OH-Initiated Oxidation of Acetylacetone: Implications For

Article

Cite This: Environ. Sci. Technol. 2018, 52, 11169−11177 pubs.acs.org/est

OH-Initiated Oxidation of Acetylacetone: Implications for Ozone and Secondary Organic Aerosol Formation † ∥ ‡ ∥ † § † † † Yuemeng Ji, , Jun Zheng, , Dandan Qin, Yixin Li, Yanpeng Gao, Meijing Yao, Xingyu Chen, † † § Guiying Li, Taicheng An,*, and Renyi Zhang*, † Guangzhou Key Laboratory of Environmental Catalysis and Pollution Control, School of Environmental Science and Engineering, Institute of Environmental Health and Pollution Control, Guangdong University of Technology, Guangzhou 510006, P. R. China ‡ Collaborative Innovation Center of Atmospheric Environment and Equipment Technology, Nanjing University of Information Science & Technology, Nanjing 210044, P. R. China § Department of Atmospheric Sciences and Department of Chemistry, Texas A&M University, College Station, Texas 77843, United States

*S Supporting Information

ABSTRACT: Acetylacetone (AcAc) is a common atmospheric oxygenated volatile organic compound due to broad industrial applications, but its atmospheric oxidation mechanism is not fully understood. We investigate the mechanism, kinetics, and atmospheric fate of the OH-initiated oxidation for the enolic and ketonic isomers of AcAc using quantum chemical and kinetic rate calculations. OH addition to enol- AcAc is more favorable than addition to keto-AcAc, with the total rate constant of 1.69 × 10−13 exp(1935/T) cm3 molecule−1 s−1 over the temperature range of 200−310 K. For the reaction of the enol-AcAc with OH, the activation energies of H-abstraction are at least 4 kcal mol−1 higher than those of OH-addition, and the rate constants for OH-addition are by 2−3 orders of magnitude higher than those for H-abstraction. Oxidation of AcAc is predicted to yield significant amounts of acetic acid and methylglyoxal, larger than those are currently recognized. A lifetime of less than a few hours for AcAc is estimated throughout the tropospheric conditions. In addition, we present field measurements in Beijing and Nanjing, China, showing significant concentrations of AcAc in the two urban locations. Our results reveal that the OH-initiated oxidation of AcAc contributes importantly to ozone and SOA formation under polluted environments.

■ INTRODUCTION Furthermore, emissions of AcAc in developing countries (such as China) are anticipated to be substantially increased because Ketones represent an important class of oxygenated volatile 14,15 organic compounds (OVOCs) and are emitted from natural of their rapid industrialization and economic development. AcAc is a prototype of β-diketone that exists in the two

See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. and anthropogenic sources, e.g., from industrial activities and Downloaded via GUANGDONG UNIV OF TECHNOLOGY on October 17, 2018 at 07:18:05 (UTC). oxidation reactions of both biogenic and anthropogenic tautomeric forms: the enol form contains an intramolecular − hydrocarbons.1 5 The atmospheric oxidation of ketones has hydrogen bond and resonance stabilization through a ﬁ conjugated π-system, whereas the diketo form contains two been identi ed as important sources for HOx radical species 6,7 6,8 carbonyl groups with an ∼140° dihedral angle between the (i.e., OH and HO2) and secondary organic aerosol (SOA), profoundly impacting air quality, human health, and climate. oxygens. The atmospheric oxidation mechanism of AcAc is Acetylacetone (also known as 2,4-pentanedione or AcAc) is complex, including multiple pathways and steps. Oxidation of 13,16−18 highly reactive and broadly used for various industrial AcAc is mainly initiated by the hydroxyl radical (OH). applications. For example, AcAc is an important reagent in An earlier experimental study by Zhou et al. has evaluated the preparation of chelate compounds for a wide range of atmospheric chemistry for AcAc; using a relative kinetic transition metals,9,10 an industrial additive,11 and a building method the authors determined the temperature dependent block for synthesis of heterocyclic compounds and raw rate coeﬃcients over the temperature range of 285−310 K, materials for sulfonamide drugs.12 In developed countries such as Japan, the United States, and Europe, the global Received: July 18, 2018 capacity of AcAc is estimated to be approximately 20 000 t Revised: August 13, 2018 − a 1,13 mainly from industrial activities, while its formation from Accepted: August 30, 2018 in situ atmospheric photochemical production is negligible. Published: August 30, 2018

Figure 1. Optimized geometries of the key stationary points of OH-AcAc reaction at the M06-2X/6-311G(d,p). The bond length is in Å.

