Laser spectrometry and kinetics of selected elementary reactions of the acetonyl radical

The laser induced fluorescence excitation spectrum (LIF) and the ultraviolet absorption spectrum (TA) of the acetonyl radical (CH3C(O)CH2) were remeasured by using the time-resolved fast discharge flow (DF) and laser flash photolysis (LP) experimental techniques (T = 298 K).

The absorption cross section of σ(acetonyl, 207 nm) = (3.16 ± 0.61) × 10−18 cm2 molecule−1 was determined calibrated against the acetyl-peroxyl radical (CH3C(O)O2) in LP/TA measurements.

The kinetics of the reactions of CH3C(O)CH2 with the open shell reaction partners O2 (1), NO (2), NO2 (3) and H (4) were studied by using the DF method with LIF detection of the acetonyl radical at 298 ± 1 K and 2.85 ± 0.05 mbar He pressure.

The rate constants for the overall reactions were determined in units of cm3 molecule−1 s−1 to be k1 = (3.49 ± 0.51) × 10−13, k2 = (1.04 ± 0.19) × 10−11, k3 = (3.25 ± 0.65) × 10−11 and k4 ≥ 3 × 10−10 with 2σ accuracy given.

The acetonyl radical was found to react similarly to alkyl radicals by comparison with literature results.

A reduced reactivity was observed toward O2 and NO that might be attributed to the resonance stabilisation of the acetonyl radical.

No such effect was observed for the NO2 and H atom reactions.


Kinetic studies of the reactions of acetonyl radical, CH3C(O)CH2, are of interest for both practical and fundamental reasons.

One of the practical reasons is related to atmospheric chemistry.

The acetonyl radical is an important primary intermediate formed in the atmospheric degradation of acetone, CH3C(O)CH3, which has recently been found in surprisingly high concentration in the atmosphere.1,2

The oxidative degradation of acetone contributes significantly to the global budget of HOx (HO and HO2) radicals, particularly in the upper troposphere.1,2

Atmospheric loss processes for acetone are photolysis3 and reaction with OH,4,5 the relative importance of which depends on season, location and altitude.3–5

In the OH-reaction, the acetonyl radical is formed via hydrogen abstraction:OH + CH3C(O)CH3 → CH3C(O)CH2 + H2OThe further fate of CH3C(O)CH2 is determined essentially by its reaction with O2 in the atmosphere.

The acetonyl radical has relevance to combustion chemistry as well as through the reactions of oxygen atoms with butenes, e.g. it is formed in the reaction:6,7CH2 = C(CH3)CH3 + O(3P) → CH3C(O)CH2 + CH3The fundamental interest stems substantially from the recognition that the radical site in acetonyl is partially delocalized over the carbonyl double bond.8–11

This property of the acetonyl radical may give rise to unusual kinetic features allowing the investigation of structure–activity relationships.

Despite this importance and interest, only few studies have been performed so far on the kinetics of the reactions of the acetonyl radical.12–15

As part of an ongoing wider program, we have studied the reactivity of the acetonyl radical toward open shell molecules and report here the results for reactions (1)–(4) at room temperature.

We report also a reassessment of the optical spectra of the acetonyl radical6,7,12,16 with the main objective of characterising them for the purpose of further kinetic studies.

This paper is organised as follows: First, the experimental techniques are presented in quite some detail because they contain several new elements to our previous publications.

The absorption and fluorescence spectra of the acetonyl radical are discussed next, together with the determination of the absorption cross section at a single wavelength.

The rate constants for reactions (1)–(4) are reported after that and are compared with literature data.

Finally, some of the reactivity properties of the acetonyl radical are discussed and compared with those of the alkyl and alkoxyl radicals from the literature.

One of the conclusions from our current work is that the CH3C(O)CH2 radical behaves similarly in its reactions to the alkyl radicals and therefore we prefer its name as “acetonyl” instead of the more common “methylvinoxyl” which implies more an alkoxyl radical character.CH3C(O)CH2 + O2 → productsCH3C(O)CH2 + NO → productsCH3C(O)CH2 + NO2 → productsCH3C(O)CH2 + H → products

Experimental section

The discharge flow apparatus

Kinetics of the reactions (1)–(4) were investigated with the discharge flow method (DF) coupled with laser induced fluorescence (LIF) detection of CH3C(O)CH2.

A diagram of the DF/LIF apparatus is shown in Fig. 1.

Characteristic for our set-up is the vertically positioned flow-tube which is equipped with a moveable injector and operated by a versatile gas-handling and vacuum system.

The flow-tube was constructed of Pyrex and had an inner diameter of 40.3 mm and an overall length of 600 mm.

Its internal surface was coated with halocarbon wax (Halocarbon Corporation, Series 1200) or a thin film of Teflon (DuPont, 856-200 FEP) to reduce the effect of heterogeneous wall reactions.

Wall coatings were prepared similarly to the methods described by Kaufman in his classic work, e.g. in .ref. 17

A thermostatting jacket surrounded the reactor through which temperature-regulated fluids could be circulated by means of a thermostat (Julabo, F81); in the present work the reactor was thermostatted to room temperature.

The reaction pressure was measured using a calibrated capacitance manometer (MKS Baratron, 10 Torr head) at the downstream end of the flow-tube.

The sliding Pyrex double-injector consisted of an outer tube of 16 mm o.d. surrounding a coaxial shorter tube of 6 mm o.d.

Gases entered the flow-tube from the injector through a Teflon cap with radially directed holes to ensure rapid mixing.

Two flexible stainless steel tubes (1 m × 12 mm i.d.) connected the injector to the gas-handling manifold.

Acetonyl radicals were produced by reacting fluorine atoms with acetone in the flow reactor when the LIF spectra were recorded and inside the moveable injector during the kinetic experiments.

The latter arrangement had the advantage of providing constant initial acetonyl concentration at each position of the injector, which was moved to achieve time resolution in the kinetic measurements.

F atoms were generated at the upper end of the injector in an alumina-lined quartz tube by microwave discharge of F2 highly diluted in He.

