A new source of methyl glyoxal in the aqueous phase

Introduction Conclusions References


Introduction
In the last decades, carbonyl compounds have been a subject of intense research due to their ubiquitous abundance and their effect on atmospheric chemistry and human health.They are emitted directly from biogenic and anthropogenic sources or formed through the oxidation of hydrocarbons (e.g., Atkinson, 1997;Matthews and Howell, Figures Back Close Full 1981; Lipari et al., 1984;Ciccioli et al., 1993;Mopper and Stahovec, 1986;Carlier et al., 1986;Hallquist et al., 2009).One carbonyl compound that is emitted from numerous and mainly biological sources is methyl ethyl ketone (MEK).It is released from grass, clover (Kirstine et al., 1998;de Gouw et al., 1999), different types of forests, and biomass burning processes (Khalil and Rasmussen, 1992;Warneke et al., 1999;Isidorov et al., 1985).Anthropogenic emissions are also important MEK sources, such as artificial biomass burning (Andreae and Merlet, 2001;Akagi et al., 2011;Yokelson et al., 2013;Brilli et al., 2014) and tobacco smoke (Buyske et al., 1956;Yokelson et al., 2013).In addition, MEK is emitted into the atmosphere through the application as solvent for the production of glue, resins, cellulose, rubber, paraffin wax and lacquer (Ware, 1988).Tropospheric MEK gas-phase concentration was found to be in the range of 0.02-15 ppbv, depending on the region (Grosjean et al., 1983(Grosjean et al., , 2002;;Riemer et al., 1998;Singh et al., 2004;Snider and Dawson, 1985;Goldan et al., 1995;Grosjean, 1982;Müller et al., 2005;Feng et al., 2004).Singh et al. (2004) measured a concentration in a remote region of 0.02 ppbv, whereas Grosjean et al. (1983) observed a MEK concentration of 11.3 ppbv in Los Angeles.Brown et al. (1994) concluded that MEK is one of the major volatile organic compounds (VOCs) in indoor air.
In addition to the gas-phase measurements, the concentrations measured in bulk water samples collected at an open station near the Bahamas reached a concentration of < 0.5 nmol L −1 (Zhou and Mopper, 1997).Furthermore, an enrichment of MEK in the surface micro layer was found with concentrations up to 2.28 nmol L −1 (Zhou and Mopper, 1997).MEK was also investigated in ice, fog, and rain samples (Grosjean and Wright, 1983).It was not found in fog but there were traces in rain water.In cloud water, a concentration of up to 650 nmol L −1 was measured.This is supported by van Pinxteren et al. (2005), who measured a concentration of 70 to 300 nmol L −1 in cloud water.These studies concluded that the liquid-phase fraction of MEK is higher than the expected fraction calculated according to the Henry constant.Introduction

Conclusions References
Tables Figures

Back Close
Full The Henry constants at a temperature of 25 • C were found to vary between 7.7 and 21 M atm −1 in numerous studies (Buttery et al., 1969;Snider and Dawson, 1985;Ashworth et al., 1988;Zhou and Mopper, 1990;Morillon et al., 1999;Karl et al., 2002).However, Schütze and Herrmann (2004) estimated the Henry constant to be between 23 and 50 M atm −1 , which is higher than the previous measured values found in the literature.This higher Henry constant supports the conclusion from van Pinxteren et al. (2005) and tends to support the investigation of MEK in the liquid phase as aqSOA precursor compound.AqSOA is formed through the oxidation of organic compounds in the aqueous particle phase and is often related to missing SOA sources.These missing sources are most likely responsible for the huge discrepancies between measured and calculated SOA burden.As model results usually underestimate the SOA burden (Kanakidou et al., 2005;Goldstein and Galbally, 2007), missing SOA sources have to be considered in such models to close this gap (Ervens et al., 2011;Herrmann et al., 1999Herrmann et al., , 2015)).
In the present study, the reaction of MEK with OH radicals in water was investigated.
Based on the experimentally obtained data, a reaction mechanism was developed to explain methyl glyoxal formation.The mechanism was included in a COPASI (Complex Pathway Simulator) model and evaluated by comparing the experimentally obtained data and the model results.

