Reactions of Atmospheric Particulate Stabilized Criegee Intermediates

May 17, 2016 - Aging of organic aerosol particles is one of the most poorly understood topics in atmospheric aerosol research. Here, we used an aeroso...
15 downloads 7 Views 1MB Size
Subscriber access provided by UCL Library Services

Article

Reactions of atmospheric particulate stabilized Criegee intermediates lead to high-molecular-weight aerosol components MingYi Wang, Lei Yao, Jun Zheng, XinKe Wang, Jianmin Chen, Xin Yang, Douglas R. Worsnop, Neil M. Donahue, and Lin Wang Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b02114 • Publication Date (Web): 17 May 2016 Downloaded from http://pubs.acs.org on May 24, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 35

Environmental Science & Technology

1

Reactions of atmospheric particulate stabilized Criegee intermediates

2

lead to high-molecular-weight aerosol components

3 4

MingYi Wang†, Lei Yao†, Jun Zheng‡, XinKe Wang†, JianMin Chen†, Xin Yang†,

5

Douglas R. Worsnop∆, Neil M. Donahue§, Lin Wang†,*

6 7



Shanghai Key Laboratory of Atmospheric Particle Pollution and Prevention (LAP3),

8

Department of Environmental Science & Engineering, Fudan University, Shanghai

9

200433, China

10



11

Control, Nanjing University of Information Science & Technology, Nanjing 210044,

12

China

13



14

§

15

Avenue, Pittsburgh, Pennsylvania 15213, United States

Jiangsu Key Laboratory of Atmospheric Environment Monitoring and Pollution

Aerodyne Research, Billerica, MA 01821, United States

Center for Atmospheric Particle Studies, Carnegie Mellon University, 5000 Forbes

16

ACS Paragon Plus Environment

Environmental Science & Technology

17

ABSTRACT

18

Aging of organic aerosol particles is one of the most poorly understood topics in

19

atmospheric aerosol research. Here, we used an aerosol flow tube together with an

20

iodide-adduct high-resolution time-of-flight chemical-ionization mass spectrometer

21

equipped with a Filter Inlet for Gases and AEROsols (FIGAERO-HRToF-CIMS) to

22

investigate heterogeneous ozonolysis of oleic acid (OL), developing a comprehensive

23

oxidation mechanism with observed products. In addition to the well-known

24

first-generation C9 products including nonanal, nonanoic acid, azelaic acid, and

25

9-oxononanoic acid, the iodide-adduct chemical ionization permitted unambiguous

26

determination of a large number of high-molecular-weight particulate products up to

27

670 Da with minimum amounts of fragmentation. These high-molecular-weight

28

products are characterized by a fairly uniform carbon oxidation state but stepwise

29

addition of a carbon backbone moiety, and hence continuous decrease in the volatility.

30

Our results demonstrate that heterogeneous oxidation of organic aerosols has a

31

significant effect on the physiochemical properties of organic aerosols and that

32

reactions of particulate SCIs from ozonolysis of an unsaturated particulate species

33

represent a previously underappreciated mechanism that lead to formation of

34

high-molecular-weight particulate products that are stable under typical atmospheric

35

conditions.

36

ACS Paragon Plus Environment

Page 2 of 35

Page 3 of 35

Environmental Science & Technology

37

1. INTRODUCTION

38

Aerosols exert large and uncertain climate effects1 and are thought to kill more than 7

39

million people annually2. Organics constitute a large fraction of the ambient particle

40

mass with rich and poorly understood chemistry. Their chemical evolution may well

41

govern aerosol physiochemical properties and environmental impacts3. The

42

atmosphere is strongly oxidizing, and unlike well-controlled laboratory synthesis,

43

complexity is an essential characteristic of atmospheric oxidation chemistry. Until

44

recently, however, we have lacked the experimental tools to probe and understand that

45

chemistry in anything like full detail, even for relatively simple model systems. A key

46

challenge is to simultaneously describe these complex mechanisms in full detail and

47

to synthesize those descriptions within integrative frameworks that bring order and

48

meaning to the complexity. Here we present results for one such model system – fatty

49

acid oxidation by ozone – using new mass spectrometric methods to illuminate the

50

mechanism with unprecedented detail.