The alumina insert tube prevented the formation of contaminating transient species in the discharge and increased the effective degree of dissociation of the fluorine molecule.

A double-discharge arrangement was applied for the investigation of the kinetics of reaction CH3C(O)CH2 + H (4) by discharging H2/He in a quartz side-arm at the upstream end of the flow-tube.

The microwave energy was obtained from 2450 MHz microwave generators (Medical Supplies, Microtron 200 or AHF-Analysentechnik, GMW 24-303 DR) feeding 40–50 W power into the discharges.

The flow-tube was coupled to an Al-block fluorescence cell which was black-anodised to reduce scattered light.

The acetonyl radicals were detected by tunable pulsed-laser excitation using a frequency-doubled dye laser (Continuum, ND-6000+UVX doubler) pumped by a Nd:YAG laser (Continuum, Surelite II-10).

The laser was operated at a repetition rate of 10 Hz with a power of 5–20 mW in the wavelength range of 330–365 nm.

Two laser dyes, Pyridin-1 (Radiant Dyes Chemie) and Oxazin-170 (Lambda Physik), were used to cover this wavelength range.

The pulse energy was continuously monitored with a UV-photodiode (EG&G, FND100Q).

The excitation laser beam entered the detection cell through a Brewster-angled window and a series of blackened baffles.

The induced fluorescence was collected at right angles to both the laser beam and the gas stream and imaged onto the photocathode of a photomultiplier (EMI, 9781B); a Wood’s horn light trap was positioned opposite the photomultiplier tube port.

The photomultiplier viewed the acetonyl fluorescence through a combination of glass filters (Schott, 2 mm BG-12 + 3 mm GG-385 + 3 mm KV-418) comprising a band-pass filter with the optical properties of λmax = 436 nm, Tmax = 48% and FWHM = 50 nm.

All optical windows and lenses both in the DF and LP set-ups were made of non-fluorescing fused silica (Heraeus, Suprasil).

The output current from the photomultiplier tube was fed into a purpose-built gated boxcar-integrator/data-acquisition system (constructed by Peter A. Halford-Maw, The University of Leeds).

The Q-switch output of the Nd:YAG laser was used to trigger the gate; the delay, with respect to the laser, and the gate-width were adjustable.

The analogue signal was digitalized and transferred to a laboratory PC for averaging and further processing.

Typically 1000 laser shots were averaged.

In some of the experiments, the LIF signal was observed with a 100 MHz digital storage oscilloscope (Hewlett-Packard, HP54601A) when the signal strength was equated with the maximum of the averaged fluorescence traces corrected for background.

Helium was the carrier gas which was passed through liquid-nitrogen cooled activated silicagel traps for further purification (two traps in series in each He line).

The main gas flows were regulated and monitored by mass-flow controllers (Tylan, FC-260) calibrated by bubble flow meters.

Reactant flows were regulated by needle valves (Hoke, Micromite) and measured by timing the pressure rise of the gas in a calibrated volume by means of a pressure transducer (Data Instruments, Model SA) connected to a strip-chart recorder.

One-stage rotary vacuum pumps (Edwards, E1M40 and E1M80) were used for passing the bulk gas flow through the reactor.

The overall volumetric flow rate was in the range 17–48 standard cm3 s−1.

The gas-handling vacuum system was operated with two-stage rotary vacuum pumps (Tungsram, BP-8 and BP-25) which also served for the evacuation of the flow-tube section between sets of experiments.

All vacuum pumps were connected to the apparatus through large N2(l)-cooled traps to prevent back-diffusion of oil vapour.

The vacuum-tightness of the system was monitored with a Pirani gauge (Edwards, AGD 101M).

The vacuum lines were grease-free and metal parts were avoided wherever possible.

Glass tubings were cleaned with 5% HF solution and rinsed with distilled water before installation.

The laser flash photolysis apparatus

Fig. 2 shows our recently installed laser flash photolysis (LP) apparatus which is furnished with UV-VIS transient absorption (TA) and pulsed laser induced fluorescence (LIF) monitoring assemblies.

Here we present a detailed description of only the LP/TA part of the apparatus which was used to obtain the absorption spectrum of the acetonyl radical.

For the purpose of the LP/TA studies, the CH3C(O)CH2 radicals were generated by laser flash photolysis of Cl2/acetone/He mixtures using the 351 nm XeF line of an excimer laser (Lambda Physik, Compex 201).

The generation of radicals and their detection were carried out in a 1020 mm cylindrical Pyrex cell of 25 mm internal diameter which was surrounded by a thermostatting gown and equipped with Suprasil end-windows at Brewster’s angle.

As seen in Fig. 2, the laser beam was reflected into the cell and passed longitudinally in a single path along its axis by using a 351 nm dielectric mirror (Laseroptik, HT: 200–340 nm).

A second dielectric mirror behind the cell prevented the laser radiation from entering the analytical light source.

Only the central homogeneous section of the laser beam was employed for photolysis to ensure homogeneous radical distribution.

The repetition rate of the laser was (1/6)–(1/3) Hz and the pulse energy delivered to the reactor was typically 20–40 mJ pulse−1.

The monitoring light source was a 150 W Xe arc lamp (Oriel, Model 6356) housed in a ventilated projector housing (Oriel, Model 6137) and powered by a low-ripple power supply (Oriel, Model 68806).

The photolysis beam and the collimated analysis beam were counter-propagating and overlapped along the cell axis.

The analysis beam leaving the cell was focused onto the entrance slit of a 640 mm focal length Czerny–Turner monochromator (Jobin-Yvon, HR 640) with a 1200 lines mm−1 grating blazed at 250 nm.

Both the entrance and exit slits were set to 0.5 mm resulting in a spectral resolution of 0.6 nm.

The monochromator was calibrated regularly against the 253.64 nm line of a low-pressure Hg lamp (Ultraviolet Products, Pen Ray).

The dispersed probe light was viewed by a photomultiplier (EMI, 9783R) and the arising transient voltage signal was captured by a 500 MHz digital-storage oscilloscope (Tektronix, TDS 640 A) and then digitised and transferred to a personal computer for averaging and analysis.