Conclusions References
Tables Figures

Back Close
Full and methyl ethyl ketone (99.7 %) were obtained from Riedel-de Haen (Seelze, Germany), and hydrogen peroxide (30 % Suprapur ® ) was obtained from Merck KGaA (Darmstadt, Germany).Ultrapure water was used to prepare the reaction solutions for the bulk reactor experiments and the stock solutions of the authentic standard compounds (Milli-Q gradient A 10, 18.2 MΩ cm −1 , 3 ppb TOC, Millipore, USA).

Bulk reactor experiments
The aqueous-phase oxidation of MEK was conducted in a 300 mL batch reactor using the photolysis of hydrogen peroxide (H 2 O 2 ) as an OH radical source (Set 1).For the experiments, 0.1 mmol L −1 of the precursor compound was mixed with 2 mmol L −1 of H 2 O 2 .The solution was then irradiated at λ = 254 nm for 4 h to continuously generate OH radicals (500 W xenon-mercury lamp, Andover Corporation Optical bandpass filter: L.O.T. -Oriel GmbH & Co. KG, Darmstadt, Germany).To obtain time-resolved data, samples were taken once per hour (0-4 h; number of repetitions n = 3) or in steps of 15 min (n = 1) and analyzed after derivatization using GC/MS.To avoid further reactions of the organics present in the sample with remaining H 2 O 2 , 100 µL of catalase (4 mg mL −1 in water) were added to each sample.Furthermore, a set of blank experiments was conducted to exclude that (a) MEK reacts with H 2 O 2 (Set 2), (b) identified oxidation products originate from photolysis (Set 3), (c) the photolysis of contaminants in the H 2 O 2 solution results in the formation of organics (Set 4), and (d) oxidation of 2,3-butanedione or hydroxyacetone also forms methyl glyoxal (Set 5/6).To determine the photolysis rate constants of 2,3-butanedione, methyl glyoxal, and hydroxyacetone, the photolysis of these products was investigated as well with the setup used (Set 7).
A complete overview of the experiments conducted is given in Table 1.

Sample preparation
Two types of samples were taken over a period of 4 h.For the first type of samples, 60 µL of the reaction mixture were diluted with 2940 µL of water to avoid saturation Introduction

Conclusions References
Tables Figures

Back Close
Full of the GC/MS detector during the quantification of MEK, 2,3-butanedione or hydroxyacetone.For the second type of samples, 3 mL of the reaction mixture were taken and injected without any dilution to enable the identification and quantification of the formed carbonyl compounds.Samples of all sets were derivatized with 300 µL of o- (2,3,4,5,6pentafluorobenzyl)hydroxylamine hydrochloride (PFBHA, 5 mg mL −1 ) at room temperature (Rodigast et al., 2015).Cyclohexanone-2,2,6,6-d4 was used as an internal standard (150 µL, 100 µmol L −1 ), and after 24 h, a pH value of 1 was adjusted by adding hydrochloric acid (37 %) to the reaction mixture.The target compounds were extracted for 30 min with 250 µL of dichloromethane using an orbital shaker (1500 rpm, revolutions per minutes).Finally, 1 µL of the organic phase was used for GC/MS analysis.A 5-point calibration was performed for each chromatographic run using a series of the standard solutions (MEK, methyl glyoxal, 2,3-butanedione, and hydroxyacetone) with concentrations ranging from 1 to 50 µmol L −1 .

