Particle Partitioning and Mass

Subscriber access provided by - Access paid by the | UCSB Libraries

Article

Direct measurements of gas/particle partitioning and mass accommodation coefficients in environmental chambers Jordan Edward Krechmer, Douglas A. Day, Paul J. Ziemann, and Jose L. Jimenez Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b02144 • Publication Date (Web): 31 Aug 2017 Downloaded from http://pubs.acs.org on September 3, 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 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Environmental Science & Technology

Direct measurements of gas/particle partitioning and mass accommodation coefficients in environmental chambers Jordan E. Krechmer%, Douglas A. Day, Paul J. Ziemann, Jose L. Jimenez* Cooperative Institute for Research in Environmental Sciences (CIRES) and Department of Chemistry and Biochemistry, Boulder, Colorado 80309, United States Corresponding author: * Jose-Luis Jimenez University of Colorado; UCB 216, Boulder, CO 80309-0216 Phone: 303-492-3557 Fax: 303-492-1149 [email protected] Keywords: Aerosols, SOA, environmental chambers, gas/particle partitioning Abstract Secondary organic aerosols (SOA) are a major contributor to fine particulate mass and wield substantial influences on the Earth’s climate and human health. Despite extensive research in recent years, many of the fundamental processes of SOA formation and evolution remain poorly understood. Most atmospheric aerosol models use gas/particle equilibrium partitioning theory as a default treatment of gas-aerosol transfer, despite questions about potentially large kinetic effects. We have conducted fundamental SOA formation experiments in a Teflon environmental chamber using a novel method. A simple chemical system produces a very fast burst of lowvolatility gas-phase products, which are competitively taken up by liquid organic seed particles and Teflon chamber walls. Clear changes in the species time evolution with differing amounts of seed allow us to quantify the particle uptake processes. We reproduce gas- and aerosol-phase observations using a kinetic box model, from which we quantify the aerosol mass accommodation coefficient (α) as 0.7 on average, with values near unity especially for low volatility species. α appears to decrease as volatility increases. α has historically been a very difficult parameter to measure with reported values varying over three orders of magnitude. We use the experimentally-constrained model to evaluate the correction factor (Φ) needed for chamber SOA mass yields due to losses of vapors to walls as a function of species volatility and particle condensational sink. Φ ranges from 1 – 4.

1 ACS Paragon Plus Environment


Page 2 of 31

37

Introduction

38

Gas/particle partitioning (GPP) is a fundamental physical process describing the interaction

39

between atmospheric organic gases and particles. Originally treated as an adsorptive process,1,2

40

most atmospheric models currently treat GPP of organic compounds as an absorptive process.3–5

41

In this interpretation, semivolatile organic gases condense into and evaporate from particles until

42

quickly reaching a volatility- and available organic aerosol mass-dependent equilibrium.3,4

43

Semivolatile equilibrium partitioning is often parameterized using a volatility basis set, which

44

lumps the amount of condensable material in the air in multiple bins spaced typically by a decade

45

of the saturation mass concentration of the compound.5 GPP has been incorporated into

46

numerous box, regional, and global models6–10 as a default treatment of organic aerosol

47

formation and evolution.

48

Still, modeling of GPP by rapid absorptive equilibrium remains controversial. There are

49

now several real-time analysis techniques capable of measuring the gas/particle distribution of

50

trace organic compounds at relatively high time resolution in the field.11–14 Many of these

51

measurements have shown substantial disagreements with semivolatile equilibrium partitioning

52

theory and one another, however.15,16 Some recent works have even suggested that semivolatile

53

equilibrium partitioning is not reached under certain conditions due to strong kinetic limitations,

54

and more detailed treatments are required to match atmospheric observations.17–21 GPP model treatment is expected to have a large influence on secondary organic aerosol

55 56

(SOA)

mass

concentrations

predicted

by

atmospheric

models.

Most

models

use

57

parameterizations of the SOA yield, i.e. the mass of SOA formed per unit mass of volatile

58

organic compound (VOC) precursor reacted.22 Almost all experiments quantifying SOA mass

59

yields that have been used in models were performed in Teflon-walled environmental


Page 3 of 31


60

chambers.22–26 However, it has recently been reported that Teflon chambers suffer from losses of

61

semivolatile gaseous compounds to chamber walls.27–30 Wall losses can have strong effects on

62

aerosol mass yields,30 and could also affect conclusions about chemical composition and

63

processes gleaned from chamber experiments. The question even arises: if wall losses are

64

substantial for a certain range of compounds, and fast enough to be on similar time scales as the

65

approach to equilibrium GPP, is it possible to observe and accurately characterize GPP kinetics

66

and equilibrium in environmental chambers?