− with an Arrhenius expression of k = 3.35 × 10−12 exp[(983 ± has been demonstrated.33 36 Furthermore, no theoretical 130)/T] cm3 molecule−1 s−1.13 That experimental work also results are available on the atmospheric chemistry of AcAc. identified several products from the OH-initiated oxidation of In this work, we have investigated the detailed oxidation AcAc, including methylglyoxal (MG), acetic acid (AA), and mechanism of AcAc with OH employing quantum chemical peroxyacethyl nitrate (PAN).13 On the basis of their measured and kinetic rate calculations within the tropospheric temper- products, a mechanism for the reaction of AcAc with OH has ature range of 200−310 K. We also present field measurements been postulated, involving the initial OH addition to C2 and of AcAc in Nanjing and Beijing, China using proton-transfer 13 C3 positions followed by O2 addition. reaction mass spectrometry (PTR-MS). The atmospheric fate In contrast to investigation on the industrial applications of and oxidation products of AcAc are assessed, and the − AcAc,9 12 limited previous work exists on the atmospheric implications of our results for ozone and SOA formation are oxidation mechanism of AcAc, hindering accurate assessment discussed. of its roles in the formation of ozone and fine particulate matter (PM). For example, MG, organic acids, and PAN have ■ METHODS been identified as the critical species leading to SOA 14,15,19−24 fi The electronic structures and energy calculations were carried formation. Speci cally, organic acids play important out with the Gaussian 09 program suite.37 Geometrical 25,26 − roles in new particle formation and growth and acid base optimization of all stationary points (SPs), such as the 27,28 α reactions, while oligomerization of small -dicarbonyls reactants, transition states (TSs), complexes, intermediates, represents a major source of SOA on the urban, regional, and and products, was performed using the M06-2X level with the 29,30 α global scales. The atmospheric sources of small - 6-311G(d,p) basis set denoted as the M06-2X/6-311G(d,p) dicarbonyls, organic acids, and PAN, however, remain poorly level. The M06-2X functional is a high-nonlocality functional fi 14 quanti ed. The current atmospheric chemical mechanism for with double the amount of nonlocal exchange (2X), with the AcAc oxidation has been proposed on the basis of the reliable performance for the thermochemistry, hydrogen 13 environmental chamber experiment, which has provided bonding, kinetics, and weak interactions.38 In addition, the critical information on the initial kinetic and products of the MPWB1K/, B3LYP/, MP2/and QCISD/6-311G(d,p) levels AcAc oxidation. However, extrapolation of the kinetics and were also employed to optimize the geometry and to validate mechanism of the AcAc reactions from the measured product the convergence of the predicted geometries. The MPWB1K yields is challenging, since a product formation typically and B3LPY methods and the MP2 method represent the involves multiple possible steps and pathways and the product classic density functional theory and the classic Ab Initio is subject to secondary reactions or photolysis. In addition, theory, respectively, while the QCISD method corresponds to there exist additional intricate difficulties using the chamber a higher electronic correlation method. Frequency calculations method. Noticeably, the limitations of the chamber method were carried out at the M06-2X/6-311G(d,p) level to include a long reaction time, higher reactant concentrations, determine all SPs as a real local minima (without any and wall loss.31,32 In particular, the significance of wall loss for imaginary frequency) or a TS (with only one imaginary reactive and condensable species using the chamber method frequency). The evaluation of the vibrational frequencies

11170 DOI: 10.1021/acs.est.8b03972 Environ. Sci. Technol. 2018, 52, 11169−11177 Environmental Science & Technology Article

Figure 2. PES for the OH-initiated reactions of keto- and enol-AcAc (in the unit of kcal mol−1). confirmed that all optimized geometries represented the campus of Nanjing University of Information Science & minima on the potential energy surfaces (Table S1 of the Technology (NUIST) in Nanjing and on the campus of Supporting Information, SI). Intrinsic reaction coordinate Tsinghua University in Beijing. (IRC) calculations were performed to confirm the connection between the TSs and their corresponding reactants and ■ RESULTS AND DISCUSSION 7,39 products. The potential energy surface (PES) was further Initial Reaction of AcAc with OH. There exists an fi re ned by the M06-2X/6-311++G(3df,3pd) level to yield equilibrium between enolipc and ketonic isomers of AcAc more accurate energetics. Because kinetic calculations of the (Figure S2), with the enolic isomer representing the main form organic reaction systems were highly sensitive to the predicted in the gas-phase.44 To systematically assess the OH-initial energetics, single point energy calculations were performed to oxidation of AcAc, we considered both the reactions of enolic refine the PES using the QCISD(T)/6-311+G(2df,p) and and ketonic isomers, denoted by enol-AcAc and keto-AcAc, CCSD(T)/6-311+G(2df,2p) levels. The dual-level approach respectively. The geometries of the two isomers optimized at was denoted as X//Y, where a single-point energy calculation the M06-2X/6-311G(d,p) level are displayed in Figure 1. For at level X was carried out for the geometry optimized at a lower comparison, the geometries and frequencies using other level level Y. In all cases, the energies were calculated relative to the calculations, including MPWB1K/, B3LYP/, MP2/, and Δ corresponding reactants including ZPE corrections. Ea is QCISD/6-311G(d,p), are included in Figure S2 and Table fi Δ − de ned as the activation energy (i.e., Ea = ETS Ereactants), S1. The structural parameters obtained by the five levels are Δ fi Δ − while Er is de ned as the reaction energy ( Er = Eproducts similar, and the largest discrepancies are within 0.9° in the Ereactants). The natural bond orbital (NBO) analysis was carried bond angles and 0.07 Å in the bond lengths. The calculated out to confirm the favorable reaction pathways. All barrierless frequencies at the M06-2X level agree with those obtained by processes were verified by performing the pointwise potential the B3LYP and MP2 levels, with the maximum error of less curve (PPC) scan. Furthermore, the effect of the basis set than 10%. Hence, the M06-2X level theory accurately describes superposition error on the energies was considered using the the geometry optimization and vibrational frequency calcu- counterpoise method described by Boys and Bernardi40 to lations for the AcAc reaction system. Figure 1 shows that enol- evaluate the stability of the complexes. All precomplexes AcAc exhibits a conjugated π-electron character that enhances existed, when BSSE correction was included (see SI). On the the reactivity of the C atoms. As a result, the C atoms in enol- basis of the predicted PES, the kinetics calculations, i.e., the AcAc are more susceptible to attack by OH than those in keto- rate constants and product distributions, were performed using AcAc. Figures 1 and S2 depict the optimized geometries of SPs the Polyrate program41 with the generalized transition-state in the OH-AcAc reaction at the M06-2X/6-311G(d,p) level. theory (more details in SI).7,31,39 The most stable structure for The absolute energies, ZPEs and Cartesian coordinates are also each pathway was used in the kinetic study. included in Table S1. In addition, we measured the concentration of AcAc in The PESs for possible pathways of the OH-initial reaction of Nanjing and Beijing, China on the basis of the proton-transfer enol-AcAc are presented in Figure 2. In addition, the methods 42 + → • + reaction, i.e., H3O + AcAc H2O + AcAc H . The of QCISD(T) and CCSD(T) are performed. As discussion in measurements in Nanjing were made using a high-resolution SI, the M06-2X method is suitable to predict the energies and time-of-flight chemical ionization mass spectrometer (HR- represents a compromise between the computational accuracy ToFCIMS, Aerodyne Res. Inc., U.S.A.),43 while the measure- and efficiency. The reaction for enol-AcAc with OH occurs via ments in Beijing were made using a quadruple mass two distinct pathways (Figure 2): H-abstraction from the two 30 fi spectrometer. Figure S1 shows the high-resolution tof methyl groups (e-Rabs1 and e-Rabs2) and OH-addition to either fi the AcAC peak at m/z = 101.06, con rming the detection of C2 or C3 position (e-Radd1ore-Radd2). For each pathway, a AcAc. In this work, the observation sites were located on the prereactive complex is identified prior to the corresponding TS