Instrumentation
Derivatized carbonyl compounds were analyzed using a GC System (6890 Series Agilent Technologies, Frankfurt, Germany) coupled with an electron ionization quadrupole mass spectrometer in splitless mode at a temperature of 250

Experimental results from bulk reactor
MEK was oxidized with OH radicals, and the decay of MEK was monitored by GC/MS. Figure 1a shows the consumption of the precursor compound MEK.As can be seen, MEK was almost consumed after 180 min of reaction time.From the analysis of the collected samples, 2,3-butanedione, hydroxyacetone, and methyl glyoxal were observed as the most dominating oxidation products.The formation of methyl glyoxal from the oxidation of MEK was unexpected as it has not been reported in the literature before.
Due to the relevance of methyl glyoxal for aqSOA formation, its formation was comprehensively characterized in the present study.
Nevertheless, in Fig. 1b, 2,3-butanedione was found as the main oxidation product of MEK reaching a maximum yield of ≈ 29.5 ± 6.0 % after 60 min.The small variations of the molar yields over the course of the experiment might result from slight temperature changes during the experiment despite the reactor temperature being controlled.However, the curve shape shows that the concentration of 2,3-butanedione starts to decrease after 60 min, indicating further reactions of 2,3-butanedione.This decrease results in a molar yield of 18.9 ± 3.9 % at the end of the experiment (240 min).
A similar trend was observed for hydroxyacetone and methyl glyoxal, as the concentration of methyl glyoxal was the highest after 15 min (≈ 9.5 %) and started to decrease afterwards.In comparison, hydroxyacetone reached the highest concentration after 60 min with a molar yield of 3.0 ± 2.6 %.The standard deviation of the determined molar yields of hydroxyacetone was high, and therefore, only after reaction times of 60 and 120 min could a molar yield of hydroxyacetone be determined.
Methyl glyoxal and hydroxyacetone were completely consumed at the end of the experiment.The strong decrease of the concentrations of the detected carbonyl compounds might result from the reaction with OH radicals and/or from photolysis.Both mechanisms are most likely, as it has been demonstrated that the detected carbonyl compounds react quickly with OH radicals (Lilie et al., 1968;Gligorovski and Herrmann, Introduction Conclusions References Tables Figures

Back Close
Full 2004; Doussin and Monod, 2013;Monod et al., 2005;Ervens et al., 2003;Herrmann et al., 2005;Tan et al., 2010;Stefan and Bolton, 1999) and they are prone to photolysis (Faust et al., 1997;Tan et al., 2010).The photolysis rate constants of the detected carbonyl compounds were determined in the present study because of the dependency on the setup used (Set 7; see Supplement S2 for more details).Methyl glyoxal and hydroxyacetone showed higher photolysis rate constants of k p = 3 × 10 −5 s −1 and ).This indicates a faster decomposition of methyl glyoxal and hydroxyacetone and confirms the experimental results.Furthermore, during the photolysis of 2,3-butanedione and hydroxyacetone, methyl glyoxal was found with molar yields of ≈ 17 and ≈ 19 %, respectively (see Fig. S2, Supplement S2).Due to the low molar yield of hydroxyacetone during the oxidation of MEK (≈ 3 %) and the slow photolysis rate constant of 2,3-butanedione, these processes are of minor importance for the methyl glyoxal yield.
Since 2,3-butanedione is the main oxidation product, it was necessary to investigate the contribution of 2,3-butanedione to the product distribution, especially for the formation of methyl glyoxal.In the oxidation of 2,3-butanedione (Set 5), no methyl glyoxal was detected in the GC/MS chromatogram over a reaction period of 240 min (Fig. 2).Introduction

