheric nightglow.1,2

The spectral characteristics of these emissions are important to provide a comprehensive analysis of the chemistry taking place in the earth’s mesosphere and mesopause.5,6

Analysis of the temporal and spatial structures in these emissions can be used to study the propagation of atmospheric gravity waves through the mesosphere.7

Accurate vibrational deactivation coefficients of highly vibrationally excited OH are a critical component of models of nighttime airglow emissions.2,8

However the database on vibrational deactivation of OH with potential atmospheric quenchers is limited.

There have been several measurements of deactivation of OH (v = 1,2,3) with H2O, including one from our laboratory.9

There have been measurements on deactivation of OH with up to 12 quanta of vibrational excitation in collision with O2 but the database is limited,10–15 and there are no reported measurements of deactivation by the most abundant atmospheric gas, nitrogen, for vibrational levels below v = 8.

In this work we report deactivation coefficients for OH with up to five quanta of vibrational excitation by O2 and N2 and, in addition, some preliminary measurements on deactivation of OH (v = 1,2) with O3.


Our experimental approach has been described in detail elsewhere:9 We utilized the pulsed laser photolysis-pulsed laser induce fluorescence (PLP-PLIF) technique to measure temporal profiles of OH(v = 1–5) under pseudo-first-order conditions at different pressures.

OH radicals were produced by reaction of O(1D) with an H-atom donor: H2, H2O, or CH4.

Experiments were performed in an octahedral stainless steel cell having internal volume of about 3.2 L. Two Pyrex glass cylindrical side arms, 4 cm internal diameter and 5 cm long, were attached on opposite sides of the cell.

Photolysis and probe lasers were aligned on the same axis, entering the cell through a quartz window.

The O(1D) was produced by the pulsed laser photolysis of O3 using the 266 nm fourth harmonic output from a Nd-YAG laser (Quanta-Ray GCR 16).

Typical initial O(1D) concentrations were calculated to be 8 × 1012 atoms cm−3, based on laser fluence and ozone concentration.

The OH radical, formed by the reaction between the O(1D) and the H-atom donor, was then excited by a Nd:YAG pumped, frequency doubled, tunable dye laser or, alternatively, by an optical parametic oscillator (OPO: Quanta-Ray MOPO-730), pumped by the third harmonic of a Nd-YAG laser (Quanta-Ray PRO 270).

Vibrationally excited OH radicals were monitored by LIF exciting Δv = −1 or −3 transitions in the A–X band at the wavelengths listed in Table 1.

The fluorescence was monitored via the A–X (0–0, 1–1 and 2–2) transitions at 308–320 nm using a photomultiplier tube, and using appropriate filters to minimize scattered light.

The photomultiplier output was appropriately terminated and fed to a 500-MHz digital oscilloscope to obtain the integrated voltage averaged for (typically) 100 laser shots.

Kinetic information was obtained by varying the delay between the photolysis and the probe lasers using a digital delay generator.

Experiments were performed under “slow-flow” conditions in order to avoid buildup of by-products.

Flows were monitored using calibrated mass flow meters.

N2 and O2 were used without further purification.

Bulbs containing dilute mixtures of the H-atom donors were made up manometrically.

Ozone was produced by flowing pure oxygen through a commercial ozonizer and freezing the product on silica gel in an ozone trap cooled by a dry-ice–methanol slush (196 K).

Ozone concentrations were measured by photometry at 254 nm.

The ozone cross section used to calculate the concentration was σ = 1.15 × 10−17 cm−2.16

Concentrations of O2 were calculated from flows.

Concentrations of N2 were calculated from flows and, because of the large partial pressures of N2, were measured barometrically.

Agreement between the concentrations measured barometrically and those based on flows was excellent.

The pure gases and chemicals used in this study had the following stated minimum purities: H2, 99.999%; CH4, 99.99%, O2 99.99%, He 99.99%.

They were used without any further purification.

OH(v) production and detection

OH radicals were produced by the following sequence of reactions:O3 + hv266nm → O2(1Δ) + O(1D)O(1D) + CH4 → OH(ν) + CH3O(1D) + H2 → OH(ν) + HO(1D) + H2O → 2 OH(ν ≤ 3)As noted above LIF was excited by long wavelength off-diagonal transitions with fluorescence detection on the strong (0,0), (1,1) and (2,2) transitions.

This excitation/detection approach has two advantages, firstly we observe fluorescence which is blue shifted relative to the excitation wavelengths minimizing probe laser scatter.