The digitised oscilloscope trace consisted of 2000 data points.

The 200–400 points constituting the pre-trigger baseline were averaged at each laser shot and this value was subtracted from the decay points to obtain the transient light absorption of the acetonyl radical (It).

The transmitted analysis light intensity in the absence of the transient (I0) was obtained before the laser flash by transferring the PMT signal to the PC via an interface.

The signal-to-noise ratio was improved by signal averaging: typically 400–2000 acetonyl decay traces were averaged.

The measurement was automated by means of an integrated hardware and software system (H-Interorg Ltd., Budapest).

The experimental observable in the LP/TA investigations was the time- and wavelength resolved absorbance with the averaged intensities Ī0(λ) and Īt(λ), that is At(λ) = ln (Ī0(λ) / Īt(λ)) was recorded from the measurements.

The photolysis cell was attached to a conventional vacuum line which was equipped with Teflon valves, operated with a two-stage rotary pump (Tungsram, BPL-25) through a liquid-nitrogen cooled trap and monitored with a vacuum indicator (MLW Labortechnik, Pirani-Vakuummeter).

Acetone and acetaldehyde gas mixtures in helium were prepared manometrically on the vacuum line and stored in blackened 10 litre bulbs.

The gas mixtures were prepared and the reaction pressure was set by means of a 1000 Torr absolute pressure gauge (MKS Baratron, 122A).

The photolysis gas mixture was slowly flown in the cell in order to avoid the accumulation of the reaction products.

The overall flow rate of the gases was such that the content of the photolysis cell was replenished in every 4–12 laser shots.

All gas flows were regulated by needle valves (Hoke, Micromite and Millimite) and were mixed-up in the main line of the vacuum system before entering the photolysis cell.

The composition of the photolysis mixture was established from the partial flows and the overall pressure.

The partial flows were determined by measuring the rate of pressure rise in the cell using the MKS pressure gauge.

The concentrations of Cl2, acetone and acetaldehyde were checked also by absorption measurements performed at 330, 274 and 281 nm, respectively.

The two methods gave results in good agreement.

In a few instances, we applied both the TA and LIF detection methods simultaneously for monitoring of the acetonyl radical in time-resolved laser flash photolysis experiments.

In this case, the kinetic apparatus shown in Fig. 2 was used in its full extension and the time delay between the pump and probe laser pulses was varied by means of a pulse generator (EMG, TR0361).


The suppliers and purity of the gases used were as follows: He (Linde, 99.996%), H2 (Messer-Griesheim, 99.9995%), O2 (Messer-Griesheim, 99.995%), F2 (Messer-Griesheim, 5% in 99.9995% purity He), Cl2 (Matheson, 2% in 99.996% purity He), NO (Messer-Griesheim, >99%), and NO2 (Messer-Griesheim, 98%).

NO and NO2 were purified by repeated trap-to-trap distillations in vacuum until the colour of the samples frozen at N2(l) became ‘off-white’ and ‘bright-white’, respectively.

Acetone (Aldrich, 99.9+% HPLC grade) and acetaldehyde (Fluka, >99.5%) were carefully degassed by freeze–pump–thaw cycles prior to use.

Further chemicals used in some test runs were Ar (Linde, 99.9990%), acetone-d6 (Aldrich, 99.9 atom% D) and 2,5-hexadione (Loba Chemie, puriss).

Results and discussion

All experiments were carried-out at room temperature (T = 298 ± 1 K) in He buffer gas.

The reaction pressure was P = 2.85 ± 0.05 mbar and P ≈ 200 mbar in the DF/LIF and LP/TA investigations, respectively.

Absorption and fluorescence spectra of the acetonyl radical

TA spectrum of CH3C(O)CH2

In the present work, we have measured the transient absorption spectrum for the acetonyl radical in the wavelength range of 200–340 nm and determined the absorption cross section at 207 nm.

As described in the experimental section, the acetonyl radical was generated by the pulsed laser photolysis of Cl2 at 351 nm in the presence of acetone.

The following main reactions occurred in the photolysis system:Cl2 + hν (351 nm) → 2 ClCl + CH3C(O)CH3 → CH3C(O)CH2 + HClCH3C(O)CH2 + Cl2 → CH3C(O)CH2Cl + Cl2 CH3C(O)CH2 → CH3C(O)(CH2)2C(O)CH3CH3C(O)CH2 → productsUnder our experimental conditions the laser-generated Cl atoms were converted completely to CH3C(O)CH2.

Acetonyl radicals were consumed predominantly in the self-combination reaction (8); the fast chain propagation step (7) regenerates Cl atoms and so does not lead to acetonyl consumption.

Reaction (9) represents essentially the diffusion loss of CH3C(O)CH2 out of the detection volume.

A representative absorbance vs. reaction time profile as obtained in one of the LP/TA experiments of acetone/Cl2/He mixtures is presented in Fig. 3.

As it is seen in Fig. 3, the absorbance decayed first fast, as a result of the dominance of the self-combination reaction (8), but levelled-off at longer reaction times due, at least in part, to the formation of secondary absorbing species.

At long reaction times (above ∼3 ms) and below about 220 nm and at around 280 nm detection wavelengths, a residual signal was observed that amounted to about 5–15% of the initial absorbance (see also in Fig. 3).

Probably, the combination product 2,5-hexadione (acetonyl-acetone) caused this residual light absorption as it was established by UV-spectrometric measurements of 2 mbar samples of pure acetonyl–acetone in a static system.

Acetonyl-acetone has been found to show the characteristic absorption spectrum of a carbonyl compound with absorption cross-sections of, e.g., σ (acetonyl-acetone, 207 nm) ≈ 1.5 × 10−19 cm2 molecule−1 and σ (acetonyl-acetone, 275 nm) ≈ 4 × 10−20 cm2 molecule−1.

In order to construct the absorption spectrum of the acetonyl radical, the absorbance values at zero reaction time (A0) are required where the light absorption is only due to CH3C(O)CH2.

