Modeling Photosensitized Secondary Organic Aerosol Formation in

Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Article

Modeling Photosensitized Secondary Organic Aerosol Formation in Laboratory and Ambient Aerosols William Gang Tsui, Yi Rao, Hai-Lung Dai, and V. Faye McNeill Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b01416 • Publication Date (Web): 12 Jun 2017 Downloaded from http://pubs.acs.org on June 14, 2017

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 20

Environmental Science & Technology

1

Modeling Photosensitized Secondary Organic Aerosol Formation in Laboratory

2

and Ambient Aerosols

3 William G. Tsui,† Yi Rao,‡ Hai-Lung Dai,‡ V. Faye McNeill*,†

4 5 6

†

Department of Chemical Engineering, Columbia University, New York, NY 10027 ‡

Department of Chemistry, Temple University, Philadelphia, PA 19122

7

*Corresponding author. E-mail: [email protected], Telephone: (212)-854-2869

8

Abstract

9

Photosensitized reactions involving imidazole-2-carboxaldehyde (IC) have been

10

experimentally observed to contribute to secondary organic aerosol (SOA) growth.

11

However, the extent of photosensitized reactions in ambient aerosols remains poorly

12

understood and unaccounted for in atmospheric models. Here we use GAMMA 4.0, a

13

photochemical box model that couples gas-phase and aqueous-phase aerosol chemistry,

14

along with recent laboratory measurements of the kinetics of IC photochemistry, to analyze

15

IC-photosensitized SOA formation in laboratory and ambient settings. Analysis of the

16

laboratory results of Aregahegn et al. (2013) suggests that photosensitized production of

17

SOA from limonene, isoprene, α-pinene, β-pinene, and toluene by 3IC* occurs at or near

18

the surface of the aerosol particle. Reactive uptake coefficients were derived from the

19

experimental data using GAMMA 4.0. Simulations of aqueous aerosol SOA formation at

20

remote ambient conditions including IC photosensitizer chemistry indicate less than 0.3%

21

contribution to SOA growth from direct reactions of 3IC* with limonene, isoprene, α-

22

pinene, β-pinene, and toluene, and an enhancement of less than 0.04% of SOA formation

23

from other precursors due to the formation of radicals in the bulk aerosol aqueous phase.

ACS Paragon Plus Environment


24

Other, more abundant photosensitizer species, such as humic-like substances (HULIS),

25

may contribute more significantly to aqueous aerosol SOA production.

26 27

Introduction

28

Aerosols are ubiquitous in the atmosphere and are known to affect human health,

29

climate, and visibility. The composition of atmospheric aerosols can vary greatly, with

30

organic material and liquid water comprising major fractions of aerosol mass.1,2 Current

31

models are unable to accurately predict the amount and composition of organic aerosols

32

observed in the atmosphere, largely due to uncertainties surrounding the production of

33

secondary organic aerosol (SOA).3–5 SOA formation involves the transfer of organic mass

34

from the gas phase to the particle phase, via oxidation of volatile organic compounds

35

(VOCs) followed by partitioning into the organic or aqueous condensed phase. Multiphase

36

reactions in cloud and aerosol water have recently gained attention as important SOA

37

formation pathways.6

38

One aqueous SOA formation precursor that has received considerable attention in

39

the literature is glyoxal, a gas-phase oxidation product of isoprene, toluene, and other

40

biogenic and anthropogenic VOCs. Glyoxal has a high affinity for water and exhibits

41

salting-in, which, given the high ionic strength of many aerosol particles, means that the

42

uptake of glyoxal to aerosols can be very efficient.7,8 The annual global source of SOA

43

from glyoxal has been estimated to be as much as 2.6 Tg a-1.9 This pathway has been

44

estimated to contribute 15% of the total ambient SOA in Mexico City.10 Glyoxal undergoes

45

both oxidative and non-radical chemistry in the aqueous phase, to form products including


Page 2 of 20

Page 3 of 20


46

the nitrogen-containing species imidazole-2-carboxaldehyde (IC).11 Recently, IC was

47

identified as a component of ambient aerosols at concentrations up to 3.2 ng/m3.12

48

IC absorbs light in the UV/visible range and has recently been shown to participate

49

in SOA formation in the role of photosensitizer (Figure 1).13–16

50

photosensitizers,17 the triplet excited state of IC (3IC*) can either: (1) reduce O2 to form

51

oxidizing radicals such as ·OH and ·HO2, or (2) directly oxidize other organic material.