11171 DOI: 10.1021/acs.est.8b03972 Environ. Sci. Technol. 2018, 52, 11169−11177 Environmental Science & Technology Article Δ Δ or products. The prereactive complexes are consistently more Considering the Ea and Er values, the reactivity of the H stable than their corresponding reactants (Figure 2). The atom in the two groups is dominantly affected by the steric − ff reaction energies of e-PCadd1 and e-PCadd2 are by 5 kcal e ect rather than its stability, attributable to the presence of −1 − − mol lower than those of the reactants, verifying that the enol- the CH2 group in the middle of AcAc. AcAc oxidation proceeds via the prereactive complex and TS The rate constants of the OH-initial reaction pathways of prior to the product formation. The structures of the keto- and enol-AcAc are calculated and summarized in Table prereactive complexes are similar to those of the reactants, S2, and the temperature dependences of the branching ratios except for the forming bond. For instance, the forming C−O (Γ) are shown in Figure S5. For the reaction of OH with enol- distances are 2.78 and 2.58 Å for the e-PCadd1 and e-PCadd2 AcAc, the rate constants of e-Radd1 and e-Radd2 pathways are (Figure S4a), respectively, while the other bond distances are 3.78 × 10−11 and 7.46 × 10−11 cm3 molecule−1 s−1 at 298 K, similar to those of the reactants. The existence of the respectively, which are by 2−3 orders of magnitude higher 45 prereactive complexes also impacts the reaction kinetics, than those of the two H-abstraction pathways (e-Rabs1 and e- and the OH-AcAc oxidation is expected to exhibit a negative Rabs2) (Table S2). The contribution of the combined H- temperature effect. abstraction pathways to the total rate constant is less than 1%, Δ Figure 2 shows that the activation energies ( Ea) of the two suggesting that the H-abstraction pathway is of minor Γ H-abstraction pathways (e-Rabs1 and e-Rabs2) are 3.04 and 0.89 importance. As shown in Figure S5a, the of e-Radd2is kcal mol−1, respectively, which are at least 4 kcal mol−1 higher greater than 66% in the temperature range of 237−298 K. than the two OH-addition pathways (e-Radd1 and e-Radd2). The Hence, OH addition at C3 position (e-Add-2) is more Δ − −1 Γ calculated reaction energy ( Er)of 28.96 kcal mol for e- favorable, while the value for OH addition at C2 position (e- −1 Radd1 is by about 11 kcal mol lower than that of e-Radd2. The Add-1) is 33% at 298 K. lower exothermicity for e-Radd1 is explained by its structural The rate constants of the OH-addition pathway for keto- characteristics according to the Hammond postulate,46 since AcAc contributes negligibly to the total rate constants (Γ = 0), this pathway proceeds with an earlier TS due to an elongated and the sum Γ of the H-abstraction pathway equals to 1, in − Δ C O distance (Figure 1). However, the Ea value of e-Radd2is contrast to the case of enol-AcAc (Figure S4b). For example, −1 × −13 1.05 kcal mol lower than that of e-Radd1(Figure 2), the rate constants of k-Rabs1 and k-Rabs2 are 4.31 10 and × −14 3 −1 −1 indicating that e-Radd1 is thermodynamically favored but e- 5.86 10 cm molecule s at 298 K, respectively, Radd2 is kinetically favored. The natural bond orbital (NBO) contributing 88% and 12% to the rate constant for keto-AcAc. charges are 0.496 and 0.573 e for the C2 and C3 positions, Our results of dominant H-abstraction for k-Rabs1are respectively. Since the more positive potential bond is easily consistent with those obtained by Holloway et al.16 but are attacked by the nucleophiles (OH), the OH addition to the C3 in contrast with those by Zhou et al.13 The reactivity of the − − − position (e-Radd2) is more favorable than that to the C2 CH2 group is likely overestimated using the structure ff position (e-Radd1). activity relationship, because the impact of the steric e ect is The reaction of keto-AcAc with OH also occurs via two not considered.13 pathways, i.e., H-abstraction from the methyl and methylene The calculated total rate constants (i.e., ktotal, the sum of groups (k-Rabs1 and k-Rabs2) and OH-addition to C4 position calculated rate constants for all pathways) for the OH-AcAc (k-Radd1). OH addition to keto-AcAc proceeds via the reaction are presented in Figure 3, along with comparison with Δ prereactive complex prior to the TS, with a Ea value of the available experimental data. Our derived Arrhenius −1 × −13 3 −1 6.11 kcal mol . The OH-addition to keto-AcAc possesses a expression is ktotal = 1.69 10 exp(1935/T) cm molecule higher barrier (by about 10 kcal mol−1) than those of the H- s−1 over the temperature range of 200−310 K. The rate abstraction pathways; the distinct behaviors between OH constant for the keto-pathway (kketo) is much smaller than that additions to enol- and keto-AcAc are explained because of the of the enol-pathway (kenol). Furthermore, the previous studies conjugative effect. The occurrence of the H-abstraction TSs is further evaluated δ  using the L parameter (L = (C H) ) according to our δ(H O) previous study.47 This parameter not only quantifies whether the TS structure exhibits a product-like (L > 1) or reactant-like (L < 1) character but also reflects whether the pathway is exothermic or endothermic. The L value of 0.34 for k-TSabs1is two times higher than that of k-TSabs2(Figure 1). Hence, the Δ k-Rabs2 pathway with an earlier TS corresponds to a larger Er. Δ − −1 As shown in Figure 2, the Er of k-Rabs2is 26.57 kcal mol , which is by about 4.41 kcal mol−1 more negative than that of k- Δ Rabs1. However, the k-Rabs1 pathway corresponds to a Ea value of 0.56 kcal mol−1, and there exists a hydrogen bond with the distance of 2.02 Å in the TS (Figure 1). To further assess the occurrence for H-abstraction, the dissociation energies 0 − − − − (D298(C H)) of methylene ( CH2 ) and methyl ( CH3) groups are calculated at the M06-2X//M06-2X level. The − 0 − 1 − − − − (D298(C H)) values are 89.77 kcal mol for the CH2 Figure 3. Rate constants (cm3 molecule 1 s 1)ofketo- and enol-AcAc −1 − group and 94.14 kcal mol for the CH3 group, respectively, with OH radical against the temperature. The experimental results − − indicating that the H atom in the CH2 group (k-Rabs2) are (i.e., Expt. 1, 2, and 3) are from Zhou et al., Bell et al., and Holloway − more reactive than that in the CH3 group (k-Rabs1). et al., respectively.