A0 values were obtained by extrapolating second-order fits of the measured absorbance–time profiles at low-percentage depletion of CH3C(O)CH2 back to t = 0.

The extrapolation was done over a period of 100–200 μs after the laser pulse, where the complications due to secondary absorbing species are minimised.

While this extrapolation procedure did serve well to obtain the initial absorbances, the rate constant, k8, derived by the second-order plots increased with acetonyl concentration, contrary to the fact that the second-order kinetics were always obeyed well at the short observation times.

Thus, only a lower limit of k8 > 1.8 × 10−11 cm3 molecule−1 s−1 is proposed from our present study.

A more accurate value could have been obtained by working at larger CH3C(O)CH2 concentrations and performing rate parameter estimation by computer modelling of a complex reaction mechanism.

We are aware of a single other determination of the rate constant for reaction (8) by direct means which is k8 = (4.8 ± 0.3) × 10−11 cm3 molecule−1 s−1 (T = 298 K).12

The scaled absorption spectrum of CH3C(O)CH2 constructed by the initial absorbances is shown in Fig. 4.

The absorption cross section of the acetonyl radical was determined at a single wavelength in time-resolved spectrometric measurements calibrated against the acetyl-peroxyl radical, CH3C(O)O2.

The CH3C(O)O2 radicals were generated in separate flash photolysis experiments by replacing acetone for acetaldehyde and O2, whereupon the following main reactions took place:Cl + CH3CHO → CH3CO + HClCH3CO + O2 + M → CH3C(O)O2 + M2 CH3C(O)O2 → productsCH3C(O)O2 → productsThe spectrometric experiments were made back-to-back by maintaining the same Cl2 concentrations, laser intensity, gas flow-rate, etc., in the acetonyl and acetyl-peroxyl runs, so that the Cl atom concentrations were equal.

In fact, reaction (11) is not the only possible channel for CH3CO + O2.

As reported by Blitz et al18. and Tyndall et al.,19 this reaction also has a channel that generates OH radicals.

The yield is pressure dependent, and is a few percent at moderate pressures.

However, the OH radicals that might be formed were converted in a fast reaction with acetaldehyde to CH3CO and subsequently to CH3C(O)O2 in our acetyl-peroxyl system.

Therefore, no correction was needed to get [CH3C(O)O2]0 for the calibration of the absorption cross section of the acetonyl radical (see the concentrations below).

The measurements were performed at 207 nm analysis wavelength which corresponds to the site of the maximum for the stronger absorption band of the acetyl-peroxyl radical.20–21

Typical concentrations in the back-to-back experiments were the following (all in molecule cm−3): [Cl2] ≈ 1 × 1016, [acetone] ≈ 4 × 1016, [CH3CHO] ≈ 5 × 1016, [O2] ≈ 6 × 1017 and [CH3C(O)CH2]0 = [CH3C(O)O2]0 ≈ (1–4) × 1013.

At these concentrations the laser-generated Cl atoms were converted within a few microseconds to CH3C(O)CH2 and CH3C(O)O2 in the acetonyl and acetyl-peroxyl systems, respectively.

Examples of transient CH3C(O)CH2 and CH3C(O)O2 absorbance profiles are given in Fig. 5, where the corresponding second-order transforms of At−1vs. t are also presented.

The absorption cross section of CH3C(O)CH2 was obtained by the Lambert–Beer law for a single absorbing species using the expressionwhere A0(acetonyl, 207 nm) and A0(acetyl-peroxyl, 207 nm) are the extrapolated absorbances at t = 0 in the CH3C(O)CH2 and CH3C(O)O2 systems, respectively.

The initial absorbance values for the reference acetyl-peroxyl radical had to be extrapolated back to time zero also from short observation times (Fig. 5) because the secondary absorbing radicals CH3O2 and HO2 began to build up as CH3C(O)O2 consumed (see also in ref. 20).

Altogether six pairs of back-to-back experiments were carried out with the variation of the Cl2 concentration, laser intensity and laser repetition rate.

Neither of them was found to systematically influence the derived spectrometric results.

We have accepted the well-established absorption cross-section of σ(acetyl-peroxyl, 207 nm) = 6.75 × 10−18 cm2 molecule−1 recommended by Tyndall et al. for the reference radical in a recent review paper.21

In this way, we have obtained (3.16 ± 0.31) × 10−18 cm2 molecule−1 absorption cross section for the acetonyl radical, where the error given is 2σ statistical uncertainty of the average of the experimental determinations.

Propagating an estimated 10% uncertainty for the acetyl-peroxyl radical and adding an estimated 8% error to account for possible systematic errors in our experiments, an overall uncertainty of ∼20% can be assigned to the absorption cross section of the acetonyl radical.

That is,σ(acetonyl, 207 nm) = (3.16 ± 0.61) × 10−18 cm2 molecule−1is proposed from our current study with an accuracy given at the 95% confidence level.

The ultraviolet absorption spectrum of CH3C(O)CH2 obtained in the present work, and displayed in Fig. 4, is composed of a weak band with a maximum at around 315 nm and a stronger band below about 230 nm excitation wavelength.

The longer-wavelength absorption band corresponds to the B̃ ← X̃ electronic transition that can be viewed as a Π* ← Π transition on the carbonyl group of the radical.16,22

This same transition is involved in the LIF spectrum of the acetonyl radical (see below).

The strong absorption band at short wavelengths has not yet been assigned to our knowledge.

This might involve excitation to a higher-lying electronic state, C̃, in analogy with the unsubstituted vinoxyl radical (CH2C(O)H).23

The absorption spectrum of the acetonyl radical determined in the present work agrees favourably with the spectrum reported by Cox et al.12

They applied the technique of pulsed radiolysis combined with multipath transient absorption spectrometry and determined an absorption cross section of σ(acetonyl, 310 nm) = (8.73 ± 0.50)×10−19 cm2 molecule−1.

The relative absorption cross section we have obtained at 310 nm converts to ∼7×10−19 cm2 molecule−1 absolute value.