51

Mechanisms for evolution of organic aerosols can be broken down into three key

52

classes: functionalization, fragmentation, and oligomerization. These govern

53

trajectories of organic-aerosol components within a space defined by carbon number

54

(nC) and carbon oxidation state (OSC)4. Functionalization reactions add functional

55

groups to a carbon backbone without altering nC. Fragmentation reactions cleave the

56

carbon backbone, generating two or more products with lower nC and (often) higher

57

OSC. Oligomerization reactions, the association of two organic molecules, preserve

58

OSC but increase nC, and hence lead to a large reduction in volatility. Evidence for the

ACS Paragon Plus Environment

Environmental Science & Technology

59

important role of such accretion reactions in the formation of secondary organic

60

aerosols includes identification of high-molecular-weight species such as hemiacetals

61

and acetals5, peroxyhemiacetals and peroxyacetals6,

62

organosulfates10. Reactions of stabilized Criegee intermediates (SCIs) may be another

63

class of accretion reaction leading to formation of oligomers in organic aerosols11, 12.

7

, aldols8, esters9, and

64

SCIs come from collisional stabilization of Criegee intermediates formed via

65

decomposition of primary ozonides following the reaction of alkenes and ozone.

66

Gaseous SCIs react with water vapor13, sulfur dioxide14, nitrogen dioxide15, and small

67

organic molecules15. The fate of SCIs in atmospheric particles is less elucidated, even

68

though solution phase ozonolysis has been studied much more historically16. Oleic

69

acid (OL) is a C18 ω-9 unsaturated carboxylic acid and a major constituent of

70

triglycerides making up olive oil. It is a model low-volatility organic compound and is

71

commonly used as a tracer for cooking activities17. Following heterogeneous

72

ozonolysis of oleic acid, secondary reactions of particulate SCIs formed from the

73

primary ozonolysis are thought to lead to the formation of high-molecular-weight

74

oligomers18-22. The reaction mechanisms based on a few randomly identified

75

oligomers from their fragmentation patterns, although covering a number of reaction

76

pathways, are fragmentary and seemingly contradictory in some cases. Since this is a

77

benchmark for heterogeneous oxidation of unsaturated organics by ozone, a

78

comprehensive reaction mechanism unifying the seemingly disparate fragmentary

79

mechanisms is important. A quantitative understanding on the branching ratios of

80

different reaction pathways and on the concentration evolution with excellent mass

ACS Paragon Plus Environment

Page 4 of 35

Page 5 of 35

Environmental Science & Technology

81

closure is necessary to permit accurate treatment of heterogeneous aging processes of

82

atmospheric unsaturated particles in global models.

83 84

2. MATERIALS AND METHODS

85

As illustrated in Figure S1, we carried out heterogeneous ozonolysis reactions in an

86

aerosol flow tube coupled to a High-Resolution Time-of-Flight Chemical-Ionization

87

Mass Spectrometer equipped with a Filter Inlet for Gases and AEROsols

88

(FIGAERO-HRToF-CIMS) using the iodide ion (I-) as the reagent ion. Details of the

89

experimental setup and the analytical methods follow.

90

We generated pure organic particles via homogeneous nucleation by passing a flow

91

of ultra-high purity (UHP) N2 (99.99%, Pujiang Specialty Gases Factory, China) at

92

0.1 slpm (standard liters per minute) over an insulated organic reservoir kept at 110 ±

93

1 °C. The outflow was immediately diluted by 1.0 slpm zero air generated by a pure

94

air generator (Model 737, Aadco, USA), and then directed to a cylindrical VOC

95

scrubber filled with a mixture of KMnO4/activated charcoal (1:1, v/v) to scavenge

96

excess organic vapors. We measured the size distribution of the organic particles using

97

a Scanning Mobility Particle Sizer (SMPS, consisting of one DMA 3081 and one CPC

98

3775, TSI, USA). The geometric mean diameter, standard deviation, and integrated

99

number concentration of oleic acid (OL, ≥ 99.0%, Sigma-Aldrich) particles were 110

100

nm, 1.45, and 1.8×105 particles cm-3, respectively. The geometric mean diameter,

101

standard deviation and integrated number concentration of erucic acid (EA, ≥ 99.0%,

102

Sigma-Aldrich) particles were 76 nm, 1.32, and 6.8×105 particles cm-3, respectively.

ACS Paragon Plus Environment

Environmental Science & Technology

Page 6 of 35

103

In the case of the two-component mixture particles, we dissolved a mixture of 37.5%

104

oleic acid / 62.5% 1-dodecanol (or 1-octadecene / 1-dodecyl aldehyde /

105

2-hexyldecanoic acid) (mol/mol) in ethanol and then atomized the mixture. The

106

outflow was then directed through a VOC scrubber to eliminate ethanol before it

107

entered the aerosol flow tube.