Conclusions References
Tables Figures

Back Close
Full Consequently, a contribution of 2,3-butanedione to the methyl glyoxal formation could be excluded.Despite the low molar yield of hydroxyacetone during MEK oxidation, the oxidation of hydroxyacetone was investigated for methyl glyoxal formation as well (Herrmann et al., 2005;Schaefer et al., 2012;Set 6).During the oxidation, a molar yield of 100 % was found after a reaction time of 60 min (see Supplement S3, Fig. S3).After 60 min of reaction time, the molar yield of methyl glyoxal decreases through further reactions, as was observed during MEK oxidation.However, due to the low molar yield of hydroxyacetone (3.0 %), the oxidation has only minor importance for the observed molar yield of methyl glyoxal.
To ensure methyl glyoxal was only formed during the oxidation of MEK, an experiment was conducted to investigate the non-radical reaction of MEK with H 2 O 2 in the dark (Set 2).During this experiment, no decomposition of MEK was observed, excluding the non-radical reaction of MEK with H 2 O 2 as a source of methyl glyoxal.Furthermore, the photolysis of MEK has to be considered as a source for methyl glyoxal.
Hence, the photolysis of MEK was studied at λ = 254 nm (Set 3).As can be seen in Fig. 3, the photolysis of MEK leads to the formation of 2,3-butanedione with a molar yield of ≈ 2.2 % indicating that the photolysis of MEK is an additional source.No further carbonyl compounds were detected from the photolysis of MEK, and thus, the photolysis of MEK can be excluded as a methyl glyoxal source.The blank experiment (Set 4) indicated the absence of methyl glyoxal as well.
In summary, the oxidation of MEK constitutes a source for methyl glyoxal, and due to the high concentration of MEK in cloud water (70-650 nmol L −1 ; Grosjean and Wright, 1983;van Pinxteren et al., 2005), this source is of atmospheric relevance.Due to the molar yields of 9.5 % for methyl glyoxal, 29.5 % for 2,3-butanedione, and 3.0 % for hydroxyacetone, ≈ 42 % of the oxidation products of MEK could be elucidated in the present study with only these carbonyl compounds (Table 2).This highlights the importance of carbonyl compounds for the aqueous phase chemistry.Based on the Introduction

Conclusions References
Tables Figures

Back Close
Full experimental findings, a reaction mechanism was developed to describe the formation of methyl glyoxal (Fig. 4).

Oxidation mechanism and model description
According to the structure of MEK, the OH radical attack can proceed at three different positions (Fig. 4; H-atoms at carbons 1, 3, and 4).For the present study, only the attack at carbons 3 and 4 is considered because these processes lead to the formation of the observed products (Fig. 4).Note that the abstraction of a hydrogen atom at carbon 3 leads to a secondary alkyl radical (A), whereas at the terminal carbon, a primary alkyl radical is formed (B).The branching ratios for the formation of the primary and secondary alkyl radicals will be discussed in detail in the next section (Sect.3.2.1).
The primary and secondary alkyl radicals react rapidly with oxygen to form alkylperoxy radicals.The alkylperoxy radical recombines to a tetroxide and reacts further in three different ways, including the formation of an carbonyl compound and an alcohol (i), the formation of two carbonyl compounds and H 2 O 2 (ii), and the decomposition into an alkoxy radical (iii) (von Sonntag and Schuchmann, 1991).The decomposition of the tetroxide into a peroxide and oxygen was not considered further due to the minor importance of this process (von Sonntag and Schuchmann, 1997).The secondary and primary alkylperoxy radicals can react with HO 2 , forming organic hydroperoxides.The organic peroxides react with OH radicals or photolyze, resulting in the formation of an alkoxy radical or tetroxide that can react further, as described before through pathways (i)-(iii).The described mechanism was included in a COPASI model to examine the developed oxidation mechanism, the decomposition of the precursor compound, and the formation of the observed products.Table 3 shows the considered reactions, the rate constants, and their references.Only the reactions leading to the formation of the products identified are discussed in detail.Surprisingly, the products 3-oxobutanal and hydroxybutanone were not observed during the experiments.Since 3-oxobutanal contains two carbonyl groups, two derivatives can be formed, including one (mass to charge ratio m/z 281; M •+ ) or two (m/z 476; M •+ ) derivatized groups.Hydroxybu-Introduction