This data may be subject to significant uncertainty because of the poor signal-to-noise ratio of our measurements at this wavelength.

Extrapolation of the scaled absorption spectrum by Cox et al12. to 207 nm supplies an acetonyl cross section value of ∼4 × 10−18 cm2 molecule−1 which agrees well with our result.

LIF spectrum of CH3C(O)CH2

Fig. 6 shows the laser induced fluorescence excitation spectrum of the acetonyl radical observed in the region of 330–365 nm when acetone was reacted with fluorine atoms in the DF apparatus (the CH3C(O)CH2 concentration was approximately 2 × 1012 molecule cm−3 in these experiments).F2 + MW-discharge → 2FF+CH3C(O)CH3 → CH3C(O)CH2 + HFThe spectrum presented in Fig. 6 is corrected for variations in laser intensity and the peaks designated through “c” to “r” correspond to the assignment reported by Williams et al.16

The LIF spectrum we have obtained for the CH3C(O)CH2 radical is very similar to that reported by Washida and coworkers6 at higher pressures also from room temperature studies.

It consists of numerous overlapping vibrational–rotational bands of substantial intensity which are superimposed on a diffuse background.

The shortest-wavelength fluorescence is observed at 333.6 nm (29981 cm−1) laser excitation (peak “r”), whereas significant absorption still occurs in a wide range below this wavelength (cf.

Fig. 4).

This cut-off in LIF intensity is a strong indication of the onset of a non-radiative photochemical or photophysical process of the B̃ electronic state as discussed.16,22

The 0-0 band of the B̃ ← X̃ transition of the acetonyl radical is at 366.5 nm (27282.5 cm−1)16 that lies just beyond our detection range; the longest-wavelength measurable absorption occurs approximately at this band origin.12

The B̃ ← X̃ laser-induced fluorescence excitation spectrum of CH3C(O)CH2 was first published by Williams et al. from seeded supersonic molecular beam spectroscopy investigations16 and very soon after that by Washida and co-workers6 from laser flash photolysis experiments under bulk conditions.

Washida et al. have reported the LIF spectra of five methyl-substituted vinoxyl radicals including acetonyl (1-methylvinoxyl).

The authors have convincingly assigned the spectra to the corresponding radicals by using various sources of each radical.

The jet-cooled LIF spectrum of the acetonyl radical reported by Williams et al. contains more than 50 discrete vibronic bands.16

This complexity and the unusual intensity pattern of the spectrum were explained by a change in the preferred orientation of the methyl rotor upon electronic excitation.

In a subsequent publication, Williams and co-workers have presented a detailed assignment of the LIF spectrum of the jet-cooled acetonyl radical and an analysis of the dynamic features of the B̃ ← X̃ electronic transition which was supported also by high-level ab initio computations.22

To conclude this section, we note that the wavelengths of peaks “c”–“r” listed by Williams et al. for their jet-cooled LIF spectrum of the acetonyl radical16 agree well with those we have established for a hotter version of the spectrum (Fig. 6).

Therefore, these wavelengths can serve as fingerprints for identification of the acetonyl radical in complex chemical systems by LIF detection.

Reaction kinetics application

Both the laser induced fluorescence and UV transient absorption methods can be used favourably for the detection of the acetonyl radical in direct kinetic experiments.

To this end we note that several well-established methods are available for the generation of this free radical including chemical reactions, such as F/Cl + CH3C(O)CH3 (ref. 6 and this work) as well as laser flash photolysis of acetone,6 chloro-acetone,6 2,4-pentadione6,15,24 and 2-methoxy-propene.6,16

In a previous kinetic study25 we applied peak “n” and in the present work peak “e” for the detection of the acetonyl radical at 341.5 and 358.7 nm laser excitation wavelengths, respectively.

As it is seen in Fig. 6, these are the strongest transitions observed by us allowing sensitive and selective monitoring of CH3C(O)CH2.

The detection sensitivity for the CH3C(O)CH2 radical was ∼2 ×109 molecule cm3 both in the DF and LP investigations (the sensitivity is defined at the signal-to-noise ratio of 1).

In Fig. 6 the XeF excimer laser line (351 nm) and the third-harmonic generation wavelength of a Nd:YAG laser (355 nm) are also indicated.

We did observe strong LIF signals with these fixed-wavelength lasers, but the signal-to-noise ratio was much poorer compared with the frequency-doubled dye laser excitation.

There is also the possibility of detecting the perdeutero acetonyl radical, CD3C(O)CD2, by LIF and in this way to perform kinetic isotope effect studies.

In a few trial experiments we reacted F atoms with acetone-d6 in the flow reactor and observed strong fluorescence at and around 336.8 nm, 340.7 nm and 357.6 nm laser excitation wavelengths.

The LIF signal-strengths of the deutero-acetonyl radicals were found to be about three times larger than those observed for the undeuterated radicals providing an excellent detection sensitivity.

In the experimental arrangement shown in Fig. 2 we have compared the laser induced fluorescence and transient UV-absorption detection methods of the acetonyl radical.

In conformity with the general picture known for free radicals, the LIF technique was observed to be more selective and about fifty times more sensitive compared with the UV-absorption method.

We note, however, that the TA detection of CH3C(O)CH2 is also well suited for kinetic studies at around and below 207 nm, the absorption cross section being fairly large and because the absorption of many of the potentially disturbing radicals (e.g. alkyl-, acyl- and alkyl-peroxyl radicals) decreases significantly towards these short wavelengths.

Finally we mention that a combined application of the LIF and TA detection techniques has proven particularly advantageous in a series of recent kinetic studies in our laboratory on the reactions of the acetonyl radical where a knowledge of absolute radical concentrations is required24.

Rate constants for reactions (1)–(4)

The rate constants, ki, for the overall reactionsCH3C(O)CH2 + R(i) → productswith R(i) = O2(1), NO(2), NO2(3) and H(4), were determined with the DF/LIF technique by monitoring the depletion of the CH3C(O)CH2 concentration along the flow reactor for different positions of the moveable injector in the absence and the presence of an excess of the reactant R(i).