108

We produced ozone by photolysis of molecular oxygen using an Hg Pen-ray lamp

109

(Model 600, Jelight, USA) and monitored the concentration of ozone in the outflow

110

with a UV absorption ozone analyzer (Model 49i, Thermo, USA).

111

The aerosol flow tube consists of a 7.75 mm i.d. stainless-steel moveable aerosol

112

injector that traverses the centerline of a quartz tube with an i.d. of 8 cm and a length

113

of 1.2 m, and two upstream side injectors for the introduction of the gaseous reactant.

114

In this study, we introduced the organic aerosol and the ozone into the flow tube

115

through the movable injector and the side injectors, respectively. The total flow

116

through the reaction zone in the flow tube was typically 6.0 slpm, indicating a laminar

117

flow (Reynolds number of ~108) at 2.0 cm/s with an entrance length of ~50 cm to

118

fully achieve the laminar condition. The mixing length was 44 cm23, based on a

119

diffusion coefficient of 0.1444 cm2/s for O3 in air at the atmospheric pressure24. Hence,

120

the interaction distance between organic particles and ozone was set between 40 cm to

121

80 cm, corresponding to a residence time of 20-40 s.

122

We

analyzed

the

gaseous Briefly,

and

particle-associated

HRToF-CIMS

offers

organics

a

123

FIGAERO-HRToF-CIMS.

124

measurement with high sensitivity and high mass resolution25-28. FIGAERO is a

ACS Paragon Plus Environment

mass

with

spectrometry

Page 7 of 35

Environmental Science & Technology

125

manifold with two operation modes29. In gas mode, we directly analyzed the gas

126

sample with the HRToF-CIMS while simultaneously collecting particles via a separate

127

dedicated port, using a PTFE filter (5 µm, Millipore, USA) that collected organic

128

particles with ~99.5% efficiency as measured by an SMPS. In particle mode, we

129

analyzed vapors evolved from the temperature-programmed thermal desorption of the

130

collected particles by heated UHP N2. We used a moveable tray to switch

131

automatically between the two modes. Note that before filter collection we directed

132

the aerosol sample flow through a cylindrical diffusion ozone scrubber filled with a

133

mixture of Na2SO3/activated charcoal (1:1, v/v). The particle loss in the ozone

134

scrubber was negligible30. The typical collection time for particle samples was 3 min.

135

We used iodide-adduct ionization in this study for its large negative mass defect,

136

which provides an added degree of separation, as well as minimal fragmentation

137

during adduct formation31. We formed iodide ions (I-) by passing a 1.0 slpm flow of

138

UHP N2 over a diffusion tube filled with methyl iodide (CH3I, Xiya Reagent, China),

139

and then through an 241Am ion source (0.1 mCi). We then mixed the reagent ion flow

140

with a sample flow in the ion-molecule reactor (IMR) at ~110 mbar, resulting in I-

141

(m/z 126.9050 amu) as the most abundant ion and (I·H2O)- (m/z 144.9156 amu) as the

142

next dominant one. We tuned the axial electric field strengths ( 99.97 % of entering particles but leaving semi-volatile gases largely

155

unperturbed, and then through the FIGAERO filter for the same collection duration as

156

for a normal particle sample. We then subtracted the background signals from the

157

sample signals.

158

We calibrated the instrument by spiking the FIGAERO filter with solutions of

159

nonanal (NN, ≥ 95.0%, Sigma-Aldrich), nonanoic acid (NA, ≥ 97.0%,

160

Sigma-Aldrich), and azelaic acid (AA, ≥ 98.0%, Sigma-Aldrich) at three known

161

concentrations. The linear correlation coefficients of all compounds were larger than

162

0.95 (Figure S2).

163 164

3. RESULTS AND DISCUSSION

165

3.1 Kinetics, Products and Mechanism. We generated oleic-acid particles with

166

diameters near 100 nm and exposed them to excess ozone in an aerosol flow tube

167

(Figure S1). We measured the uptake coefficient  by examining the decay of oleic

168

acid as a function of the ozone exposure, and determined the products using an

ACS Paragon Plus Environment

Page 8 of 35

Page 9 of 35

Environmental Science & Technology

169

iodide-adduct

High-Resolution

170

Spectrometer

171

(FIGAERO-HRToF-CIMS)29,

172

ozonolysis reaction of oleic acid is calculated to be (1.19 ± 0.09) × 105 atm-1 s-1 for

173

ozone exposures ranging from 0 to 1.2 × 10-5 atmsec, as plotted in Figure S3. The

174

derived uptake coefficient , (1.44 ± 0.11) × 10-3, is reported together with values

175

from previous studies in Table S1.

equipped

with 31

.