Conclusions References
Tables Figures

Back Close
Full tanone has one carbonyl group with a derivatized m/z 283 (M •+ ).As can be seen in Fig. 5, m/z 283 was not found in the extracted ion chromatogram (EIC).Furthermore, m/z 281 and 476 can be detected in the EIC of the bulk reactor samples, but they can also be found in the EIC of the authentic standard compounds.The sample of the authentic standard compounds does not contain hydroxybutanone and 3-oxobutanal.
According to the comparison of the EIC of the bulk reactor samples and the authentic standard, it can be assumed that m/z 281 and 476 are fragments of the internal standard and 2,3-butandione.No additional signals in the EIC of the samples were detected at m/z 281, 283, and 476, and thus, 3-oxobutanal and hydroxybutanone are not formed during the experiment.Hence, their formation pathways were excluded from the model.

HO x chemistry and OH radical attack
The relevant reactions for HO x chemistry according to R1-R6 are included in the mechanism.OH radicals were formed through the photolysis of H 2 O 2 at λ = 254 nm with measured photolysis rate constants of k p = 7.6 × 10 −6 s −1 (R1, see Supplement S1).
The formed OH radicals react further with MEK in a first oxidation step, leading to the formation of a primary (R7) and secondary alkyl radical (R8).The rate constants were postulated by Herrmann et al. (2005) to be k = 1.17 × 10 8 and k = 1.3 × 10 9 M −1 s −1 for the formation of the primary and secondary alkyl radicals, respectively.This is in good agreement with the model study by Sebbar et al., who postulated that the H-atom abstraction most likely proceeds at the secondary carbon of MEK (Sebbar et al., 2014(Sebbar et al., , 2011) ) due to the lower C-H bond dissociation energy (≈ 377 kJ mol −1 , C4) in comparison to the primary carbon atom (≈ 423 kJ mol −1 , C3).In the present study, a rate constant k = 1.5 × 10 9 M −1 s −1 (Gligorovski and Herrmann, 2004) was used, and the branching ratio was varied from 60/40 to 10/90 for the primary/secondary H-atom abstraction.Introduction

Conclusions References
Tables Figures

Back Close
Full The results are shown in Fig. 6 and discussed based on the molar yields of the products.As can be seen for 2,3-butanedione (Fig. 6a), a branching ratio of 60 % for the primary H-atom abstraction and 40 % for the secondary H-atom abstraction leads to lower molar yields, whereas the molar yields start to increase with an increasing fraction of secondary H-atom abstraction.According to the mechanism (Fig. 4), 2,3butanedione is only formed via secondary H-atom abstraction, and thus, it is feasible to reach higher molar yields with a higher fraction of secondary H-atom abstraction.However, with an increasing secondary H-atom abstraction, the experimentally determined concentration was increasingly overestimated, especially at the beginning of the experiment.After 60 min of reaction time, the highest experimentally determined molar yield (29.5 ± 6.0 %) was overestimated by a factor of ≈ 2 with a ratio of 10/90 for primary/secondary H-atom abstraction (molar yield ≈ 49.9 %), whereas a ratio of 60/40 resulted in reasonably good agreement (molar yield ≈ 23.7 %).Based on this, a branching ratio of 60/40 for primary/secondary H-atom abstraction was used in the present model.
In contrast, methyl glyoxal molar yields were increasingly underestimated with an increasing fraction of secondary H-atom abstraction (Fig. 6b).Thus, after 15 min of reaction time, the experimental molar yield (≈ 9.5 %) was underestimated by a factor of ≈ 5 with a higher fraction of the secondary H-atom abstraction (molar yield ≈ 2 %).In comparison with a ratio of 60/40 for primary/secondary H-atom abstraction, a molar yield of ≈ 11.4 % was observed, which is in good agreement with the experiment.Overall, the branching ratio of 60/40 (primary/secondary H-atom abstraction) resulted in reasonable agreement for 2,3-butanedione and in an excellent conformity to the methyl glyoxal molar yields.This shows the importance of the primary H-atom abstraction as the main decomposition pathway of MEK and thus for methyl glyoxal formation.
The primary and secondary alkyl radicals react further with oxygen (R9/R21) with a rate constant of k = 3.1 × 10 9 M −1 s −1 (Zegota et al., 1986;Glowa et al., 2000), which was reported for the formation of acetonylperoxy radicals.The rate constants of the acetonylperoxy radical were used for the formation of the primary alkylperoxy radical Introduction