Secondly, ozone does not absorb at the excitation wavelengths, eliminating the production of OH by the probe laser.

The LIF signal from OH generated by the probe laser can be a significant source of interference when excitation transitions utilizing Δv = 0, +1 branches are used since the absorption coefficient of ozone is significant at these wavelengths.

OH (v = 1,2) were monitored exciting the hydroxyl radical on Δv = −1 transitions at 345.9 nm and 350.9 nm respectively.

The OH (ν = 4, 5) were monitored exciting the hydroxyl radical on Δv = −3 transitions at 451 nm and 453.3 nm.

OH (v = 3) was monitored using both approaches at 356.5 nm and 449 nm.

The transitions and wavelengths are summarized in Table 1.

We used Lifbase17 simulations to model and assign the observed OH LIF spectra and found very good agreement between the simulations and the experimental spectrum assuming that the OH is rotationally thermalized.

Figs. 1 and 2 show sections of the LIF spectra of the (0–3),(1–4) and (2–5) transitions of the OH radical produced in reaction (4).

The lines used for kinetic studies of v = 4 and 5 are labeled.

Simulations computed with Lifbase are included for comparison.

The rotational temperatures used in the simulations are 298 K. The relative vibrational populations of v = 3, 4 and 5 were taken to be 5:1:0.4 and give good agreement between the experimental spectra and the simulations.

Knowledge of the vibrational product distributions in reactions (1), (4) and (5) is important in determining an experimental strategy for the excitation of the various vibrational levels and for assessing possible complications from vibrational cascading.

Product distributions for the reaction of O(1D) with hydrogen and methane have been extensively studied using LIF and infrared emission spectroscopy although attempts to determine the full product vibrational distribution are limited.18–21

We calculate ΔHr,298° = −43.0 kcal mol−1 for reaction 4 with CH4 as the H donor and −43.7 kcal mol−1 for reaction (5) with H2.

These values of ΔHr,298° would limit OH production to levels with 4 quanta of vibrational excitation or less.

However the translational energy of of O1D produced by reaction 3 is considerable.

Dylewski et al22. have measured the energy and angular distribution of O1D produced from ozone photolysis at wavelengths between 235 nm and 305 nm.

At 265 nm they find that O2(1Δ) is formed predominantly in v = 0 with O1D being produced with 0.6 eV of translational energy.

This additional translational energy makes production of OH (v = 5) exothermic and we observe its production in both reactions.

This appears to be the first observation of v = 5 production in reactions (4) and (5).

The LIF and infrared emission studies suggest the vibrational distributions are similar for both reactions (4) and (5).

The population distribution peaks in v = 2 with about 25% of the population, approximately 20% is in v = 0,1,3 and approximately 10% in v = 4.

We have not attempted to accurately quantify the relative vibrational distributions and the distributions noted above for reaction (4) may be partially relaxed.

We did not perform extensive spectral scans for reaction (5).

We observed production of vibrationally excited OH in v = 1, 2, and 3 in the reaction of O(1D) with H2O. As noted above the dynamics of the reaction between 16O(1D) with H218O has been studied extensively.

Sauder et al4. measured the vibrational distribution in the newly formed 16OH and reported relative populations 0.39 (v = 0): 0.29 (v = 1): 0.3 (v ≥ 2) while the OH radical with the old bond was formed predominantly in the ground state.

Again, while we made no attempt to accurately quantify the nascent OH product distribution from this latter reaction, we were able to observe the production of OH(v = 3).

Population of OH(v = 3) has been previously observed in this laboratory.9

The formation of OH(v = 3) from thermal reactants would be slightly endothermic using ΔfH298(OH) = 9.31 ± 0.29 kcal/mol as suggested in the JANAF compilation.23

However ΔfH298(OH) has been recently re-evaluated by Ruscic et al.24

Using their revised ΔfH298(OH) = 8.89 ± 0.09 kcal mol−1, the formation of OH(v = 3) is slightly exothermic.

As for reactions (4) and (5), the translational energy associated with the O1D reactant could allow exothermic production of OH (v = 3) using either thermochemistry.


The kinetic analysis that is appropriate to our experimental conditions has been described in detail by Silvente et al.9

O1D is produced by photolysis of ozone and reacts rapidly to produce vibrationally excited OH.

The timescale for reaction of 99% of the O1D was typically 15 μs at the lowest reactant concentrations, producing both vibrationally and rotationally excited OH.