11172 DOI: 10.1021/acs.est.8b03972 Environ. Sci. Technol. 2018, 52, 11169−11177 Environmental Science & Technology Article

Figure 4. PES for (a) the subsequent pathways of e-Add-2 and (b) the competing decomposition and combination with O2 for e-Add-1. The Δ Δ −1 3 −1 −1 number denotes the value of Ea and Er for each reaction step. (unit: kcal mol for energies and cm molecule s for the rate constants).

Figure 5. Schematic representation of the preferred pathways of the OH-AcAc reactions leading to formation of methylglyoxal and acetic acid. have reported AcAc in gas phase exists predominantly in the method provides a reliable description for the initial kinetics of enol-form at room temperature.13,52 Hence, OH addition to the atmospheric oxidation of AcAc. enol-AcAc is favorable than addition to keto-AcAc. As is evident Subsequent Oxidation of the OH-AcAc Adduct. The from Figure 3, our results at the M06-2X//M06-2X level also attack of e-Add-1 and e-Add-2 by O2 yields two peroxy radicals compare favorably with the available experimental data, (RO2-1 and RO2-2), which further undergo NO-addition to considering the respective uncertainties. For example, the form peroxy nitrites (RO2NO-1 and RO2NO-2) followed by − total rate constant at 298 K is calculated to be 11.1 × 10 11 NO2-elimination to form alkoxy radicals (RO1 and RO2). cm3 molecule−1 s−1, consistent with the rate constant of (9.05 Such multistep processes are described as the following (Figure − − − 13 4): ± 1.81) × 10 11 cm3 molecule 1 s 1 measured by Zhou et al. 17 R11or R21 :++ O R12or R22 : NO but somewhat higher than those reported by Bell et al. and ee‐‐Add 1 or ‐‐⎯ Add 2⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ →2 RO ‐ 1 or RO ‐ 2 ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ → 16 22 Holloway et al. Our calculations show a strong negative R13or R23 :− NO ‐‐⎯→⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯2 ‐ ‐ temperature-dependence for the rate constants in the temper- RO22 NO 1 or RO NO 2 RO 1 or RO 2 ature range of 200−310 K, which is attributable to the The vibrational frequencies, absolute energies, ZPEs, and presence of the prereactive complex. Also for comparison, Bell Cartesian coordinates of the relevant species are included in 17 Δ et al. reported report an activation parameter (Ea/R) of 1260 Table S3. As shown in Figure 4, the Ea of pathway R11 is K, comparable to our theoretical value. Clearly, the M06-2X 21.33 kcal mol−1, but the TS of the association of e-Add-2 with

11173 DOI: 10.1021/acs.est.8b03972 Environ. Sci. Technol. 2018, 52, 11169−11177 Environmental Science & Technology Article fi fi O2 (R21) is not identi ed. The PPC was performed to con rm peroxy radicals (ER-RO2 and AR-RO2). The subsequent Δ Δ a barrierless process for R21 (Figure S5). The Er values of reaction of AR-RO2 with NO2 is also barrierless, with a Er − − −1 −1 Δ R11 and R21 are 26.24 and 36.98 kcal mol , respectively. value of 24.76 kcal mol to yield PAN. The Ea value of ER- Δ −1 The large Ea and instability of RO2-1 indicate that its RO2 decomposition to form AA is 6.59 kcal mol . The formation is thermodynamically and kinetically less favorable. pathway from MR to MG corresponds to successively To assess the competition between decomposition and O2 endothermic processes (R25 and R26) but occurs promptly, addition for e-Add-1, we calculated the rate constants for both considering the large exothermicity for the formation of e-Add- × −7 −1 × −26 pathways, with the values of 4.80 10 s and 7.75 10 2 and RO2NO-2. cm3 molecule−1 s−1, respectively. Hence, decomposition of e- It is plausible that there are additional pathways to those Add-1 to enol-AcAc and OH is more favorable than the investigated in our present work, which require further fi combination with O2 (with an equivalent rst-order rate experimental and theoretical studies. Our results indicate that constant of 3.81 × 10−7 s−1). Considering the branching ratios MG and AA represent the most favorable products from the between e-Add-1 and e-Add-2, it is estimated that about 15% of oxidation of AcAc initiated by OH, both arising from the e-Add-1 reacts with O2 to further propagate the oxidation pathways via e-Add-1 and e-Add-2. For comparison, the (Figure 5). previous experimental measurements by Zhou et al. obtained ± ± The association reactions of RO2-1 and RO2-2 with NO the yields of (20.8 4.5)% and (16.9 3.4)% for MG and AA, Δ (R12 and R22) are barrierless and exothermic, with the Er respectively. In addition, we predict a minor pathway via e- values of −22.42 and −21.12 kcal mol−1, respectively. The Add-1 to form PAN, consistent with the work by Zhao et al. produced peroxy nitrites (RO2NO-1 and RO2NO-2) further for a small yield for the formation of PAN (about 2%). undergo NO-elimination via a barrierless process to form the Atmospheric Lifetime of AcAc. The lifetime of VOCs is Δ alkoxy radicals (RO-1 and RO-2), with the Er values of 13.88 a key parameter to assess their roles in the formation of ozone and 9.44 kcal mol−1, respectively. According to previous and SOA. The OH initiated oxidation represents the dominant studies,48,49 there are three primary reactions for the alkoxy daytime mechanism in regulating the lifetimes of VOCs.1,50 τ radicals, i.e., dissociation, isomerization, and reaction with O2. We evaluate the atmospheric lifetime ( ) for AcAc at the 1 The reaction of with O2 is competitive only if isomerization is ff τ = di erent altitude and [OH] according to [], where impossible and dissociation forms primary alkyl radicals. The ktotal OH isomerization occurs only if there exists a H atom located at [OH] and ktotal are the OH concentration and total rate constant, respectively. Using a lapse rate of 6.5 K km−1 for the four carbons away from the radical center (which is absent in 47 our reaction system. Hence, we focus on the subsequent typical tropospheric condition, we determine the lifetime of dissociation of RO-1 and RO-2 involves the following stepwise AcAc, and results are presented in Table S4. At the ground × 6 −3 processes, level with [OH] = 1 10 molecules cm (12 h daytime average), the lifetime of AcAc is 1.54 h. The AcAc lifetime R14‐‐ 1or R14 2 RO‐⎯ 1⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ →TS12 or TS13 → ER and MG or acetyl radicaland PD decreases rapidly with increasing height and [OH]. For example, the τ value decreases to 0.16 h as [OH] increases R24‐‐ 1or R24 2 − RO‐⎯ 2⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ →TS22 or TS23 to 1.5 × 107 molecules cm 3.