Conclusions References
Tables Figures

Back Close
Full (R9) and further reactions of the alkylperoxy radical because of their structural similarity.They were also applied by Glowa et al. (2000) for the formation of the secondary alkylperoxy radical of MEK.Further reactions of the primary and secondary alkylperoxy radicals resulting in different oxidation products will be discussed in detail in the following section (Sect.3.2.2 and 3.2.3).
In addition to the discussed pathway, the primary alkylperoxy radical has the opportunity to react with HO 2 instead of the recombination, forming an organic hydroperoxide (R14).The rate constant was reported by von Sonntag and Schuchmann (1991) to be from 10 7 up to 10 9 M −1 s −1 .For the primary alkylperoxy radical, k = 1×10 7 M −1 s −1 was used as it led to the best agreement of the experimental data and the model results.
The formed peroxide can react with OH radicals (R15) or photolyze (R16).The same rate constant (k = 2.7 × 10 7 M −1 s −1 ) as that for the OH radical oxidation of H 2 O 2 was used for R15 due to a lack of literature data.According to the assumption by Monod et al. (2007), the rate constant for the photolysis of the formed peroxide (R16) was defined to be 1/10 of the photolysis rate constants of hydrogen peroxide.Thus, a photolysis rate constant of k p = 7.6×10 −7 s −1 was included in the model.This was used as a first approximation due to a lack of literature data for the formed hydroperoxides.The 31903 Introduction

Conclusions References
Tables Figures

Further reactions and photolysis of formed oxidation products and MEK
The products 2,3-butanedione, methyl glyoxal, and hydroxyacetone positively identified by GC/MS analysis might also react further, forming a variety of oxidation products.The rate constant of methyl glyoxal with OH radicals is given in the range of k = 5.3 × 10 8 to Monod et al., 2005;Ervens et al., 2003;Herrmann et al., 2005; Introduction

Conclusions References
Tables Figures

Back Close
Full   et al., 2010), whereas a rate constant of k = 5.3 × 10 8 M −1 s −1 (R29; Monod et al., 2005) leads to the best agreement between the experimental data and the model results.In comparison, the rate constants determined by Ervens et al. (2003), Herrmann et al. (2005), and Tan et al. (2010) resulted in an underestimation of the molar yield.
Methyl glyoxal also has the opportunity to photolyze.The photolysis is included with 2,3-Butanedione is also prone to OH radical oxidation and photolysis.As discussed, a huge discrepancy exists in the rate constants for the reaction of OH radicals with 2,3butanedione (Lilie et al., 1968;Doussin and Monod, 2013;Gligorovski and Herrmann, 2004).In the present study, the value of k = 1.4 × 10 8 M −1 s −1 (R31) determined by Gligorovski and Herrmann (2004) was used.The higher rate constants reported by Lilie et al. (1968) and Doussin and Monod (2013) resulted in consumption that was too fast.The photolysis was included with a rate constant of k p = 9 × 10 −6 s −1 in the COPASI model (R32).It should be mentioned that methyl glyoxal was formed during the photolysis of 2,3-butanedione.However, the photolysis was too small to contribute significantly to methyl glyoxal formation.For more details, see Supplement S2.
The oxidation of hydroxyacetone with OH radicals was also considered in the model study with a rate constant of k = 8×10 8 M −1 s −1 (R33; Stefan and Bolton, 1999).During the experiment, methyl glyoxal was formed with 100 % molar yield.Thus, the reaction of hydroxyacetone to methyl glyoxal was included in the model study.The photolysis of hydroxyacetone was measured with a rate of k p = 2 × 10 −5 s −1 , leading to methyl glyoxal with a molar yield of ≈ 19 % (R34).The photolysis rate constant of MEK was measured as k p = 5 × 10 −5 s −1 (see Supplement S2, Fig. S1).During the experiment, 2,3-butanedione was found with a molar yield of 2.2 %.Thus, the reaction of MEK leading to 2,3-butanedione was included in the model study (R35).Introduction