67

In mathematical representations of the kinetics of GPP, the mass accommodation

68

coefficient (α; also known as the sticking coefficient; and equal to the evaporation coefficient at

69

equilibrium) is a critical parameter that defines the fraction of gas/particle collisions that result in

70

a colliding species being taken up by a particle. α for organic species was historically assumed to

71

be 1, mostly from liquid evaporation measurements.31–33 More recent measurements of mass

72

accommodation coefficients for aerosol-phase compounds34–36 range across two orders of

73

magnitude. A very wide range of values (as low as 0.001) has been used in models and data

74

fitting, sometimes as a tuning parameter which might obscure other model structural

75

limitations.30,37–40 Comparing these literature values is complicated by the different methods used

76

to derive α. Some models34,41 and measurements34,35 conflate particle diffusion kinetics and

77

surface accommodation as one effective mass accommodation coefficient. Other models have

78

attempted to specifically separate mass accommodation and particle-phase diffusion

79

coefficients.36,42,43

80

Given the discrepancies that persist between ambient SOA measurements and

81

models,44,45 the substantial fraction of organic material in the atmosphere which is thought to be

82

semivolatile,46 and the difficulty of quantifying GPP in field studies,16 it is important to closely



Page 4 of 31

83

examine in detail the kinetics of approach to equilibrium GPP. . It is also important to further

84

evaluate the impact that losses of vapors to Teflon chamber walls may have in SOA yield

85

measurements. Herein, we conduct fundamental gas/particle partitioning experiments in an

86

environmental chamber with a particular focus on GPP kinetics. A well-characterized simple

87

chemical system is used to produce low-volatility organic compounds very rapidly, which are

88

taken up by liquid organic seed particles and/or the Teflon chamber walls. We use liquid organic

89

seed aerosol specifically to establish an experimental method under which particle-phase kinetic

90

limitations are not present. Both gas-phase products and total aerosol volume and surface area

91

concentrations are continuously monitored. Analysis of these experiments allow us to thoroughly

92

analyze the gas-phase and accommodation kinetics of GPP in the presence of walls for semi- and

93

low-volatility gas-phase compounds. A box model is used to quantify c*and α. We use liquid

94

aerosol particles as seeds, which removes particle-phase diffusion as a limiting parameter and

95

allows us to infer the true value of α. We discuss the implications for quantifying gas/particle

96

partitioning under more complex conditions.

97 98

MATERIALS AND METHODS

99

Environmental chamber experiments

100

A schematic outlining the chamber and instruments used in this work is shown in Figure

101

1. We conducted all of our experiments in the one of the two chambers in the CU Environmental

102

Chamber (CUEC) Facility. The CUEC bag was constructed of FEP Teflon with a maximum

103

volume of 20 m3. The enclosure temperature was maintained at 26 ± 1° C for all experiments.

104

The chambers were filled with clean, dry (< 1% RH) air from two AADCO Model 737-15 clean

105

air generators (Cleaves, OH, USA). The chambers were equipped with an automated flushing


Page 5 of 31


106

system that alternately evacuated and refilled the bag when it was not in use (~5% of full volume

107

every 5 minutes), while maintaining the bag volume at ~95% of maximum volume using

108

differential pressure between the bag and climate-controlled enclosure. Running the automated

109

flushing sequence overnight before each experiment ensured that each experiment started with

110

aerosol concentrations of 0 particles cm-3 as measured with an ultrafine condensation particle

111

counter (UCPC, TSI model 3776 with a diameter cutoff of 2.5 nm). The experimental procedure is similar to the one detailed in Krechmer et al.27 to measure

112 113

vapor wall losses to Teflon walls, but with the addition of seed aerosol. Dioctyl sebacate (DOS)

114

seed was chosen for this set of experiments because it has been shown19 to be reliably liquid

115

under the experimental conditions used here and should not impose particle-phase diffusion

116

limitations on time scales relevant to the experiment. DOS also has a low vapor pressure (2 × 10-

117

6

118

reaction, thus not affecting GPP through particle-phase reactions.

119

Pa)47 and, as a diester, should be unreactive with any of the products formed in the OH˙

First, we injected dioctyl sebacate (DOS; >97%; Sigma-Aldrich; saturation concentration

120

at 298 K (c*) < 0.1 µg m-3)48 liquid seed aerosol into the chamber using an evaporation-

121

condensation generator modeled after Sinclair and Lamer.49 The aerosol generator was

122

configured to seed the chamber to 100 µg m-3 of DOS in ~15 minutes. Particle-phase wall losses

123

occurred on a ~10 hr timescale, but it was not necessary to account for them explicitly in this

124

model since the time-dependent aerosol size distribution measurements were used to constrain

125

the box model. If needed, we diluted the chamber with clean air until the DOS seed surface area

126

reached a desired level. '-condensation generator was tuned to produce aerosol with a lognormal

127

surface area distribution centered at 200 nm. A scanning mobility particle sizer (SMPS)

128

comprised of a TSI model 3080 electrostatic classifier and model 3775 condensation particle



129

counter monitored aerosol number, surface area, and volume concentrations. The SMPS scan

130

was programmed to last 120 s, with 15 s to return to the starting voltage, resulting in a time

131

resolution of 135 s. We provide a summary of different experiments and seed concentrations in

132

Table S1.