→‐‐‐‐MR and AA or butane 2,4 dione 3,4 diol and CH3 Decomposition of RO-1 via TS12 or TS13 yields (R14-1) ■ ATMOSPHERIC IMPLICATIONS MG and ethanediol radical (ER) or (R14-2) acetyl radical Ketones are generally less reactive because the less reactive Δ 1 (AR) and propanal-2,2-diol (PD), with the Ea values of 0.40 keto-forms are preferred than the more reactive enol-forms, −1 Δ ﬀ and 3.73 kcal mol , respectively. The small Ea di erence while for AcAc the enol-form is dominant due to intra- between R14-1 and R14-2 indicates that the products of the molecular hydrogen bonding.44 We have assessed the Δ two pathways are both accessible. The Ea values of the two mechanism, kinetics, and atmospheric fate of the OH-initiated decomposition pathways of RO-2 are 1.48 and 8.64 kcal mol−1 oxidation for the enolic and ketonic isomers of AcAc using to form (R24-1) AA and methylglyoxal radical (MR) and quantum chemical and kinetic rate calculations (Figure 5). OH (R24-2) butane-2,4-dione-3,4-diol and methyl radical (CH3), addition to enol-AcAc is more favorable than addition to keto- Δ Δ π ﬀ respectively. The large Ea and small Er for R24-2 indicate a AcAc, because of the conjugated -electron e ect. Con- minor importance to form butane-2,4-dione-3,4-diol and CH3. sequently, the rate constant for the keto-pathway is much Although the transition states of R14 and R24-2 pathways have smaller than that of the enol-pathway, and the reaction of the lower energies than those of the corresponding products enol-AcAc with OH represents the dominant pathway for AcAc (Figure 4), there exist the product complexes (COM12 and oxidation in the troposphere, largely determining the fate and COM13) at the exit channels. Hence, AA and MG are the impacts of its products. For the reaction of the enol-AcAc with major products from both R14-1 and R24-1. OH, the rate constants for OH addition are by 2−3 orders of Subsequently, ER, PD, and MR further react with O2 via the magnitude higher than those for H-abstraction, and OH following stepwise processes, addition to enol-AcAc occurs via a prereactive complex. Our R15:O+− R17 HO derived total rate constant for the reaction of AcAc with OH is 22 × −13 3 −1 −1 ERERROA⎯→⎯⎯⎯⎯⎯⎯⎯⎯ ‐2 ⎯⎯⎯ → TS15 ⎯⎯⎯⎯⎯ → A ktotal = 1.69 10 exp(1935/T) cm molecule s over the temperature range of 200−310 K and has a value of 11.1 × R16:O++22 R18:NO −11 3 −1 −1 ARARROPA⎯→⎯⎯⎯⎯⎯⎯⎯⎯ ‐2 ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ → N 10 cm molecule s at 298 K, consistent with the previous experimental measurements.13,16,17 +− R25:O22 R26 HO For the subsequent reactions of the OH-enol-AcAc adduct, MRMRROMG⎯→⎯⎯⎯⎯⎯⎯⎯⎯TS23 → ‐2 ⎯⎯⎯ → TS24 ⎯⎯⎯⎯⎯ → O2 addition to the C3-adduct proceeds barrierlessly with a Δ − − The Er values of R15 and R16 are 41.94 and 35.53 kcal large negative reaction energy (e-add-2), while O2 addition to mol−1, respectively, via the barrierles processes to form two the C2-adduct occurs with a high activation barrier (about 21