Conclusions References
Tables Figures

Back Close
Full

Model results and comparison with the experimental dataset
The described reactions are included in a model, and the decomposition of MEK and molar yields of formed products were compared to the experimentally obtained data (Fig. 7).The model was not validated with the time course of hydroxyacetone due to the high standard deviation of the experimental results.The comparison of the model study and the experiment showed very good agreement for the consumption of MEK (Fig. 7a).There is also good agreement for the formation of methyl glyoxal and limited agreement for the molar yields of 2,3-butanedione.The initial high molar yield of 2,3butanedione is reflected well (Fig. 7b).Thus, after 60 min of reaction time, molar yields of 23.7 % in the model study and 29.5 ± 6.0 % in the experiment were reached.Under consideration of the standard deviation, this is in good agreement with the COPASI model.
The determined molar yields up to a reaction time of 120 min showed very good conformity with the experiment (18.2 % and 22.1 ± 9.0 %).The temporal behavior shows that the determined molar yield in the model study is somewhat lower towards the end of the experiment.Hence, it is possible that there are other reaction pathways that lead to higher molar yields of 2,3-butanedione at the end of the experiment, which are not considered in the model study as of now.Good agreement between the model results and experiment was observed for methyl glyoxal (Fig. 7c).The curve shapes are very similar, and hence, fast formation and decomposition of methyl glyoxal were found in the model study and in the experiment.After 15 min of reaction time, a molar yield of 9.5 % was found in the experiment.In comparison, the model study resulted in a molar yield of 11.4 %, which is in good agreement with the experiment and validates the COPASI model.

Summary
In the present study, MEK was identified as a new source for methyl glyoxal in the aqueous phase.It was demonstrated that methyl glyoxal originates directly from MEK oxidation and not from side reactions such as photolysis or non-radical reactions.A molar yield of ≈ 9.5 % was determined during the oxidation.Based on the experimental results, a reaction mechanism could be developed.The calculations with a COPASI model supported the experimental results and confirm MEK as a precursor compound for methyl glyoxal in aqueous medium.
Further carbonyl compounds could be identified and quantified.2,3-Butanedione was found as the main oxidation product (molar yield ≈ 29.5 %) and was formed during the photolysis of MEK as well.As a further oxidation product, hydroxyacetone was identified and was formed with a molar yield of ≈ 3.0 % during the oxidation of MEK.
The oxidation mechanism of MEK in aqueous solution was elucidated, and MEK was demonstrated to be a precursor compound for methyl glyoxal in the aqueous phase.Regarding the important role of methyl glyoxal for the aqSOA formation, MEK has to be considered for aqSOA as well, which could be a next step in reducing the underestimation of the SOA burden by model studies.Full  Full

The
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | • C (Agilent 5973 Network mass selective detector, Frankfurt, Germany).They were separated with an HP-5MS UI column (Agilent J & W GC columns, 30 m × 0.25 mm × 0.25 µm) using the following temperature program: 50 • C isothermal for 2 min and elevated to 230 • C with 10 • C min −1 .The temperature of 230 • C was held for 1 min, and the temperature gradient ended with 320• C, which was held constant for 10 min.
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Supplement related to this article is available online at doi:10.5194/acpd-15-31891-2015-supplement.Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Table 1 .
Conducted experiments in the bulk reactor.

Table 2 .
Maximal molar yields of the oxidation products.

Table 3 .
Reaction mechanism and rate constants for the modelling of the experiment with COPASI.