The initial acetonyl concentration was typically [CH3C(O)CH2]0 ≈ 6 × 1011 molecule cm−3 in these kinetic experiments.

In the absence of reactants, the heterogeneous reactionCH3C(O)CH2 + wall → productswas the major acetonyl loss process with a small contribution of the homogeneous self-combination reaction2 CH3C(O)CH2 → products (2,5-hexadione)The depletion of the CH3C(O)CH2 concentration without the reactants could be described by an effective first-order “wall rate constant”, kw.

Values were measured, with the exception of the NO2 reaction (see below), in the range of kw = 11–41 s−1 (P = 2.85 mbar) showing only a small increase with the initial acetonyl concentration which was varied by a factor of about 20 in the experiments.

These wall rate constants are of the usual magnitude observed for organic free radicals in discharge flow reactors coated with PTFE or halocarbon wax.

In the presence of the reactants R(i), the corresponding homogeneous gas-phase reactions (1)–(4) were the major consumption reactions of the acetonyl radical.

No indications for the occurrence of interfering parallel or consecutive reactions were found at low [CH3C(O)CH2]0 applied in the investigations.

Pseudo-first-order kinetics were presumed on evaluating the experimental observables.

Thus, the consumption rate of acetonyl radicals is given byor, with the premise that the wall activity is the same in the presence and absence of the reactants R(i)26 and assuming the validity of plug-flow conditions, by

Where Son and Soff are, respectively, the amplitudes of the CH3C(O)CH2 LIF signals with and without added R(i), z is the length of the reaction zone, v is the linear flow velocity (varied between 530 and 1510 cm s−1), ki′ (=ki [R(i)]) is the pseudo-first-order rate constant and t is the reaction time.

The semilogarithmic decay plots constructed according to eqn. (III) obeyed straight lines, with the exception of the H atom reaction (4) (see below), indicating the validity of first-order kinetics; the pseudo-first-order rate constants were determined as linear least-squares slopes.

The bimolecular rate constants were obtained from plots of kivs.


The inset in Fig. 7 shows some typical semilogarithmic decay plots measured for reaction (1).

In the main panel of Fig. 7, a plot of k1vs.

[O2] is displayed.

The straight line going through the origin represents a least-squares fit to the data points the slope of which gives the bimolecular rate constant k1.

In the present work we have not corrected the rate constants for viscous pressure drop and diffusion.

Instead, an 8% contribution to account for such effects was included in the error margins.

In previous studies with similar systems, we found the corrections to amount to about 2–8%.

The experimental conditions and kinetic results are summarised in Table 1.

The F2 concentrations listed are approximately equal to [CH3C(O)CH2]0 as estimated by the degree of dissociation of fluorine in the microwave discharge (∼70%) and an estimated loss of acetonyl radicals inside the moveable injector.

The error bars attached to the ki values in Table 1 are the overall uncertainties based on the observed statistical errors and the estimated systematic errors.

The overall uncertainties are proposed to represent 2σ.

Reaction CH3C(O)CH2 + O2 (1)

The rate constant we have determined for the overall reaction between CH3C(O)CH2 and O2 is k1(298 K, 2.85 mbar He) = (3.49 ± 0.51) ×10−13 cm3 molecule−1 s−1.

This is the first determination of k1 in the low-pressure regime providing further evidence for the fall-off behaviour of the reaction in comparison with literature data that were reported from kinetic studies at higher pressures.12,14

The pressure dependence was first established by Oguchi et al14. who studied the CH3C(O)CH2 + O2 reaction in the range of 20.3–449.3 mbar in He buffer at room temperature by using the LP/LIF experimental method.

In order to analyse and interpret the experimental rate data, they performed RRKM calculations based on DFT(B3LYP)-estimation of molecular and energetic parameters yielding the limiting high-pressure rate constant of k1(298 K) = (9.8 ± 0.8) × 10−13 cm3 molecule−1 s−1.

Oguchi et al. have derived the fall-off parameters according to Troe’s formalism27 as well from which a k1 value of 2.36 × 10−13 cm3 molecule−1 s−1 is obtained for the pressure of our present investigation.

This rate constant is about 30% smaller than the one we have determined, but the agreement is satisfactory in view of the fact that Oguchi et al. carried out the experiments in a region where the reaction displayed only weak pressure dependence and therefore the extrapolation to lower pressures might have substantial uncertainty.

The observed fall-off kinetics point to the dominance of a three-body combination between CH3C(O)CH2 and O2:CH3C(O)CH2 + O2 + M → CH3C(O)CH2O2 + MΔr1aH°298 = −129 kJ mol−128Further studies are required to decide whether other reaction channels also occur in the reaction between CH3C(O)CH2 and O2, such as the formation of the products CH2CO + CH2O + OH and HO2 + CO + C2H4 which are thermochemically accessible being exothermic by about −86 kJ mol−1 and −11 kJ mol−1, respectively.

Analogous reaction channels have been reported for the reactions alkyl + O229 and vinoxyl + O2.30,31

As noted in the Introduction, a major part of the atmospheric degradation of acetone involves the acetonyl radical formed by OH hydrogen abstraction.

The further fate of acetonyl is determined entirely by reaction (1) which is orders of magnitude faster than any of the competing processes due to its fairly large rate constant and the high O2 concentration in the atmosphere.

While this condition means simplification for atmospheric modelling studies, it is to be noticed that the reaction of CH3C(O)CH2 with O2 should be treated as pressure dependent practically over the whole T and P ranges in the troposphere and lower stratosphere.

Up to now, the kinetics of this important atmospheric reaction have been studied only at room temperature, over a limited range of pressures.

Reaction CH3C(O)CH2 + NO (2)

The rate constant of k2(298 K, 2.85 mbar He) = (1.04 ± 0.19) × 10−11 cm3 molecule−1 s−1 is proposed from our present work for the overall reaction of the acetonyl radical with nitric oxide.