Time-of-Flight a

Filter

Inlet

Chemical-Ionization for

Gases

and

Mass

AEROsols

The pseudo-first-order rate constant () of the

176

The uptake coefficient we determine is in fairly good consistence with the previous

177

ones obtained by directly monitoring the decay of the condensed-phase oleic acid,

178

which are, specifically, AMS studies by Morris et al.32 ((1.6±0.2)×10−3) and Katrib et

179

al.33 ((1.25±0.2)×10−3), an aerosol-CIMS study by Hearn et al.34 ((1.38±0.06)×10−3),

180

and a GCMS study by Mendez et al.35 ((1.0±0.2)×10−3). On the other hand, other

181

studies that determined the uptake coefficient by monitoring the O3 loss to oleic-acid

182

surfaces, reported a value of ~8×10−4 (ref.

183

and O3-based methods may be attributed to additional oleic acid consumption by

184

secondary reactions within the particles. Oleic acid-based methods that do not account

185

for the secondary chemistry may artificially overestimate the loss rate of oleic acid,

186

resulting in a larger value for .

36-39

). The discrepancy between oleic acid-

187

The mass spectra of initially pure oleic-acid particles after exposure to ozone are

188

complex, as shown in Figure 1. In addition to iodide adducts of unreacted oleic acid

189

and four well-known first-generation products (nonanal, NN; nonanoic acid, NA;

190

azelaic acid, AA; and 9-oxononanoic acid, OX), dozens of other iodide-adduct ions

ACS Paragon Plus Environment

Environmental Science & Technology

191

are evident. The ions can contain only carbon, hydrogen, oxygen, and iodide, so the

192

high mass-resolving power of our Time-of-Flight mass spectrometer allows us to

193

assign a molecular formula to almost all of the ions, with a mass tolerance of < 3 ppm

194

(except for five ions < 40 ppm (Figure S4)). The associated isotope distribution

195

further supports our molecular formula assignments. Without the FIGEARO

196

thermograms we would be unable to identify or separate isomers, but as we shall

197

discuss below, those thermograms provide important additional information.

198

We propose an SCI-based reaction mechanism to elucidate the formation routes of

199

all the identified products. Figure 2 presents this as a flowchart from the parent oleic

200

acid to the first-, second-, and multi-generation products, using one molecular

201

structure as an example for each category. Full details are provided in Figure S5A-C.

202

In brief, oleic acid reacts with ozone to form a primary ozonide, the decomposition of

203

which leads to the formation of four first-generation products (nonanal, nonanoic acid,

204

azelaic acid, and 9-oxononanoic acid) and two SCIs. The SCIs either isomerize to

205

form an acid (nonanoic acid or azelaic acid), react with oleic acid, react with a

206

first-generation carbonyl (nonanal or 9-oxononanoic acid) to form three secondary

207

ozonides (SOZs), react with a first-generation carboxylic acid (nonanoic acid, azelaic

208

acid, or 9-oxononanoic acid) to form six linear α-acyloxyalkyl hydroperoxides

209

(AAHPs), or react with another SCI to form three cyclic diperoxides (DPs)18-21. SCIs

210

can further react either with an SOZ to form an SOZ-AAHP (four potentially

211

identified in this study), with a DP to form a DP-AAHP (four potentially identified),

212

or with an AAHP to form either an SOZ-AAHP (four potentially identified) or an

ACS Paragon Plus Environment

Page 10 of 35

Page 11 of 35

Environmental Science & Technology

213

AAHP-AAHP (eight potentially identified) depending on whether an aldehyde or a

214

carboxylic acid group reacts with the SCI18. In some cases, we observed dehydration

215

products from these species, for example with AAHP-AAHP8 in Figure S5A. Many

216

of these species share identical molecular formulas and hence the adducts in Figure 1

217

are labelled with all possible isomers. We observed only trace signals of SOZ1, DP1,

218

and AAHP1, either because further reactions consume most of them, or because the

219

absence of terminal functional groups leads to a low clustering affinity with iodide.