11174 DOI: 10.1021/acs.est.8b03972 Environ. Sci. Technol. 2018, 52, 11169−11177 Environmental Science & Technology Article kcal mol−1) and a smaller exothermicity (e-add-1). AA and MG initiated oxidation of enol-AcAc (Figure 4), facilitating the are the most favorable products from the two peroxy radical cycling between NO and NO2 and producing tropospheric 51 pathways (RO2-1 and RO2-2), with a comparable yield. The ozone. Hence, because of its high reactivity and dominant branching ratio of about 81% leading to the formation of the yields for MG and AA, the photochemical oxidation of AcAc two peroxy radicals likely corresponds to the upper limit for may contribute importantly to ozone and SOA formation the AA and MG yields, indicating the dominant production of under polluted environments, with implications for air quality, the two species from the OH-initiated oxidation of AcAc. The human health, and climate. Our results provide the kinetic and previous experimental study for the OH-initiated oxidation of mechanistic data for inclusion of the oxidation of AcAc in AcAc by Zhou et al. obtained the yields for AA (about 17%) atmospheric models. Future studies are necessary to assess the and MG (about 21%).13 Considering possible wall loss and impacts of AcAc on ozone and SOA formation using chemical secondary reactions (including photolysis for MG) of these transport models, with the consideration of its emission − species in the chamber work,33 36 those measured yields likely inventory, chemistry, and transport. correspond to the lower experimental limits. Our results indicate a minor pathway leading to PAN, consistent with the ■ ASSOCIATED CONTENT 13 small experimental yield of PAN (2%). Other plausible *S Supporting Information products from the OH-initiated oxidation of AcAc include The Supporting Information is available free of charge on the organic nitrates (RONO2) directly arising from the peroxy ACS Publications website at DOI: 10.1021/acs.est.8b03972. nitrites (RO2NO-1 and RO2NO-2) and minor butane-2,4- dione-3,4-diol. However, direct formation of organic nitrates The structures, Cartesian coordinates, frequencies, zero- from peroxy nitrites typically constitutes a minor pathway, in point energies, and absolute energies of all relevant contrast to the dominant formation for alkoxy radicals.51 Using species involved in the title reaction, along with the the predicted temperature-dependence kinetic data, we branching ratios, rate constants, and lifetimes (PDF) estimate a lifetime of less than a few hours due the OH- initiated oxidation of AcAc throughout the entire tropospheric ■ AUTHOR INFORMATION conditions. Corresponding Authors Figure 6 shows a time series of AcAc measurements in * Nanjing and Beijing, China. The measured AcAc concentration Phone: 86-20-23883536; fax: 86-20-23883536; e-mail: [email protected] (T.A.). *Phone: 979-845-7656; fax: 979-862-4466; e-mail: renyi- [email protected] (R.Z.). ORCID Taicheng An: 0000-0001-6918-8070 Renyi Zhang: 0000-0001-8708-3862 Author Contributions ∥ These authors contributed equally to this work. Notes The authors declare no competing financial interest. ■ ACKNOWLEDGMENTS This work was financially supported by National Natural Science Foundation of China (41675122, 41425015, U1401245, and 41373102), Science and Technology Program of Guangzhou City (201707010188), Local Innovative and Research Teams Project of Guangdong Pearl River Talents Program (2017BT01Z032), Team Project from the Natural Science Foundation of Guangdong Province, China (S2012030006604), and the Special Program for Applied Figure 6. Time series of AcAc measured using the PRT- MS method Research on Super Computation of the NSFC-Guangdong in Nanjing (a) and Beijing (b), China. The major tick mark on the x- Joint Fund (the second phase), and the National Super- axis labels the local time at midnight. computing Centre in Guangzhou (NSCC-GZ). R.Z. acknowl- edged support from the Robert A. Welch foundation (A-1417). varies from ppt to ppb levels and exhibits a diurnal variation in both locations, which is likely regulated by the emission, ■ REFERENCES planetary boundary layer height, and photochemical activity. (1) Atkinson, R.; Arey, J. Atmospheric degradation of volatile − Such atmospheric concentrations of AcAc are clearly organic compounds. Chem. Rev. 2003, 103 (12), 4605 4638. significant, considering its high reactivity with OH and short (2) Dalmasso, P. R.; Taccone, R. A.; Nieto, J. D.; Cometto, P. M.; atmospheric lifetimes under tropospheric conditions. Since AA Cobos, C. J.; Lane, S. I. Reactivity of hydrohaloethers with OH 14,15,19−24,52−54 radicals and chlorine atoms: Correlation with molecular properties. and MG are important SOA precursors, Atmos. Environ. 2014, 91, 104−109. accurate determination of their yields from AcAc oxidation (3) Lewis, A. C.; Carslaw, N.; Marriott, P. J.; Kinghorn, R. M.; by OH is critical to assess SOA formation under polluted Morrison, P.; Lee, A. L.; Bartle, K. D.; Pilling, M. J. A larger pool of environments. Furthermore, organic peroxy radical (RO2) and ozone-forming carbon compounds in urban atmospheres. Nature − hydroperoxy radical (HO2) are produced from the OH- 2000, 405 (6788), 778 781.

11175 DOI: 10.1021/acs.est.8b03972 Environ. Sci. Technol. 2018, 52, 11169−11177 Environmental Science & Technology Article