Rotational equilibration is known to be fast,12 and this was confirmed by our experimentally measured LIF spectra which showed no deviation from 300 K rotational distributions on timescales as short as 1 μs at low pressure.

We are confident that all kinetic measurements monitored rotationally thermalized OH.

Fast OH production was followed by relatively slow decay of each vibrationally excited OH level.

Under our typical experimental concentrations [O3] = 1.6 × 1014 cm−3; [H2] = 6.5 × 1014 cm−3; [CH4] = 6.5 × 1014 cm−3; [H2O] < 6.5 × 1013 cm−3, relaxation by the OH precursors was small compared with the measured vibrational decay rates.

Our data analysis was identical to the procedure adopted previously in our study of deactivation rates of OH (v = 1,2,3).9

If we assume pseudo-first-order conditions ([R] ≫ [O(1D)]), the temporal decay profiles for each vibrational level can be described by an analytical expression9.[OH(v)]t = Cvexp(−kt)wherek′ is the pseudo-first order loss rate of OH (v) i.e.kv[O2] or kv[N2]

kL is the pseudo-first-order loss rate of O(1D), i.e. kO(1D)+R[R] and R is the OH precursor

pv is the fractional population produced in vibrational level v.

Solutions give a complex double exponential expression with production and loss terms.

If cascading is negligible the deactivation process produces only OH(v = 0) and all vibrational levels follow an identical expression.

If cascading, i.e. stepwise quenching, is included, a more complex expression is obtained which contains the detailed state-to-state quenching rate coefficients, as described in appendix 1 of Silvente et al.

Fitting to the detailed analytical expression which includes quenching becomes impractical but the profiles are still described to a good approximation by eqn. (I) if kq increases significantly as a function of increasing vibrational level.

In this case, the quenching rate constant for any given level becomes limiting at long times, as all higher levels have already decayed.

Under these conditions the decay portion of the temporal profile is described by a simple exponential loss.[OH]v = C exp (−kt)In some cases complete temporal profiles were obtained including both production and loss of OH(v) and these profiles were fitted to the eqn. (I) containing the exponential production and loss terms but neglecting quenching.

The decay section of these profiles was also fit by the simplified single exponential loss expression eqn. (II).

In all cases the pseudo-first-order decay rates obtained using eqn. (I) agreed well with those obtained from the single exponential fit, eqn. (II).

However this requires acquisition of many data points at short reaction times to give the fast production portion of the temporal profile and this does not give any information on the quantity of interest, the deactivation rate coefficient.

In this work most of the temporal profiles included only a few points at short reaction times and decays were evaluated visually and the exponential portions of the decay were fitted to the single exponential expression, eqn. (II) to give the pseudo first order decay rate k′.

Fig. 3 shows an example of such decays for deactivation of v = 4 in N2. The decays show good exponential behavior over greater than three 1/e decay times.

The vibrational deactivation rate coefficient, kv, was then obtained by plotting k′ as a function of the concentration of the quenching gas, as shown in Figs. 4–6.

By using three OH production schemes with different chemistry and different vibrational product distributions we attempted to ensure that the influence of secondary chemistry and vibrational cascading on our measured rates was minimal.

Individually measured rate coefficients together with 2σ errors of precision are detailed in Table 2.

The averages of these rate coefficients for each vibrational level are summarized in Table 3.

The errors in Table 3 are ±30% and represent a conservative estimate of overall accuracy based on our ability to reproduce well established thermal OH rates.

O2 Deactivation rate coefficients

Experiments using oxygen as collider, were typically carried out between 30 and 130 Torr of total pressure, helium being the buffer gas and with O2 partial pressures of up to 40 Torr being used in measurements on v = 1.

Experiments were performed using all three H-atom donors for measurements on OH(v = 1).

For v = 2–5 two production schemes were used for each level.

Deactivation rate coefficients were obtained by plotting the pseudo-first order decay rate as a function of O2 concentration.

Fig. 4 shows plots of pseudo-first-order decay rate vs. [O2] for OH(v = 2,4,5).

The deactivation rate coefficients obtained in the O2 experiments are summarized in Table 2 which lists the rate coefficients obtained with the individual H-atom donors.

Typical 2σ errors of precision are less than 5% and it can be seen that the rate coefficients obtained using the different OH production schemes are in excellent agreement suggesting that errors due to cascading effects or secondary chemistry are negligible.

