Environ. Sci. Technol. 2006, 40, 5956-5961
Acetone-h6 or -d6 + OH Reaction Products: Evidence for Heterogeneous Formation of Acetic Acid in a Simulation Chamber ESTELLE TURPIN,† A L E X A N D R E T O M A S , * ,† CHRISTA FITTSCHEN,‡ PASCAL DEVOLDER,‡ AND JEAN-CLAUDE GALLOO† De´partement Chimie et Environnement, Ecole des Mines de Douai, 941 rue Charles Bourseul, B.P. 10838, F-59508 Douai, Cedex, France, and Physico-Chimie des Processus de Combustion et de l’Atmosphe`re (PC2A), UMR CNRS 8522, Universite´ des Sciences et Technologies de Lille, F-59655 Villeneuve d’Ascq, Cedex, France
Simulation chamber experiments have been carried out at room temperature to investigate the products of the acetone + OH and acetone-d6 + OH reactions using two photoreactors made of Teflon or Pyrex and coupled to GCFTIR-FID analytical techniques. In the Pyrex chamber, the results demonstrated that the channel forming acetic acid is a minor oxidation route in the atmospheric acetoneh6 + OH reaction (yield 0.90
(0.8-1.85) × 1015 (0.05-5.0) × 1015 (8.0-25.1) × 1015 (3.2-7.9) × 1015
0.5 (He) 0.92 (He) 735-750 (He/O2) 760 (air)
>0.99 0.8-1 0.95-0.97 0.86 >0.95
Acetone-d6 + OH O3 photolysis MS (8.0-25.1) × 1015 CH3ONO photolysis GC-FTIR (1.2-2.5) × 1015
735-750 (He/O2) 760 (air)
H2 radiolysis CF4 radiolysis O3 photolysis CH3ONO photolysis
CIMS LIF MS GC-FTIR
LIF: laser induced fluorescence; MS: mass spectrometry; CIMS: chemical ionization mass spectrometry.
our experimental study focused on the acetic acid formation in the reaction of acetone with OH radical under various conditions.
Experimental Section Environmental Chambers. Two different environmental chambers have been used: (i) a collapsible Teflon chamber (FEP 50 µm thickness) with a volume of about 300 L and (ii) a 35 L cylindrical Pyrex chamber. Both setups operate at room temperature and atmospheric pressure. The chambers are placed inside a wooden box equipped with an irradiation system and a temperature regulation device, the latter consisting of two fans fixed at the top of the box and flushing laboratory air through it. The internal walls of the box are covered with aluminum sheets to homogenize direct and reflected light emitted by the fluorescent tubes. Six fluorescent lamps Philips TMX 200 LS emitting in the range 300-460 nm (λmax ) 365 nm), disposed on two parallel faces of the box, are used to irradiate the chambers. The OH radicals were generally produced by photolysis of methylnitrite (CH3ONO) in the presence of NO:
CH3ONO + hν f CH3O + NO
CH3O + O2 f HCHO + HO2
HO2 + NO f OH + NO2
Acetone was introduced in the Teflon chamber using a gently heated vacuum line. In the case of the Pyrex photoreactor, acetone was directly injected with a syringe into the chamber evacuated to a pressure of 100 mbar. Initial acetone concentrations were in the range of 130-320 ppm (298 K, 1 atm). The chamber was filled with purified air, and a few tenths cm3 of gaseous CH3ONO and NO were finally introduced in the chamber. Initial CH3ONO and NO concentrations were around 60 and 2 ppmv, respectively. Excess NO was added to the reaction mixture to ensure the conversion of HO2 to OH radicals through reaction 4. The gaseous mixture was left in the dark for 1 h to achieve a good homogeneity of the reactants. Methylnitrite was synthesized separately by slowly adding diluted H2SO4 to a mixture of NaNO2 and methanol using the method described by Taylor et al. (25). Acetone, acetic acid, acetone-d6, and acetic acid-d3 (Acros Organics with purity higher than 99%) were used without further purification. NO (5% in N2, Air Liquide) was used directly from the cylinder; purified air was produced by a zero air generator (Air Generator Claind 2301 HG).