(4) Fares, S.; Paoletti, E.; Loreto, F.; Brilli, F. Bidirectional flux of (21) Lim, Y. B.; Tan, Y.; Turpin, B. J. Chemical insights, explicit methyl vinyl ketone and methacrolein in trees with different chemistry and yields of secondary organic aerosol from methylglyoxal isoprenoid emission under realistic ambient concentrations. Environ. and glyoxal. Atmos. Chem. Phys. Discuss. 2013, 13 (17), 4687−4725. Sci. Technol. 2015, 49 (13), 7735−7742. (22) Lim, Y. B.; Tan, Y.; Perri, M. J.; Seitzinger, S. P.; Turpin, B. J. (5) Liu, Y.; Shao, M.; Kuster, W. C.; Goldan, P. D.; Li, X.; Lu, S.; Aqueous chemistry and its role in secondary organic aerosol (SOA) Gouw, J. A. D. Source identification of reactive hydrocarbons and formation. Atmos. Chem. Phys. 2010, 10, 10521−10539. oxygenated VOCs in the summertime in Beijing. Environ. Sci. Technol. (23) De Haan, D. O.; et al. Brown Carbon Production in 2009, 43 (1), 75−81. Ammonium- or Amine-Containing Aerosol Particles by Reactive (6) Arnold, F.; Bürger, V.; Droste-Fanke, B.; Grimm, F.; Krieger, A.; Uptake of Methylglyoxal and Photolytic Cloud Cycling. Environ. Sci. Schneider, J.; Stilp, T. Acetone in the upper troposphere and lower Technol. 2017, 51, 7458−7466. stratosphere: Impact on trace gases and aerosols. Geophys. Res. Lett. (24) De Haan, D. O.; et al. Nitrogen-containing, light-absorbing 1997, 24 (23), 3017−3020. oligomers produced in aerosol particles exposed to methylglyoxal, (7) Ji, Y.; Wang, L.; Li, Z.; Liu, J.; Sun, C. Theoretical dynamic photolysis, and cloud cycling. Environ. Sci. Technol. 2018, 52, 4061− studies on the reaction of CH3C (O) CH3−nFn with the hydroxyl 4071. radical and the chlorine atom. ChemPhysChem 2006, 7 (8), 1741− (25) Zhang, R.; Suh, I.; Zhao, J.; Zhang, D.; Fortner, E. C.; Tie, X.; 1749. Molina, L. T.; Molina, M. J. Atmospheric new particle formation (8) Atkinson, R.; Tuazon, E. C.; Aschmann, S. M. Atmospheric enhanced by organic acids. Science 2004, 304 (5676), 1487−1490. chemistry of 2-pentanone and 2-heptanone. Environ. Sci. Technol. (26) Altieri, K. E.; Seitzinger, S. P.; Carlton, A. G.; Turpin, B. J.; 2000, 34 (4), 623−631. Klein, G. C.; Marshall, A. G. Oligomers formed through in-cloud (9) Warneke, J.; Van Dorp, W. F.; Rudolf, P.; Stano, M.; Papp, P.; methylglyoxal reactions: Chemical composition, properties, and ̌ Matejcík,̌ S.; Borrmann, T.; Swiderek, P. Acetone and the precursor mechanisms investigated by ultra-high resolution FT-ICR mass ligand acetylacetone: Distinctly different electron beam induced spectrometry. Atmos. Environ. 2008, 42 (7), 1476−1490. decomposition? Phys. Chem. Chem. Phys. 2015, 17 (2), 1204−1216. (27) Qiu, C.; Zhang, R. Multiphase chemistry of atmospheric (10) Ho, Y. S.; Wu, C. H.; Chen, Y. S.; Liu, C. K. Effects of molar amines. Phys. Chem. Chem. Phys. 2013, 15 (16), 5738−5752. ratio of acetylacetone to aluminum precursor on the yttrium (28) Qiu, C.; Wang, L.; Lal, V.; Khalizov, A. F.; Zhang, R. − aluminum garnet (Y3Al5O12, YAG) formation by a sol gel process. Heterogeneous reactions of alkylamines with ammonium sulfate and Int. J. Appl. Ceram. Technol. 2015, 12 (S2), E53−E58. ammonium bisulfate. Environ. Sci. Technol. 2011, 45 (11), 4748− (11) Yang, H.; Sun, H.; Zhang, S.; Wu, B.; Pan, B. Potential of 4755. acetylacetone as a mediator for Trametes versicolor laccase in (29) Gomez, M. E.; Lin, Y.; Guo, S.; Zhang, R. Heterogeneous enzymatic transformation of organic pollutants. Environ. Sci. Pollut. chemistry of glyoxal on acidic solutions an oligomerization pathway Res. 2015, 22 (14), 10882−10889. for secondary organic aerosol formation. J. Phys. Chem. A 2015, 119 (12) El-Sayed, H. A.; Said, S. A.; Amr, A. E.-G. E. Synthesis of some (19), 4457−4463. fused heterocyclic systems and their nucleoside candidates. Res. Chem. (30) Zhao, J.; Levitt, N. P.; Zhang, R.; Chen, J. Heterogeneous Intermed. 2014, 40 (2), 833−845. reactions of methylglyoxal in acidic media: Implications for secondary (13) Zhou, S.; Barnes, I.; Zhu, T.; Bejan, I.; Albu, M.; Benter, T. organic aerosol formation. Environ. Sci. Technol. 2006, 40 (24), 7682− Atmospheric chemistry of acetylacetone. Environ. Sci. Technol. 2008, 7687. 42 (21), 7905−7910. (31) Ji, Y.; Li, Y.; An, T.; Zhang, R. Reply to Newland et al.: The (14) Zhang, R.; Wang, G.; Guo, S.; Zamora, M. L.; Ying, Q.; Lin, Y.; dominant phenolic pathway for atmospheric toluene oxidation. Proc. Wang, W.; Hu, M.; Wang, Y. Formation of urban fine particulate Natl. Acad. Sci. U. S. A. 2017, 114 (38), e7858−e7859. matter. Chem. Rev. 2015, 115 (10), 3803−3855. (32) Ji, Y.; et al. Reassessing the atmospheric oxidation mechanism (15) Guo, S.; Hu, M.; Zamora, M. L.; Peng, J.; Shang, D.; Zheng, J.; of toluene. Proc. Natl. Acad. Sci. U. S. A. 2017, 114, 8169−8174. Du, Z.; Wu, Z.; Shao, M.; Zeng, L. Elucidating severe urban haze (33) Zhang, X.; Cappa, C. D.; Jathar, S. H.; McVay, R. C.; Ensberg, formation in China. Proc. Natl. Acad. Sci. U. S. A. 2014, 111 (49), J. J.; Kleeman, M. J.; Seinfeld, J. H. Influence of vapor wall loss in 17373−17378. laboratory chambers on yields of secondary organic aerosol. Proc. (16) Holloway, A.-L.; Treacy, J.; Sidebottom, H.; Mellouki, A.; Natl. Acad. Sci. U. S. A. 2014, 111 (16), 5802−5807. Daele,̈ V.; Le Bras, G.; Barnes, I. Rate coefficients for the reactions of (34) Matsunaga, A.; Ziemann, P. J. Gas-Wall Partitioning of Organic OH radicals with the keto/enol tautomers of 2, 4-pentanedione and 3- Compounds in a Teflon Film Chamber and Potential Effects on methyl-2, 4-pentanedione, allyl alcohol and methyl vinyl ketone using Reaction Product and Aerosol Yield Measurements. Aerosol Sci. the enols and methyl nitrite as photolytic sources of OH. J. Photochem. Technol. 2010, 44 (10), 881−892. Photobiol., A 2005, 176 (1-3), 183−190. (35) La, Y. S.; Camredon, M.; Ziemann, P. J.; Valorso, R.; (17) Bell, P.; Nicovich, J. M.; Wine, P. H. Acetylacetone photolysis Matsunaga, A.; Lannuque, V.; Lee-Taylor, J.; Hodzic, A.; Madronich, at 248 nm: hydroxyl radical yield and temperature-dependent rate S.; Aumont, B. Impact of chamber wall loss of gaseous organic coeﬃcients for the OH+ acetylacetone reaction. Proceedings of 19th compounds on secondary organic aerosol formation: explicit International Symposium on Gas Kinetics, Orleans, France, July 22−27, modeling of SOA formation from alkane and alkene oxidation. 2006; pp 79−80. Atmos. Chem. Phys. 2016, 16 (3), 1417−1431. (18) Dagaut, P.; Wallington, T. J.; Liu, R.; Kurylo, M. J. A kinetic (36) Krechmer, J. E.; Pagonis, D.; Ziemann, P. J.; Jimenez, J. L. investigation of the gas-phase reactions of hydroxyl radicals with cyclic Quantification of Gas-Wall Partitioning in Teflon Environmental ketones and diones: mechanistic insights. J. Phys. Chem. 1988, 92 Chambers Using Rapid Bursts of Low-Volatility Oxidized Species (15), 4375−4377. Generated in Situ. Environ. Sci. Technol. 2016, 50 (11), 5757−5765. (19) Tan, Y.; Carlton, A. G.; Seitzinger, S. P.; Turpin, B. J. SOA from (37) Frisch, M.; Trucks, G.; Schlegel, H. B.; Scuseria, G.; Robb, M.; methylglyoxal in clouds and wet aerosols: Measurement and Cheeseman, J.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. prediction of key products. Atmos. Environ. 2010, 44 (39), 5218− Gaussian 09, revision D. 01; Gaussian Inc.: Wallingford CT, 2009. 5226. (38) Zhao, Y.; Truhlar, D. G. The M06 suite of density functionals (20) Tan, Y.; Lim, Y. B.; Altieri, K. E.; Seitzinger, S. P.; Turpin, B. J. for main group thermochemistry, thermochemical kinetics, non- Mechanisms leading to oligomers and SOA through aqueous covalent interactions, excited states, and transition elements: Two photooxidation: insights from OH radical oxidation of acetic acid new functionals and systematic testing of four M06-class functionals and methylglyoxal. Atmos. Chem. Phys. Discuss. 2011, 11 (11), 18319− and 12 other functionals. Theor. Chem. Acc. 2008, 120 (1−3), 215− 18347. 241.