N2 Deactivation rate coefficients

Experiments in N2 were performed using all three H-atom donors for OH (v = 1,2) and both the H2 and CH4 were used for measurements on OH (v = 3,4,5).

Due to the extremely low efficiency of nitrogen in the relaxation process, pressures of up to 400 Torr were required to measure significant decay rates for OH (v = 1).

Fig. 5 shows a plot of the pseudo-first-order decay rate vs. [N2] for OH (v = 3,4,5).

The results are summarized in Table 2 with the rate coefficients obtained using each H-atom donor listed individually, and, again, we obtain excellent agreement between the rate coefficients obtained using the different precursors.

O3 Deactivation rate coefficients

In order to be able to assess the effects of O3 on vibrational deactivation, rate coefficients were measured for deactivation of OH (v = 1,2) with O3.

These experiments were much more limited in scope with single sets of measurements being made for each level.

Reaction 4 was used for OH production.

At higher ozone concentrations there was evidence for complications from secondary chemistry manifested by an initial rise in the signal, an initial fast decay and then a slow decay.

Simulations indicated that the initial fast decay represents the vibrational deactivation rate and that the “tail” is caused by slow generation of vibrationally excited OH via the reaction of H atoms with ozone.

Fig. 6 shows a plots of pseudo first order decay rate as a function of O3 concentration for OH (v = 1,2) with the pseudo first order decay rate calculated from the initial fast decay.

The rate coefficients are listed in Table 2.

The 2σ errors of precision in these measurements are ∼20% for each measurement, considerably higher than the errors of precision in the N2 and O2 data.

In addition, given the limited number of measurements made and the complications from secondary chemistry at high O3 we consider the uncertainty in these rate coefficients to be considerably higher than for those measured with O2 and N2.

However effects of secondary chemistry should only lead to an underestimation of the actual deactivation coefficient and they should be considered as good lower limits to the actual deactivation rate coefficients.

Potential sources of error

Cascading and secondary chemistry are the main potential sources of error in these experiments and this is discussed in detail by Silvente et al.9

Cascading effects are errors caused by the relaxation from higher vibrational levels into the vibrational level being monitored.

Such an effect would result in the measured deactivation rate coefficient being too slow.

For this effect to be significant there must be an appreciable population of higher vibrational levels relaxing into the level which is being monitored and the relaxation rates must be similar.

Secondary chemistry results in the production of vibrationally excited OH on long time scales and again will result in measured deactivation rate coefficient being too slow.

Production of H atoms and their subsequent reaction with ozone is the most important potential source of artifacts due to secondary chemistry and can occur with H2 and to a lesser extent CH4 as the H-atom donor.

This should not be a problem with H2O. It can be made insignificant if low ozone concentrations are used.

An analysis of secondary chemistry using Acuchem suggests that at the ozone concentrations used here such effects are indeed negligible for the O2 and N2 measurements.

As noted above elevated O3 levels are required to measure its deactivation rate and complications from secondary chemistry are unavoidable.

For the O2 and N2 measurements, the good agreement between rate coefficients obtained using the different H-atom donors suggests that artifacts due to secondary chemistry and vibrational cascading are negligible.


Table 3 summarizes the vibrational deactivation coefficients measured by LIF during this study together with the prior results for O2.

Three prior experimental studies have reported rate coefficients on the deactivation of the lower vibrational levels of the OH radical by O2.

Rensberger et al12. selectively produced OH(v = 2) by infrared excitation and measured the vibrational deactivation rate by LIF.

This experimental approach should preclude complications from cascading from higher vibrationally excited states to v = 2.

Dodd et al. used electron beam excitation of O3/O2/H2/Ar mixtures to produce vibrationally excited OH via reaction (2).

The OH was then monitored by infrared emission.

Their initial work used a single quantum deactivation model to interpret their data and rate coefficients were reported for deactivation of OH (v = 1–3).10

In a subsequent publication11 they reported rate coefficients for deactivation of OH (v = 1–6) based on a more extensive analysis using a model which included multiquantum deactivation and more detailed chemistry.

Our results for OH(3–5) are in good agreement with the results of Dodd et al.

However our measured rate coefficients are significantly slower than the values reported by Rensberg et al. and Dodd et al. for v = 2 and by Dodd et al. for v = 1.

The reasons for this discrepancy are unclear.