This value is consistent with the results of recent experimental13,15 and theoretical15 studies, indicating the dominance of a pressure dependent combination reaction.

The reaction most likely leads to the formation of nitroso-acetone, CH3C(O)CH2NO:15CH3C(O)CH2 + NO + M → CH3C(O)CH2NO + MΔr2aH°298 = −151 kJ mol−128

(The reaction enthalpy we have estimated for reaction (2a) above agrees well with a very recent G2 computational result15 of −148 kJ mol−1.) The disproportionation reaction between acetonyl and NO giving HNO and c-propanone is considerably endothermic (by about 50 kJ mol−1) and so it has no contribution to the overall rate at room temperature.

A major part of our knowledge on the kinetics and mechanism of the reaction of acetonyl radical with NO arises from the combined experimental and theoretical work published very recently by Delbos and co-workers.15

These authors applied the complementary DF/LIF and LP/LIF experimental techniques to study the temperature and pressure dependencies of the acetonyl + NO and vinoxyl + NO reactions between 298 and 447 K in the wide pressure range of 0.8–900 mbar in He buffer gas.

They explored the potential energy surface of the reactions using ab initio molecular orbital computations at the G2 level of theory and utilised the ab initio data obtained for the stationary structures in unimolecular rate theory computations32 to estimate rate parameters.

These rate parameters were employed then to set the fall-off parameters for the analysis of the experimental rate constants in terms of Troe’s fitting procedure.27

From the final fall-off parameters Delbos et al. report, one obtains k2 (298 K) = 3.2 × 10−11 cm3 molecule−1 s−1 and k2 (298 K, 2.85 mbar He) = (1.3 ± 0.5) × 10−11 cm3 molecule−1 s−1.

The latter value is in satisfactory agreement with our experimental determination.

Reaction CH3C(O)CH2 + NO2 (3)

Quite unexpectedly, the kinetic study of the reaction between CH3C(O)CH2 and NO2 was complicated by heterogeneous effects.

In the first trial experiment a very large consumption of the acetonyl radicals was observed on the wall of the reactor.

This was accompanied by hysteresis: the wall consumption was found to be different when determined in measurements from short-to-long reaction time or viceversa (halocarbon wax was the wall coating in the NO2 experiments).

Even under such conditions, however, the “reagent NO2 on–off” technique has corrected very well for the heterogeneous wall reactions at deriving the bimolecular rate constant for reaction (3) as it is seen by the linearity of the semilogarithmic decay plot presented in Fig. 8.

The hysteresis could be completely eliminated by a 30 min conditioning of the reactor with the reaction itself prior to the kinetic runs.

The wall consumption was also reduced, but still remained significantly larger (kw ≈ 70 s−1) than observed with the other reaction systems.

The pseudo-first-order kinetics were obeyed well in all experiments conducted in the pre-treated reactor providing the bimolecular rate constant of k3 (298 K) = (3.25 ± 0.65) × 10−11 cm3 molecule−1 s−1.

As far as we are aware, there is a single other determination of k3 reported in the literature.13

Sehested et al13. used pulse radiolysis combined with transient UV-absorption spectrometry to study the kinetics of the reaction between CH3C(O)CH2/CH3C(O)CH2O2 radicals and both NO and NO2 at 295 K in the presence of a large excess of SF6 (∼1000 mbar).

They determined k3 = (1.6 ± 0.4) × 10−11 cm3 molecule−1 s−1 which is only half of the value we have obtained with the DF/LIF experimental technique.

At present, we can not offer any reasonable explanation for the large disparity in the results.

Heterogeneous effects were found to be significant in our study, but were controlled in the regular experiments and therefore we believe them not to give rise to a significant systematic error.

Sehested et al. carried out the experiments at very large acetonyl concentration (∼1 × 1015 molecule cm−3) that might have led to complications through secondary chemistry.

One would expect, however, the secondary reactions to result rather in an overestimation of the rate constant.

No experimental or theoretical information is available on the product channels of the reaction between CH3C(O)CH2 and NO2.

Barnhard and co-workers33 studied the related C2H3O (vinoxyl) + NO2 reaction by using the LP/LIF method in the temperature and pressure ranges of 295–374 K and 3.3–133 mbar (N2), respectively.

They found no pressure dependence which might indicate that the nitro-acetaldehyde adduct did not stabilise in their reaction system.33

Thus, it is not unreasonable to assume that the CH3C(O)CH2 + NO2 reaction occurs also not just via a simple radical–radical recombination mechanism in contrast to the O2- and NO reactions discussed above.

One possibility is that there exists a fast open reaction pathway via a series of chemical activation reaction steps (reaction (3b)).

The intermediate (CH3C(O)CH2O)*, if it is formed, is expected to undergo a fast decomposition reaction.34

The radical disproportionation (reaction (3c)) is also feasible thermochemically.

CH3C(O)CH2 + NO2 + M → CH3C(O)CH2NO2 + M Δr3aH°298 = −238 kJ mol−128 CH3C(O)CH2 + NO2 → (CH3C(O)CH2NO2)* (CH3C(O)CH2NO2)* → (CH3C(O)CH2O)* + NO (CH3C(O)CH2O)* → CH3C(O) + CH2O Overall: CH3C(O)CH2 + NO2 → CH3C(O) + CH2O + NO Δr3aH°298 = −29 kJ mol−1 CH3C(O)CH2 + NO2 → c-propanone + HNO2 Δr3aH°298 = −86 kJ mol−1

Reaction CH3C(O)CH2 + H (4)

Double-discharge experiments were carried out to study the kinetics of the reaction of CH3C(O)CH2 with H atoms (see Fig. 1).

The absolute hydrogen atom concentration was determined by the well-established gas-titration method with NO235 by observing the equivalence point in the reaction H + NO2 → OH + NO with LIF detection of OH.

We have found this reaction of the acetonyl radical to be extremely fast, in fact beyond the limit of measurability with the available gas pumping capacity in the current study.