11176 DOI: 10.1021/acs.est.8b03972 Environ. Sci. Technol. 2018, 52, 11169−11177 Environmental Science & Technology Article

(39) Zhang, D.; Zhang, R. Mechanism of OH formation from ozonolysis of isoprene: A quantum-chemical study. J. Am. Chem. Soc. 2002, 124, 2692−2703. (40) Boys, S. F.; Bernardi, F. The calculation of small molecular interactions by the differences of separate total energies. Some procedures with reduced errors. Mol. Phys. 1970, 19, 553−566. (41) Zheng, J.; Zhang, S.; Lynch, B.; Corchado, J.; Chuang, Y.; Fast, P.; Hu, W.; Liu, Y.-P.; Lynch, G.; Nguyen, K. POLYRATE, version 2010-A; Software Manager; University of Minnesota: Minneapolis, MN, 2010. (42) Zhao, J.; Zhang, R. Y. Proton transfer reaction rate constants + between hydronium ion (H3O ) and volatile organic compounds. Atmos. Environ. 2004, 38 (14), 2177−2185. (43) Zheng, J.; Ma, Y.; Chen, M.; Zhang, Q.; Wang, L.; Khalizov, A. F.; Yao, L.; Wang, Z.; Wang, X.; Chen, L. Measurement of atmospheric amines and ammonia using the high resolution time- of-flight chemical ionization mass spectrometry. Atmos. Environ. 2015, 102 (0), 249−259. (44) Nakanishi, H.; Morita, H.; Nagakura, S. Electronic structures and spectra of the keto and enol forms of acetylacetone. Bull. Chem. Soc. Jpn. 1977, 50 (9), 2255−2261. (45) Alvarez-Idaboy, J. R.; Mora-Diez, N.; Boyd, R. J.; Vivier-Bunge, A. On the importance of prereactive complexes in molecule-radical reactions: Hydrogen abstraction from aldehydes by OH. J. Am. Chem. Soc. 2001, 123 (9), 2018−2024. (46) Hammond, G. S. A correlation of reaction rates. J. Am. Chem. Soc. 1955, 77 (2), 334−338. (47) Ji, Y.; Wang, H.; Gao, Y.; Li, G.; An, T. A theoretical model on the formation mechanism and kinetics of highly toxic air pollutants from halogenated formaldehydes reacted with halogen atoms. Atmos. Chem. Phys. 2013, 13 (22), 11277−11286. (48) Atkinson, R. Rate constants for the atmospheric reactions of alkoxy radicals: an updated estimation method. Atmos. Environ. 2007, 41, 8468−8485. (49) Kroll, J. H.; Seinfeld, J. H. Chemistry of secondary organic aerosol: Formation and evolution of low-volatility organics in the atmosphere. Atmos. Environ. 2008, 42, 3593−3624. (50) Wang, H.; Ji, Y.; Gao, Y.; Li, G.; An, T. Theoretical model on the formation possibility of secondary organic aerosol from OH initialed oxidation reaction of styrene in the presence of O2/NO. Atmos. Environ. 2015, 101,1−9. (51) Zhang, D.; Zhang, R.; Park, J.; North, S. W. Hydroxy peroxy nitrites and nitrates from OH initiated reactions of isoprene. J. Am. Chem. Soc. 2002, 124 (32), 9600−9605. (52) Zhang, R.; Khalizov, A. F.; Wang, L.; Hu, M.; Xu, W. Nucleation and growth of nanoparticles in the atmosphere. Chem. Rev. 2012, 112, 1957−2011. (53) Zhang, R.; Lei, W.; Tie, X.; Hess, P. Industrial emissions cause extreme urban ozone diurnal variability. Proc. Natl. Acad. Sci. U. S. A. 2004, 101 (17), 6346−6350. (54) Iglesias, E. Determination of the keto-enol equilibrium constants and the kinetic study of the nitrosation reaction of β- dicarbonyl compounds. J. Chem. Soc., Perkin Trans. 2 1997, 431−439.

11177 DOI: 10.1021/acs.est.8b03972 Environ. Sci. Technol. 2018, 52, 11169−11177