220

Figure 2 and Figure S5B indicate that SCIs react with an intact oleic acid to form

221

an OL-AAHP (two potentially identified) or a dioxolane (DO, four potentially

222

identified), depending on whether the SCIs react with the carboxylic acid group or the

223

double bond in oleic acid. This is presumably why the observed acid:ozone

224

stoichiometry is greater than 1:133,

225

OL-AAHP-AAHP (one potentially identified) and with DO to form DO-AAHP (six

226

potentially identified)22. In addition, the dioxolane ring in DO cleaves, giving rise to

227

oxooctadecanoic acid (OO, two potentially identified), which then reacts with SCIs

228

stepwise to form OO-AAHP (four potentially identified) and OO-AAHP-AAHP (two

229

potentially identified). Finally, Figure 2 and Figure S5C show that AAHPs formed

230

from the SCIs react with a first-generation carbonyl (nonanal or 9-oxononanoic acid)

231

to form twelve peroxyhemiacetals (PHAs), which then react with a first-generation

232

carboxylic acid (nonanoic acid, azelaic acid, or 9-oxononanoic acid) to form

233

bis(α-acyloxy-α-alkyl)peroxides (BAAPs)20.

234

39

. SCIs also react with OL-AAHPs to form

We believe that the SCI-based reactions dominant in this system based on our

ACS Paragon Plus Environment

Environmental Science & Technology

Page 12 of 35

235

product analysis, though our current analytical technique does not allow a direct

236

observation of the unimolecular/bimolecular reactions of CIs. If there were significant

237

non-thermalized CI reactions, a significant fraction of the C9 products with multiple

238

oxygen-containing functional groups (oxygen number more than that of azelaic acid)

239

should appear, but those products are not observed in our study. Besides, particulate

240

CIs (including most of the surface CIs) are surrounded by a large amount of “solvent

241

molecules”, and the “cage effect” makes them highly likely to be stabilized due to

242

collisional energy transfer40, 41.

243

We confirmed this mechanism with several tests. We conducted experiments with

244

erucic acid (EA), a C22 ω-13 unsaturated carboxylic acid common in rape-seed oil

245

triglycerides. This reaction forms SCIs with nine and thirteen carbon atoms, but the

246

identified products from erucic acid ozonolysis are fully consistent with the

247

SCI-based

248

mass-to-charge-ratios greater than 800 amu because the signal-to-noise-ratio was

249

insufficient (Figure S6).

mechanism

we

propose.

We

did

not

observe

products

with

250

We also studied ozonolysis of particles consisting of two-component mixtures in

251

order to further test the proposed mechanism. Ozonolysis of 37.5% oleic acid / 62.5%

252

1-octadecene (mol/mol) confirms the previous proposition that SCIs can add to a

253

double bond22, eventually leading to a decomposition product from a DO analogue

254

(Figure S7.). Ozonolysis of 37.5% oleic acid / 62.5% 1-dodecyl aldehyde (mol/mol)

255

and 37.5% oleic acid / 62.5% 2-hexyldecanoic acid (mol/mol) confirms that SCIs

256

from ozonolysis of oleic acid react with aldehyde and carboxylic acid moieties,

ACS Paragon Plus Environment

Page 13 of 35

Environmental Science & Technology

257

resulting in two new SOZ and two new AAHP species, respectively (Figure S8-

258

Figure S9). For ozonolysis of 37.5% oleic acid / 62.5% 1-dodecanol mixtures

259

(mol/mol), we did not observe products other than those from ozonolysis of pure oleic

260

acid, despite a previous report that anti SCI can react with methanol42.

261 262

3.2 Thermogram Analysis. We carefully examined the FIGAERO thermograms,

263

finding evidence for thermal decomposition of second- and multi-generation products

264

supporting our proposed mechanism. We show an example in Figure 3A-C. The

265

thermogram of azelaic acid contains two peaks: the lower-temperature peak

266

corresponds with pure azelaic acid samples, while the higher-temperature peak

267

corresponds to the evaporation and decomposition of the SCI2-OX association

268

product AAHP5 (with a loss of H2O) and also the tertiary products

269

DP-AAHP2/DP-AAHP3/AAHP-AAHP5. In addition, although no authentic standard

270

of oxooctadecanoic acid (OO) is available, the thermogram of OO suggests that the

271

two peaks likely arise from the decomposition of DO1/DO2 (evidently dominant) and

272

OO-AAHP3/OO-AAHP4 (minor), consistent with our mechanism.