As we note above vibrational cascading is a potential problem in chemical production of vibrationally excited OH and this might influence the results for low vibrational levels when CH4 or H2 are used as precursors, however the amount of v = 3 produced with H2O as a precursor appears to very small and both cascading effects and complications from secondary chemistry should be negligible for our v = 2 measurement.

If cascading were producing an artifact in our measurements the very different vibrational product distributions for reaction (1) as compared with reactions (4) and (5) should produce significantly different deactivation rate coefficients.

As we show below the variation of the deactivation rate coefficients for N2 are very well described by a power law dependence.

For O2 neither an exponential nor a power law fit gives a good functional fit to the dependence of the deactivation rate coefficient on vibrational level.

There have been no previous reports of rate coefficients for vibrational deactivation of v = 1–5 by N2 although upper limits of kdeactivation < 1 × 10−14 cm3 molecule−1 s−3 have been reported for v = .2–412,14,19

Deactivation rate coefficients for v = 8, 10 and 12 have been reported.15,25

Fig. 7 shows the rate coefficients measured here together with prior values for v = 8,10 and 12.

Kaufman and coworkers26 used a power law dependence to fit deactivation rate coefficients of HF and HD and such a dependence provides an extremely good fit to our data for v = 1–5, although the high v data is not well represented.

Fig. 7 shows the best fit of our data to a power law function which gives:kv,N2 = 1.47 × 10−15v2.22 cm3 molecule−1 s−1We are unaware of any previous results on deactivation of lower vibrational levels by O3.

There have been two reports of deactivation coefficients of v = 9 by O3, both of which reports rates in excess of 1 × 10−10 cm3 molecule−1 s−1.27,28

One theoretical study reports rates for deactivation of OH (v = 1,2) which are approximately an order of magnitude smaller than our measured rates.29

As noted above the rates reported here may be influenced by secondary chemistry which would have the effect of producing anomalously low rates.

We discussed the role of chemical bonding in mediating the deactivation processes in our previous work which was limited to deactivation of OH (v = 1–3).

The limiting case appears to be a situation in which the deactivating molecule can form a stable molecule via three body recombination.

In this case the rate coefficient for vibrational deactivation appears to be a good proxy for the high pressure limit for three body recombination.

In this case one might expect the deactivation rate coefficient to be independent of vibrational level and the mechanism of deactivation to proceed predominantly through multiquantum deactivation.

In recent work in our laboratory we have measured the deactivation of OH (v = 1–5) with NO2.

We measure a fast deactivation rate coefficient, kdeactivation = (6.4 ± 0.1) × 10−11 cm3 molecule−1s−1, which is independent of vibrational level and is in very good agreement with the directly measured high pressure limit for the recombination reaction.

Our experimental approach precludes any determination of the deactivation mechanism.

Theoretical studies have examined the deactivation efficiencies of both O2 and N2.

Shalashilin et al30. and Caridade et al.,31 have considered the role of the HO3 surface in promoting the efficiency of vibrational deactivation by O2.

In both cases they calculate deactivation efficiencies that are considerably greater than the experimental values, particularly for lower vibrational levels.

It appears that while the HO3 surface clearly influences the deactivation dynamics any complex that is formed it is too short lived and weakly bound to allow a statistical randomization of energy.

Shalashilin et al. examined the deactivation efficiency of N2 in collisions with vibrationally excited OH and OD.

They find that deactivation of OH is extremely inefficient and calculate rate coefficients which are significantly slower than our experimental values.

Atmospheric implications

We can now quantitatively evaluate the relative efficiencies of significant atmospheric colliders in the deactivation of vibrationally excited OH formed in the troposphere.

Our results confirm that O2 is much more efficient in deactivation of the lower vibrational levels than N2.

However the contribution of N2 is not negligible and accounts for 13% of the deactivation of v = 1 by N2 + O2.

Our results suggest that under typical conditions in the tropical marine boundary layer, collision with H2O is the dominant deactivation process.

O2 and H2O contribute equally to deactivation at H2O partial pressures of ∼0.6 Torr.


We have used the pulsed laser photolysis-laser induced fluorescence technique to measure the vibrational deactivation coefficients for OH(ν = 1–5) with O2 and N2.

Our results in O2 are in excellent agreement with the only previous experimental result present in the literature for OH(v = 3–5), while they are approximately a factor of two lower than previous measurements for OH(v = 1,2).

This is the first work to report vibrational deactivation coefficients for OH (v = 1–5) with N2.

Our results confirm that oxygen is much more efficient in deactivating vibrationally excited hydroxyl radical than nitrogen.