52

Both pathways can contribute to SOA formation.

Like other triplet

53

54 55 56 57

Figure 1- SOA formation from photosensitizers and other organic material in photosensitized reactions

58

In aerosol flow tube experiments performed by Aregahegn et al., significant SOA

59

formation was observed when aerosols containing IC were exposed to isoprene, limonene,

60

α-pinene, β-pinene, and toluene in the presence of UV light.14 No particle growth was

61

observed in these experiments under dark conditions. Mass spectrometry analysis of SOA

62

products showed IC-limonene or limonene-limonene recombination products, as well as

63

limonene oxidation products, suggesting that both pathways of SOA formation from 3IC*



64

were active.14,15 Additionally, Aregahegn et al. observed that SOA growth was limited

65

when oxygen was not present.14

66

Detailed laboratory studies using the laser flash photolysis technique have revealed

67

additional information about the photophysics of 3IC* and the rates and mechanisms of its

68

reactions with O2 and organics in the aqueous phase. Tinel et al. showed the formation of

69

3

70

al. observed the transient triplet-triplet IC absorption and showed that adding increasingly

71

higher concentrations of limonene to an IC solution led to faster decay.15 Our recent work

72

characterized the rates of reaction of 3IC* with O2 and isoprene. Mass spectrometry results

73

showed IC-isoprene adducts but no oxidized isoprene species, suggesting that pathway (2)

74

dominates for the IC-isoprene system.16

IC* under UV radiation and characterized its quenching by I-, Br- and Cl-.13 Rossignol et

75

Despite its potential significance, due to a prior lack of mechanistic and kinetic

76

information for modeling studies, the extent to which photosensitizer chemistry contributes

77

to ambient atmospheric aerosol growth is unclear. Now that mechanisms for the

78

photochemical processing of IC and aqueous phase reaction rates of 3IC* with O2 and

79

organics are available, a modeling study to estimate aerosol growth from photosensitized

80

IC is possible. Here, we use an updated version of GAMMA (Gas-Aerosol Model for

81

Mechanism Analysis), a photochemical box model developed by the McNeill group,18 to

82

investigate the production of SOA material via photosensitized reactions of IC under both

83

experimental and ambient conditions. GAMMA couples detailed gas phase and aqueous

84

aerosol phase chemistry, enabling us to simulate experimental and atmospheric conditions

85

to predict SOA formation.

86


Page 4 of 20

Page 5 of 20


87

Methods

88

Two categories of simulation were carried out in this study: 1) simulation of laboratory

89

aerosol growth experiments by Aregahegn et al.14 and 2) simulation of IC photosensitizer

90

chemistry under ambient conditions. GAMMA 4.0 was used for all simulations discussed

91

in this work.

92 93

GAMMA 4.0

94

The McNeill Group photochemical box model of gas and aqueous aerosol

95

chemistry and interphase mass transfer, GAMMA, features detailed, coupled gas and

96

aerosol/cloudwater aqueous chemistry for organic species.18 Following Schwartz (1986),

97

the temporal evolution of the partial pressure of species i in the gas phase (Pi) is given by:

98

,

,

∗

,

1

99

where aL is the aerosol liquid (aqueous) volume fraction (cm3 cm-3), Ci is the aerosol-phase

100

concentration of species i, Hi* is the effective Henry’s Law constant, rij,gas is the rate of gas

101

phase reaction j that species i participates in, and Ei and Di are the emission and deposition

102

rates of species i, respectively.19 The gas-aerosol mass transfer coefficient for species i,

103

kmt,i, is given by:

104

1 ,

3

,

3

4

2

105

where R is the aerosol particle radius, Dg,i is the gas-phase diffusion coefficient, i is the

106

thermal velocity, and i is the sticking coefficient.

107

The aerosol-phase concentration of species i is described by:



,

108

,

,

∗

3

109

where R is the gas constant, T is temperature, and rij,aq is the rate of aqueous-phase reaction

110

j for species i.