133

Page 6 of 31

The chemical system used here was the oxidation of long-chain alcohols by the OH

134

radical under high-NO conditions. This system is the same used by Krechmer et al.27 because the

135

oxidation products are simple and well-characterized (and with consistent chemical functionality

136

in a progression of carbon numbers) as opposed to those of other gaseous precursors such as α-

137

pinene or isoprene.

138

After the seed aerosol concentration stabilized, we injected 50 µL of 1-hexanol, 40 µL of

139

1-octanol, 25 µL of 1-nonanol, 18 µL of 1-decanol, and 12 µL of 1-dodecanol (all >99% purity;

140

Sigma-Aldrich) into the chamber by gently heating the liquids into a clean nitrogen stream (UHP

141

liquid nitrogen evaporate) flowing into the chamber. Injection took ~5 minutes. This resulted in a

142

total VOC concentration in the chamber of 1.4 ppm. We fully mixed the bag contents by turning

143

on a Teflon-coated fan inside the bag for 1 min to mix the contents. Finally, we injected 4 ppm

144

of NO (99%, Matheson Tri-gas) and 4 ppm of methyl nitrite (MeONO, synthesized after the

145

method of Taylor et al.50) into the chamber as pure gases using a small glass bulb and mixed the

146

contents again for 1 min.

147

To initiate chemical production, we turned on the chamber UV black lights (~300-400

148

nm) at 100% intensity (JNO2 = 0.01 s-1) for precisely 10 s via computer-control. MeONO formed

149

OH radicals, which then oxidized the alkanols forming ppt-levels of several oxidized products,27

150

including hydroxynitrates (HN), dihydroxynitrates (DHN), trihydroxynitrates (THN), and

151

carbonyl dihydroxynitrates (CDHN). A list of the compounds, along with their exact masses and


Page 7 of 31


152

estimated saturation concentrations (c*) is provided in Table S2. We did not observe new particle

153

formation. We were able to continuously monitor the latter three types of products with a

154

chemical ionization mass spectrometer (CIMS). This “rapid burst” method, developed

155

previously,27 is critical to these experiments because it produces compounds over a wide range of

156

volatilities over the entire volume of the chamber much faster than the vapors can equilibrate

157

with the particles and walls. After each burst, the system was left to come to equilibrium for at

158

least 1 hr.

159

The amount of aerosol produced by each photooxidation burst comprised an average of

160

7% of the total aerosol volume in the chamber (Figure S1). Therefore, the system can be

161

modeled assuming that the gas-phase compounds are partitioning into pure DOS particles and

162

not into a product compound mixture. The specific molecules studied here are estimated to form

163

~ 20 ppt and account for < ~5% of the SOA. The majority of the aerosol formed is likely to be

164

semivolatile hydroxynitrates, which are not detectable by the nitrate CIMS.27 The commercial

165

iodide CIMS, which would likely detect the majority of aerosol-forming compounds, suffers

166

from a significant time delay for semivolatile products due to gas-wall partitioning in the

167

instrument source.27 This makes it unsuitable for a study such as this where it is critical to

168

measure the gas-phase decays as close to real time as possible.

169

NO3-CIMS

170

A chemical ionization time-of-flight mass spectrometer equipped with a nitrate ion

171

ionization source (NO3-CIMS) monitored gas-phase product concentrations. The instrument51,52

172

and nitrate anion source53 have been described extensively in previous works. We used the

173

instrument in a configuration similar to the one in Krechmer et al.27. Notably here, we placed the

174

instrument inside the CUEC enclosure directly adjacent to the reactor bag. This allowed us to use



Page 8 of 31

175

a 0.6 m long electropolished stainless steel inlet to isothermally bring 10 standard L min-1

176

(defined as 25° C, 1 Atm) of sample air directly from the bag to the instrument without dilution.

177

The short inlet with a sub-sampled center flow minimizes wall losses and results in a residence

178

time of sampled air in the instrument inlet and source of < 4 s. The NO3-CIMS acquired spectra

179

at 1 Hz. We then processed instrument data using Tofware (Tofwerk, AG and Aerodyne, version

180

2.5.8) toolkit within IGOR Pro 6 (Wavemetrics, Inc.) at the same time resolution. Throughout

181

this work we report CIMS results in units of ions s-1 because we are only concerned with relative

182

changes in gas-phase concentration and the CIMS response has previously been shown to be

183

linear.54,55 The NO3-CIMS detected gaseous product molecules as clusters with the nitrate ion,

184

but the NO3- prefix has been removed from reported formulas throughout this paper for clarity.

185

An example mass spectrum of some of the product compounds is provided in Figure S2.

186

Box model

187

A chemical kinetic box model was used to simulate the behavior of gas- and aerosol-

188

phase compounds, accounting for partitioning and evaporation of each oxidized products to walls

189

and particles. A schematic of the model is shown in Figure S3. The model was solved using

190

KinSim v3.24 within IGOR Pro.56

191

The vapor wall loss or condensation rate coefficient, kc was determined using the

192

procedure of Krechmer et al.,27 the results of which are shown in Figure S4. Under these

193

conditions (c*

Particle Partitioning and Mass

Recommend Documents