The semilogarithmic acetonyl decay plots determined in the experiments showed level-off with increasing reaction time indicating a significant consumption of the excess component H atom in the reaction ([H]0/[CH3C(O)CH2]0 ≈ 5 was the initial concentration ratio).

Therefore, only the lower limit of k4 (298 K) ≥ 3 × 10−10 cm3 molecule−1 s−1 can be proposed from our present study for the rate constant of reaction (4).

The reaction is believed to proceed via a radical–radical recombination mechanism to form acetone: CH3C(O)CH2 + H + M → CH3C(O)CH3 + M Δr4aH°298 = −402 kJ mol−1

Reactivity of the acetonyl radical

The unpaired electron in the acetonyl radical forms a partially delocalized π-electronic system with the carbonyl double bond portrayed as a resonance electronic structure between two localised states, the alkyl form “acetyl-methyl” (1) and the alkoxyl form “propenyl-oxyl” (2):8–11

A measure of electron delocalisation is the resonance stabilisation energy (RSE) which has recently been suggested to be 17 kJ mol−1 for the acetonyl radical.10

Note that the “prototype” resonance stabilised allyl radical has 51 kJ mol−1 RSE.29

The resonance stabilisation is expected to give rise to reduced reactivity and the question arises whether the acetonyl radical behaves more like an alkyl radical or rather like an alkoxyl radical in its reactions.

These issues are examined in some detail in Table 2 where the acetonyl reactions are compared with analogous reactions of alkoxyl, alkyl and allyl radicals.

Radicals of roughly comparable size to acetonyl have been used in the comparison whenever possible.

During the discussion of the kinetic results above we have presumed that the O2 molecule adds to the CH2-end of acetonyl forming the acetonyl-peroxyl radical (CH3C(O)CH2OO).

This adduct radical has indeed been found and identified by its UV-absorption spectrum in the reaction.12,24

The alternative alkoxyl mechanism can practically be excluded on the grounds of the relatively small rate constants of the alkoxyl + O2 reactions and the instability of the trioxy radical that might be formed if the oxygen is added to the alkoxyl site of the acetonyl radical.29,36

The high-pressure limiting rate constant of the CH3C(O)CH2 + O2 reaction is about ten times smaller than that of the n-C3H7 + O2 reaction 29,37 which may be taken as an indication of the reduction of the reactivity of the acetonyl radical by resonance stabilisation.

The allyl radical has much higher degree of resonance stabilisation and hence is expected to be even less reactive than the acetonyl radical, but the k values for the acetonyl + O2 and allyl + O2 reactions are not very different.

In other words, there is no apparent correlation with the bond dissociation energy of the R-O2 bond formed in the reactions (Table 2).

A further indication of the effect of resonance stabilisation is provided by the observation that the pressure dependent region is shifted to higher pressures in the case of the acetonyl + O2 and allyl + O2 reactions compared with the alkyl+O2 reactions.7,29,37,38

The NO molecule is believed to add to the carbon radical site of the acetonyl radical similarly to O2.

This mechanism has been supported by a recent ab initio computation, although the authors noted that addition to the oxygen radical site and a subsequent isomerisation to nitroso-acetone could not be entirely excluded.15

Similarly to the respective O2 reactions, the fall-off regimes of the acetonyl + NO and allyl + NO reactions are shifted to higher pressures.15,39

That is, the observations indicate an efficient reformation of the energised adducts to the reactants in the third-body combination mechanisms of the acetonyl and allyl radicals with both O2 and NO.

The picture is different, however, concerning the high-pressure limiting rate constants.

In analogy with the O2 reactions, one would expect smaller k values for the acetonyl + NO and allyl + NO reactions than for the C2H5+NO reaction, but this is apparently not the case (Table 2).

The alkoxyl radicals undergo combination reaction with NO2 forming organic nitrates, while the alkyl radicals are known to react via O-transfer.29,41

The resonance stabilised allyl radical reacts also by oxygen atom transfer with NO2 as it was verified experimentally by the identification of the allyl-oxyl radical (C3H5O) in the reaction.42

This result provides further argument for the validity of the chemical activation mechanism (3b) we have proposed.

There is no indication for the effect of resonance stabilisation on the kinetics of the NO2 reactions listed in Table 2.

The disproportionation-to-recombination rate constant ratios available from the literature indicate that the alkoxyl radicals react mainly by direct hydrogen atom transfer, while the alkyl and allyl radicals via a combination mechanism in their reactions with H atoms.43–45

The alkyl and allyl reactions have significantly larger rate constants as well.

Therefore, based on the observed large k4 value, we assume that the H atom adds to the CH2 moiety of the acetonyl radical to form acetone in a fast recombination reaction.

There is no sign whatsoever of an effect of resonance stabilisation on the kinetics of the compared H atom reactions.

In summary, the kinetics features presented in Table 2 clearly show that the acetonyl radical behaves like a carbon centered radical and not like an alkoxyl radical in its reactions with molecules with unpaired electrons.

No general statement can be formulated, however, concerning the reduction of reactivity due to resonance stabilisation; the effect varies from significant to non-existent at all depending on the reaction partner.

Interestingly, the acetonyl and allyl radicals show very similar reactivity contrary that the allyl radical possesses a much higher degree of electron delocalisation.

Most of the reactions compared in Table 2 are radical–radical recombinations, the high-pressure limiting rate constant of which provide the rate constants for the association steps in the third-body combination mechanisms.

These rate constants are seen in Table 2 to vary over a wide range including more than two orders of magnitude variation of the acetonyl data.

In the absence of electronic barriers, the differences in the rates of the association reactions of alkyl and allyl radicals have been explained by dynamic features of the reacting systems and the topology of the potential energy surfaces.37,46–47

These include, for instance, the effect of anisotropic long-range potential in the entrance valley of the PES,46 mixing of the neutral potential energy surface with the charge transfer surface37 and the involvement of non-dissociative triplet-state complexes in the dynamics of the reactions.47

It would be of considerable interest to perform theoretical studies on the non-barrier effects for the acetonyl reactions as well in order to explain the large differences in the reactivities observed experimentally.