111 112

Chemical mechanism. The gas and aqueous-phase mechanisms of GAMMA are described

113

in McNeill et al. 2012. GAMMA 4.0 includes the gas phase IEPOXOO + NO reaction

114

based on Xie et al. (2013),20 and methacrylic acid epoxide (MAE) chemistry following Lin

115

et al. (2013).21 Updated Henry's Law constants for GAMMA 4.0 include: H* = 3×107

116

M/atm for IEPOX22 and the introduction of salting effects for glyoxal and methylglyoxal.7,8

117

We include the possible protonation of IEPOX (aq) by ammonium as observed by Nguyen

118

et al. (2014), using a rate of 1.7×10-5 1/s*[NH4+] based on our analysis of their laboratory

119

data.22 GAMMA 4.0 includes imidazole photosensitizer chemistry as described in the

120

following sections.

121 122

3

IC* chemistry in GAMMA 4.0

123

3

IC* chemistry was incorporated using a) an explicit bulk aqueous chemistry approach or

124

b) a reactive uptake approach. We note that, although SOA formation is not exclusively an

125

aqueous process, IC formation in aerosols occurs via aqueous processes, and therefore 3IC*

126

chemistry is likely to take place in the aqueous phase.

127

Bulk aqueous chemistry approach. To model the photosensitizer reactions in the

128

experimental study of Aregahegn et al.,14 the kinetics and mechanism determined


Page 6 of 20

Page 7 of 20


129

experimentally by Li et al. (Table 1) were incorporated into GAMMA.16 The assumption

130

that 10% of IC exists in the triplet state at any time is a conservative estimate based on the

131

high triplet yield of aromatic ketones.16,17,23,24 The rate constant for isoprene oxidation by

132

OH (k4) was measured by Huang et al. (2011).25 This reaction was included previously in

133

the original GAMMA mechanism18 but is highlighted here because of its potential role in

134

SOA formation by 3IC* and isoprene. The source of ·OH for the reaction with isoprene in

135

Table 1 is downstream reactions of the ·HO2 produced by the reaction of 3IC* with oxygen.

136

Isoprene, as well as ·OH and ·HO2 generated in-particle were allowed to partition between

137

the gas and aerosol phases following Henry’s Law. All other species in the aerosol

138

(including succinic acid, IC, organic reactive intermediates and products) were assumed to

139

remain in the aqueous phase. IC is assumed to be nonvolatile for these simulations due to

140

large uncertainties in its vapor pressure.14,26 Simulations were run for 19 minutes, matching

141

the experimental conditions of Aregahegn et al.14

142

In a second set of simulations, all conditions were the same as described above, but

143

each particle was split into two separate volumes: an outer shell and an inner core. The

144

volume of each portion of each particle was determined such that the outer shell contained

145

3% of the total IC in the particle, assuming a uniform concentration of IC throughout the

146

particle. In these simulations, all IC reactions were restricted to the outer shell, so 97% of

147

the IC present in the particles did not undergo any reaction listed in Table 1.

148

Each simulation was run twice, with different particle diameters in each simulation.

149

The two particle sizes corresponded to the initial and the final diameters of each experiment

150

of Aregahegn et al.14 However, accounting for the particle size change yielded differences



151

in final SOA growth of less than 0.1%. Therefore, only the results using the initial diameter

152

are reported here.

153 154 155

Table 1- Mechanism and kinetics of photosensitized reactions of IC with isoprene from Li et al. incorporated into GAMMA16 Reaction Kinetics 3 10% of IC is converted to 3IC* IC  IC* 3 k1 = 2.7 × 109/M/s IC* + isop  isop* + IC 3 Assume 50/50 branching IC* + isop  isop-IC k2 = 3.0 × 109/M/s isop + isop*  isop-isop 3 k3 = 3.1 × 109/M/s IC* + O2   HO2 k4 = 1.4 × 1010/M/s OH + isop   isop-OH

156 157 158

Reactive uptake simulations. In another set of GAMMA simulations, a reactive

159

uptake formulation was used for each of the VOCs (limonene, isoprene, α-pinene, β-pinene,

160

and toluene) considered by Aregahegn et al., instead of Henry’s Law uptake and bulk

161

aqueous-phase reaction.14 If a species is selected for surface reactive uptake, the following

162

equation replaces equation (1):

163

164

.

,

Where 1

165

,

167

5

1 ,

166

4

,

with

,

4

3

1

0.38 1

6


Page 8 of 20

Page 9 of 20


168

Here, i is the reactive uptake coefficient, Sa is the aerosol surface area (cm3 cm-2), and Kn

169

is the Knudsen number (Kn =R/, where  is the mean free path of air).