Sampling and Analysis Method. Products were analyzed by gas chromatography coupled to two different inline techniques: Fourier transform infrared spectroscopy (FTIR), particularly suited for the detection of acetic acid, followed by flame ionization detection (FID), particularly suited for the detection of acetone. A 20 cm3 gas sample loop connected to a six-port Valco valve was used to sample sequences of 40 cm3 gas aliquots from the simulation chamber; preconcentration of the sample was obtained using a TCT thermodesorption device (Chrompack). This allowed a subsequent flash injection of the organic compounds into the chromatographic system (Varian 3300). IR spectra were collected every 1.2 s with a resolution of 16 cm-1 by a Nicolet 550 spectrometer and integrated by the OMNIC software between 600 and 4000 cm-1. The obtained so-called Gram-Schmidt chromatograms can further be treated to construct the chromatogram resulting from the spectral integration over the 1600-1900 cm-1 band presenting characteristic features of acetic acid. This method allows more precise measurements of acetic acid concentrations. Several samplings of the reaction mixture were carried out after filling the chamber in order to determine precisely the initial acetone concentration. Then the photoreactor was irradiated up to 4 h with successive samplings performed at regular intervals (about 30 min). At the end of each experiment, the chamber was evacuated, then flushed several times with zero air, and finally cleaned using irradiation over night. Test experiments were conducted to check for the stability of pure acetone and acetic acid/air mixtures in our simulation chambers, as these compounds could be either photolyzed or adsorbed on the walls during the experiment. For acetone, the variation was below 1% over 5 h. For acetic acid, pseudofirst-order decay rates of 5 × 10-6 s-1 and 3 × 10-6 s-1 were observed in the Teflon chamber and in the Pyrex chamber, respectively. On the basis of these results, losses of acetone and acetic acid to the walls were considered to be negligible within the time scale of the experiments. Calibration Curves and Acetic Acid Detection Limit. Quantification of acetone-h6 (respectively acetone-d6) and acetic acid (respectively acetic acid-d3) concentrations in the gas phase was performed by frequent calibration of the IR and FID detectors. Standard solutions of acetone in n-octane and acetic acid in ethanol were gravimetrically prepared in the range 0.7-15 mg cm-3 and 0.05-0.6 mg cm-3, respectively. About 1 µL of these solutions was injected via the split/splitless injector of the gas chromatograph set in the splitless mode. The chromatographic peak area is then plotted versus the mass of the target compound in the injected sample. A good linearity has been obtained for both deuterated VOL. 40, NO. 19, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
FIGURE 1. Chromatograms resulting from the analysis of two gas aliquots sampled at t ) 0 (full line) and t ) 60 min (dotted line) for the reaction acetone + OH investigated in either (a) the Teflon photoreactor or in (b) the Pyrex photoreactor (note the lack of acetic acid).
and nondeuterated compounds. Additional gas injections via the sample loop and the TCT injection system were carried out by sampling from the Teflon chamber filled with known amounts of either acetone or acetic acid. The measured peak areas were consistent with the expected mass of acetone or acetic acid as calculated from the concentration of the compound in the simulation chamber. From the good agreement between the liquid and gas injections, it could be concluded that adsorption of acetone and acetic acid on the surface of the simulation chamber and along the Silcosteel sample line was negligible, confirming previous test experiments. Uncertainties in the quantification of acetone and acetic acid, estimated from repeated sampling from the simulation chamber, were about 6 and 8% (1σ), respectively. The detection limit (DL) of acetic acid was determined by calculating the standard deviation (σ) of 7 successive liquid injections of a solution at 110 µg cm-3 using the Student relation: DL ) t × σ with t the Student coefficient (for 7 injections and 95% confidence interval t ) 2.45). The detection limit was 27 ng, which corresponds to about 270 ppbv 5958
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 19, 2006
of gaseous acetic acid diluted in a gas sample of 40 cm3 sampled from the simulation chamber.