273

In Figure 3D and Figure 3E, we present a comprehensive picture of the evaporation

274

and decomposition for all of the C9, C18, and C27 particulate products using the

275

molecular weight of each species as the abscissa. The ordinate in Figure 3D is the

276

thermogram temperature when the signal peaks for each compound, and in Figure 3E

277

is the average carbon oxidation state of each compound. These integrative spaces are

278

analogues of the nC-OSC space, but here the values of nC are trivial (9, 18, and 27) and

ACS Paragon Plus Environment

Environmental Science & Technology

279

so we use molecular weight because it provides more information and better

280

separation.

281

The filled symbols in Figure 3D identify the species in Figure 3A-C. Generally, the

282

temperature of the cooler (first) peak of a double-peak thermogram increases with

283

molecular weight, consistent with the fact that larger products will evaporate later.

284

The higher-temperature peak of a double-peak thermogram usually comes from

285

thermal decomposition of later-generation higher molecular-weight products. That

286

product usually is associated with a single-peak thermogram. However, multiple

287

isomers for a given molecular weight, especially in the case of later-generation

288

products, can also cause multiple-peak thermograms. Also, there is often a lag

289

between decomposition and evaporation for later-generation products when

290

decomposition happens at a lower temperature than desorption for at least one of the

291

fragments. Hence, a perfect match is rare between the thermogram temperature of an

292

early-generation product formed from thermal decomposition of a later-generation

293

oligomer and the thermogram temperature of that oligomer. The major thermal

294

decomposition zone is more like the rectangular band in Figure 3D. Note that the

295

thermogram of nonanal displayed a single peak. Nonanal is far too volatile to reside in

296

the condensed phase as a monomer, so all of the observed particulate nonanal came

297

from the thermal decomposition of later-generation products with similar

298

decomposition temperatures.

299

Our data clearly indicate that accretion reactions occur through stepwise addition of

300

SCIs to molecules with carbonyl, carboxylic, and olefin groups. The least oxidized

ACS Paragon Plus Environment

Page 14 of 35

Page 15 of 35

Environmental Science & Technology

301

SCI from oleic acid is an isomer of nonanoic acid, with OSC = -1.56, while the most

302

oxidized SCI is an isomer of azelaic acid, with OSC = -0.89. Hence, as reactions

303

proceed, the carbon oxidation state of particulate products will converge into a range

304

between those values, corresponding to the gray band in Figure 3E. Although the OSC

305

of the particulate phase is confined to a fairly narrow range, the average molecular

306

weight increases with increasing generation number, resulting in lower volatility.

307

These data further illustrate that thermal scissions of association products (except

308

for from DO to OO) always occur at the linkages between C9 units despite of the

309

different generation numbers; thermal decomposition intensively occurs within the

310

decomposition zone. The lowest-energy pathway leading to evaporation of

311

multi-generation material from the condensed phase is not evaporation of the product

312

itself but rather decomposition and subsequent evaporation of a more volatile

313

decomposition product.

314 315

3.3 Branching Ratios. We can identify the large majority of species in the complex

316

mass spectrum produced by heterogeneous ozonolysis of oleic acid. Furthermore, the

317

thermal decomposition information embedded in the FIGEARO thermograms

318

confirms the mechanism we infer to explain production of those species. The last

319

question is whether we can explain the kinetics quantitatively and obtain mass closure.

320

Figure 4 shows that we can. We developed a simplified model focusing on the

321

production and loss of first-generation species and especially the chemistry of the

322

SCIs.

ACS Paragon Plus Environment

Environmental Science & Technology

Page 16 of 35

323

During the ozonolysis of oleic acid, two SCI species are produced. These SCIs can

324

further react with the C9 products (including aldehydes, acids, and another SCI), or

325

with an intact oleic acid, or undergo isomerization. It is not currently feasible to

326

separate these reactions into independent pathways based on our data set.

327

Consequently, we simplify the reaction mechanism by ignoring the reactions of the

328

second- and multi-generation products, and focus on bimolecular reaction of the SCIs

329

with first-generation (C9) products (k1a), bimolecular reaction of the SCIs with oleic

330

acid (k1b), and isomerization of the SCIs (k2), as shown in Figure S11. The simplified

331

model does not differentiate syn- and anti-conformers of the SCIs.

332

SCI + Aldehyde/SCI/Acid  SOZ/DP/AAH

333

SCI + OL  C product

334

SCI → Aci



(1a)



(1b)

$

(2)

335

We set up a model framework to connect the reaction rates to the observed

336

time-resolved product concentrations. The fraction of SCI isomerization (into acid),

337

&'()*+,-.-/01-+2 ⁄&'(3+104 , is equal to the reaction rate of pathway (2) divided by the

338

sum of the reaction rates of all SCI consuming steps:

339

67)89:;