170

The aerosol growth due to these processes can be determined from the reactive

171

uptake coefficient due to the equivalence between the VOC uptake rate and the SOA

172

formation rate. In simulations performed under the conditions of the Aregahegn et al.

173

laboratory study, the reactive uptake coefficient was adjusted for each VOC until the

174

GAMMA results matched the experimentally observed SOA growth.14 The reactive uptake

175

coefficient was then adjusted for ambient conditions using the following relations:27

176 1

1

1

177

7

1⁄

coth

4 where

8

178

In the above equations, Dl is the liquid phase diffusivity, l is the diffuso-reactive

179

length, α is the mass accommodation coefficient, ω is the mean molecular velocity, H is

180

the effective Henry’s Law coefficient, R is the gas constant, T is the temperature, and r is

181

the radius of the particle. Using the reactions that consume aqueous phase isoprene in

182

Table 1, kI , the pseudo-first order loss rate constant for the VOC, can be expressed as

183

∗

∗

∙

184


9


Page 10 of 20

185

Isop* (aq) and ·OH (aq) are short-lived species whose concentrations are negligible

186

compared to that of 3IC*. Thus, Equation (9) can be approximated as

187

∗

10

188 189

Changes in kI due to variations in [3IC*] are the primary reason scaling of  for ambient

190

conditions is necessary.

191 192

Conditions for simulation of experiments. The experimental conditions used by

193

Aregahegn et al. for aerosols containing IC, ammonium sulfate, and succinic acid were

194

simulated in GAMMA.14 The initial aerosol composition was determined by using the ratio

195

of bulk concentrations of ammonium sulfate and succinic acid used by Aregahegn et al.

196

prior to atomizing and the relative humidity in E-AIM, a thermodynamic model which can

197

be used for calculating equilibria for aerosol systems with partitioning.14,28 IC

198

concentrations were scaled from the bulk solution to the aerosol with the calculated

199

succinic acid concentrations from E-AIM. The aerosol density was determined using

200

empirical relationships from Hyvarinen et al.29 These parameters were input into GAMMA

201

as initial conditions. For these experimental conditions, there are no emissions or

202

depositions so Ei and Di are both set to zero. Initial concentrations of all species except IC,

203

ammonium sulfate, the VOCs, succinic acid, water, O2, and H+ were set to zero.

204 205

Ambient aerosol simulations. Ambient calculations for SOA growth were run based on

206

the results of laboratory simulations using reactive uptake of VOCs. Unlike the


Page 11 of 20


207

experimental simulations where only one VOC was present, the ambient simulations

208

simultaneously consider 3IC* induced SOA growth due to isoprene, toluene, limonene,

209

isoprene, α-pinene, and β-pinene, in addition to the SOA pathways originally represented

210

in GAMMA (aqueous aerosol SOA formation by oxidation products of isoprene, acetylene,

211

toluene, cresol, and xylenes).18 The initial concentrations of species in both the gas and

212

aerosol phase are based on concentrations typically observed in the atmosphere (see

213

Supplementary Material). IC concentrations were based on the observations of Teich et

214

al.12 The aerosol pH was set constant to 1, 2, 3, or 4 for each simulation. The relative

215

humidity was also assumed to be constant at 45% for each simulation. Reactive uptake

216

coefficients determined from the experimental aerosol simulations were scaled down based

217

on Equations (7) and (8) due to the difference in concentrations of IC used by Aregahegn

218

et al. and observed in ambient conditions, as discussed above.14 A constant reactive uptake

219

coefficient, based on the initial condition of IC, was assumed, providing an upper bound

220

estimate of the SOA production from 3IC*-VOC reactions. The production of HO2 in the

221

aerosol bulk by 3IC* was simulated based on Reaction 3 (Table 1), so that the contribution

222

of this pathway to SOA formation from aqueous-phase precursors other than isoprene,

223

toluene, limonene, isoprene, α-pinene, and β-pinene could be evaluated. The aerosol

224

density was assumed to be constant throughout the simulation.

225 226

Results and Discussion

227

GAMMA simulations of bulk aerosol reactions under experimental conditions. The

228

first set of simulations used GAMMA 4.0 with the bulk 3IC* mechanism to simulate SOA

229

growth using experimental conditions of the isoprene-IC experiments of Aregahegn et al.