Results and Discussion 1. Experiments in Teflon Bags. Acetic Acid Formation Yield. A series of 7 experiments was carried out in Teflon bags. A typical chromatogram obtained before and after a 60-min irradiation period is shown in Figure 1a. Acetic acid was clearly identified by its retention time (7.2 min) and its IR spectrum. Additional GC analysis with mass spectrometric detection confirmed the formation of acetic acid. It should be noted that test experiments without acetone (i.e., methylnitrite in zero air with UV light) or without UV light irradiation (i.e., acetone with methylnitrite in air in the dark) did not reveal any formation of acetic acid. Therefore it appears that both OH and acetone are required for this formation. In addition, the high NOx (NO + NO2) concentrations in the reaction system preclude any acetic acid formation from peroxy radical reactions like CH3C(O)O2 + HO2, since all peroxy radicals are scavenged by their fast reactions with NO (26). Apart from
FIGURE 2. Plot of acetic acid formed as a function of acetone-h6 reacted in the Teflon chamber; the linear regression leads to a yield for acetic acid of (0.14 ( 0.03) (see text). the initial reactants (acetone and methylnitrite), chromatograms exhibit the presence of formaldehyde, methylnitrate, and formic acid, usual end products of the OH source. The quantification of the reacted acetone and the produced acetic acid was based on the calibration curves. In general, 10% of the initial acetone concentration has reacted at the end of the experiment. In typical concentration-time profiles, acetic acid concentrations usually reached a maximum after 60 or 90 min. The subsequent decrease of acetic acid (while acetone is still consumed) cannot be explained by the gas-phase reaction CH3C(O)OH + OH, since the corresponding pseudo-first-order rate of about 100 s-1 (for a maximum acetic acid concentration of 5 ppmv and a rate constant kCH3C(O)OH+OH ) 8 × 10-13 cm3 molecule-1 s-1 at 298 K (27)) is 5 times lower than the CH3C(O)CH3 + OH one under our typical conditions. We suspect that other unknown reactions in the gas phase or on the Teflon walls result in consumption of gaseous acetic acid. To derive the yield of acetic acid, we have thus only taken into account the data of the first two or three samplings, i.e., before 60 or 90 min of reaction time: R ) [acetic acid]formed/ [acetone]consumed ) 1 - R. The results of all experiments are reported in Figure 2. The rather large scatter of the data is due to a poor reproducibility between the experiments. A linear regression on these data leads to an acetic acid yield of (14 ( 4)%, corresponding to a branching ratio of (86 ( 4)% for the abstraction channel. The quoted uncertainty corresponds only to the 2σ statistical errors from the linear regression analysis at 95% confidence interval. However, our result was not entirely satisfying, as acetic acid had never been detected except in the last reaction chamber study (21). It should be stressed that the earlier simulation chamber study used Pyrex or stainless steel wall photoreactors (17), while the recent study (21) used a quartz reactor. Thus, we decided to further investigate the acetone + OH reaction in two ways: first by changing the OH radical source and second by repeating the above experiments in a Pyrex chamber. Further Tests with Another OH Source. Since the presence of acetic acid is still clearly controversial, we have performed supplementary experiments using a quite different OH source. In this set of experiments, OH radicals were produced via UV irradiation of CH3C(O)CH3/Cl2/NO/air mixtures in the Teflon chamber. The Cl-initiated oxidation of acetone is known to lead stoichiometrically (∼99%) to the MV peroxy radical CH3C(O)CH2O2 (28):
CH3C(O)CH3 + Cl f CH3C(O)CH2 + HCl
CH3C(O)CH2 + O2 f CH3C(O)CH2O2
OH radicals are formed in subsequent very fast reactions implying HO2 and NO. NO was added in order to avoid the formation of acetic acid through the CH3C(O)O2 + HO2
FIGURE 3. Concentration-time profiles of acetone-h6 (circles), 2-methylpropene (triangles), and acetic acid (squares) in the Teflon chamber. Filled symbols correspond to the reaction acetone-h6 + OH in the presence of 2-methylpropene and open symbols to the reaction without 2-methylpropene. 2-Methylpropene values were multiplied by a factor of 10 for clarity. reaction (26). The OH radicals were then allowed to react with the acetone in excess. In some experiments, 2-methylpropene CH3C(CH3)dCH2 was added to the reaction mixture to act as scavenger of OH radicals. By comparing the ratios kCH3C(O)CH3+Cl/kCH3C(O)CH3+OH (≈14.1 (13, 28)) and kCH3C(CH3)dCH2+Cl/k(CH3C(CH3)dCH2+OH (≈5.9 (13, 29, 30)), it proves possible to choose appropriate acetone and 2-methylpropene concentrations (about 200 and 10 ppmv, respectively) so that OH radicals react mostly with 2-methylpropene. As a consequence, this ensured that no OH radicals will react with acetone. A typical experiment is shown in Figure 3: in experiments without 2-methylpropene, acetic acid is detected immediately. Conversely, when 2-methylpropene was initially added, no acetic acid is formed as long as the 2-methylpropene concentration is sufficiently high to effectively scavenge OH radicals. These experiments demonstrate that OH radicals are essential in the formation of acetic acid and confirm our experiments using methylnitrite as OH radical source. 2. Experiments in the Pyrex Chamber. Three experiments were carried out in the Pyrex chamber. The experimental conditions (reactant concentrations, pressure, temperature) were similar to those in the Teflon chamber experiments. Two chromatograms are presented in Figure 1b; the first (full line) corresponds to a sampling carried out before turning on the lamps, and the second (dotted line) corresponds to a sampling carried out 60 min after, i.e., when about 10% of the initial acetone has reacted. In contrast to the experiments in the Teflon simulation chamber, no acetic acid was detected. Taking into account our detection limit, we estimate an upper limit of 5% for the channel (1b). These results are very consistent with the investigations of Tyndall et al. (17) achieved in Pyrex or stainless steel simulation chambers, where no acetic acid was observed and an upper limit to acetic acid production of 10% was claimed. Furthermore, they agree much better with recent measurements (16, 18), where acetic acid was searched for but not detected, and with theoretical studies of He´non et al. (23), Canneaux et al. (24), and the very recent work of Caralp et al. (31). The recent chamber study of Raff et al. (21) where acetic acid formation was observed, concludes to a branching ratio of approximately 3% at room temperature, consistent with our detection limit. 3. Experiments with Acetone-d6. Complementary experiments have been performed using the fully-deuterated acetone compound CD3C(O)CD3 with the aim at confirming the present results and providing a better insight into the chemical mechanism. VOL. 40, NO. 19, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
FIGURE 4. Plot of acetic acid-d3 formed as a function of acetone-d6 reacted in the Pyrex chamber; the linear regression leads to a yield for acetic acid-d3 of (0.20 ( 0.06) (see text).