230

and the mechanism from Li et al. (Table 1).14,16 The results, shown in Table 2, indicate that

231

these simulations greatly overestimate the amount of SOA formation as observed

232

experimentally. Additionally, this set of simulations also indicates that over 99% of the IC

233

in the system reacts within the 19-minute simulation. In these simulations, SOA formation

234

was largely insensitive to the assumed percentage of IC in the excited triplet state; for 3IC*

235

varying between 1% and 100% of the total IC, the calculated SOA formation varied less

236

than 0.4%. Predicted SOA formation also changed by less than 0.2% when the assumed in-

237

particle concentration of O2 was varied over six orders of magnitude. Since nearly all of

238

the IC reacts, the simulation results correspond to an upper bound for SOA growth due to

239

IC, as IC is limiting for this scenario. This suggests that not all of the 3IC* in the aerosols

240

used by Aregahegn et al. participates in the reactions shown in Table 1 (note that we assume

241

that only 10% of available IC is in the excited state for the base case, following Li et al.).16

242

For the shell simulations, in which only 3% of the total 3IC* is available for reaction, SOA

243

growth is consistent with particle growth observed by Aregahegn et al.14

244 245 246

Table 2- Comparison of predicted SOA growth from GAMMA simulations of experiments of Aregahegn et al. to experimentally observed SOA growth for isoprene14 GAMMA simulation Predicted SOA growth Ratio of predicted growth from GAMMA (ng/m3 air) from GAMMA to experiment (Aregahegn et al. 2013) Bulk 1.9 34 3% Shell 0.05 1.0

247 248

We therefore consider possible diffusion limitations in this system by calculating a diffuso-

249

reactive length for incoming VOC molecules, defined as

250


Page 12 of 20

Page 13 of 20


11 251 252

where Dl is the liquid-phase diffusivity and kI is the pseudo-first order loss rate constant

253

(eq 10).27 We calculate l < 1 nm for the experimental conditions of Aregahegn et al.,

254

significantly less than 24.8 nm, the radius of the particles used in the experiments.14 Thus,

255

isoprene is not expected to diffuse far into the particle before reacting, suggesting that the

256

photosensitizer reaction occurs at or near the surface of the particle. Consequently, only

257

3

258

simulations. Although 3IC* can also react with O2 in the bulk of the aerosol to form

259

radicals, the reaction of radicals with isoprene is still limited to the surface or near the

260

surface due to the rapid reaction between isoprene and 3IC*. Thus, despite possible

261

reactivity of IC throughout the entire aerosol, this contribution to SOA mass in the

262

experimental scenario only occurs at or near the surface.

IC* near the surface will react with isoprene, as predicted by the previous GAMMA

263 264

GAMMA simulations of reactive uptake of VOCs. As an alternative approach, a reactive

265

uptake formulation was used to model SOA growth in the experiments of Aregahegn et al.

266

(see the Methods section for details). The reactive uptake coefficient quantifies the

267

probability of irreversible reaction upon collision between a gas phase molecule and the

268

aerosol particle and inherently includes uptake into the particle and reaction. Reactive

269

uptake coefficients consistent with the experimental data of Aregahegn et al. were

270

determined for each VOC using GAMMA, shown in Table 3.14 As expected, VOCs that

271

form SOA more efficiently have higher reactive uptake coefficients.



Page 14 of 20

272 273

Table 3- Calculated reactive uptake coefficients for experiments of Aregahegn et al.14 Aregahegn et al. (2013) Using GAMMA VOC VOC concentration Diameter Reactive uptake (ppm) growth factor coefficient Limonene 1.8 27.6% 3.1 × 10-5 Isoprene 4 3.13% 1.4 × 10-6 α-pinene 63 19.35% 4.9 × 10-7 β-pinene 63 16.5% 3.9 × 10-7 Toluene 352 3.5% 1.3 × 10-8

274 275

Ambient aerosol GAMMA simulations. The reactive uptake coefficients which were

276

successfully determined from experimental data were scaled for ambient aerosol

277

calculations using Equations (7) and (10). Some uncertainty exists in the appropriate value

278

for the mass accommodation coefficient, , for organic gas molecules, with reported values

279

ranging from ~10-4< 

Modeling Photosensitized Secondary Organic Aerosol Formation in

Recommend Documents