CD3C(O)CD3 + OH f CD3C(O)CD2 + HDO
CD3C(O)CD3 + OH f ... f CD3 + CD3C(O)OH
Experiments were carried out either in the Teflon chamber or in the Pyrex one. Experimental conditions and analysis procedures matched those employed previously for acetoneh6, with acetone-d6 concentrations between 50 and 100 ppmv. As a first goal, relative rate experiments in the Teflon chamber using methyl formate as the reference compound (kref(HC(O)OCH3 + OH) ) 1.87 × 10-13 cm3 molecule-1 s-1 at 298 K (32)) allowed us to determine the rate constant k7. The rate constant ratio of k7/kref ) 0.184 translates into the following value k7 ) 3.45 × 10-14 cm3 molecule-1 s-1 for the CD3C(O)CD3 + OH rate constant, in good agreement with the literature (11, 12, 21, 33, 34). As a second goal, we tried to detect reaction products in the photolysis of CD3C(O)CD3/CH3ONO/air mixtures. Teflon chamber experiments clearly show the presence of acetic acid-d3 (identified by its IR spectrum), yet with very poor reproducibility between the experiments, as observed previously with CH3C(O)CH3; no consistent acetic acid-d3 yield has been obtained from this set of experiments. Conversely, experiments in the Pyrex photoreactor gave much more reproducible results; acetic acid-d3 was clearly identified by its IR spectrum; a linear regression of acetic acid-d3 versus acetone-d6 reacted (see Figure 4) leads to an acetic acid-d3 yield of (20 ( 6)%, corresponding to a branching ratio of (80 ( 6)% for the abstraction channel (channel (7a)). This result is not surprising since the barrier for D-abstraction is usually higher than that for H-abstraction (33-36). In addition, the measured acetic acid-d3 yield compares very well with the only (to our knowledge) investigation of the products of the CD3C(O)CD3 + OH reaction from Raff et al. (21), who determined an acetic acid-d3 yield of 18.7% at 298 K. Comparing the rate coefficient k7b ) 0.2 × 3.45 × 10-14 ≈ 7 × 10-15 cm3 molecule-1 s-1 for the formation of acetic acid-d3 to the upper limit of the formation of acetic acid, k1b ) 0.05 × 1.7 × 10-13 ≈ 8.5 × 10-15 cm3 molecule-1 s-1, it is noteworthy that both values match fairly well. This would indicate the absence of any isotope effect, which is consistent with an addition-elimination channel. The reason why acetic acid is observed in the Teflon bag experiments is presently not clear. The set of experiments performed using chlorine photolysis confirms that the pres5960
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 19, 2006
ence of OH radicals is absolutely necessary to observe acetic acid formation. Second, no acetic acid was observed in the Pyrex simulation chamber experiments, in contrast to those performed in the Teflon one. We tentatively suggest that this formation could be attributed to specific heterogeneous reactions of OH radicals with acetone adsorbed on the wall of the Teflon reactor. The rather large scatter of the data (see Figure 2) is consistent with such an uncontrolled heterogeneous process. It should be stressed that such catalytic effects of Teflon films have already been observed in previous smog chamber studies (37-39). In addition, the transformation of acetone into acetic acid was also shown to occur under photocatalytic oxidation conditions implying a chemisorbed propylene oxide (40, 41). Additional experiments carried out in much larger Teflon simulation chambers would have been very useful to confirm this assumption. These experiments are an example of wall-effects in Teflon simulation chambers. Our results also raise the question whether the acetic acid formation observed by Raff et al. (21) might be assigned to such heterogeneous reactions in their small quartz reactor. The increase of the yield with decreasing temperature observed by these authors (21) might also be related to the well-known increase of the efficiency of wall reactions at low temperatures, especially in the presence of polar molecules. Finally, our results obtained in the Pyrex chamber demonstrated that the channel forming acetic acid is a minor oxidation route in the atmospheric acetone + OH reaction (yield