Molecular-Size-Separated Brown Carbon ... - ACS Publications

Feb 15, 2017 - Airmass back trajectories (72 h) from the St. John,s sampling site from July 6, ... described in detail in Di Lorenzo and Young.21 Brie...
0 downloads 0 Views 2MB Size
Subscriber access provided by University of Newcastle, Australia

Article

Molecular size separated brown carbon absorption for biomass burning aerosol at multiple field sites Robert A Di Lorenzo, Rebecca A. Washenfelder, Alexis Rae Attwood, Hongyu Guo, Lu Xu, Nga Lee Ng, Rodney J. Weber, Karsten Baumann, Eric S. Edgerton, and Cora J Young Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b06160 • Publication Date (Web): 15 Feb 2017 Downloaded from http://pubs.acs.org on February 16, 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

Environmental Science & Technology

1

Molecular size separated brown carbon absorption

2

for biomass burning aerosol at multiple field sites

3

Robert A. Di Lorenzo1, Rebecca A. Washenfelder2.3, Alexis R. Attwood2,3,†, Hongyu Guo4, Lu Xu5,

4

Nga L. Ng4,5, Rodney J. Weber4, Karsten Baumann6, Eric Edgerton6, Cora J. Young1*

5

1

Department of Chemistry, Memorial University, St. John’s, Newfoundland, Canada.

6

2

Cooperative Institute for Research in Environmental Sciences, University of Colorado Boulder,

7

Boulder, Colorado, USA.

8

3

9

Atmospheric Administration, Boulder, Colorado, USA.

Chemical Sciences Division, Earth System Research Laboratory, National Oceanic and

10

4

11

USA.

12

5

13

Georgia, USA

14

6

15

School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, Georgia,

School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta,

Atmospheric Research & Analysis Inc., Cary, North Carolina, USA



Now at Droplet Measurement Technologies, Boulder, Colorado, USA

16

ACS Paragon Plus Environment

1

Environmental Science & Technology

Page 2 of 31

17

Abstract Biomass burning is a known source of brown carbon aerosol in the atmosphere. We

18

collected filter samples of biomass burning emissions at three locations in Canada and the United

19

States, with transport times of 10 h to >3 days. We analyzed the samples with size exclusion

20

chromatography coupled to molecular absorbance spectroscopy to determine absorbance as a

21

function of molecular size. The majority of absorption was due to molecules >500 Da, and these

22

contributed an increasing fraction of absorption as the biomass burning aerosol aged. This

23

suggests that the smallest molecular weight fraction is more susceptible to processes that lead to

24

reduced light absorption, while larger molecular weight species may represent recalcitrant BrC.

25

We calculate that these large molecular weight species are composed of more than 20 carbons

26

with as few as two oxygens, and would be classified as extremely low volatility organic

27

compounds (ELVOCs).

28

1. Introduction

29

The role of atmospheric aerosol in the global radiative budget is important, but poorly

30

quantified1. Direct aerosol absorption is responsible for positive radiative forcing, in contrast to

31

negative radiative forcing (RF) impacts of light scattering and indirect cloud effects. Although

32

black carbon (BC) dominates carbonaceous particle absorption, organic aerosol (OA) may also

33

include absorbing species that absorb light and are described as brown carbon (BrC). Compared

34

to BC, BrC is known to have greater chemical functionality, water solubility, and absorption that

35

increases rapidly in the near ultraviolet spectral region, with a λ-2 to λ-6 dependency, but its

36

molecular composition is poorly characterized2,3. After accounting for the absorption and lensing

37

properties of BrC, model-based RF estimates are positive and more closely aligned with

38

experimentally-constrained estimates4. Only recently have BrC properties determined from

39

experimental data been included in global radiative forcing models4–7. Many primary and

ACS Paragon Plus Environment

2

Page 3 of 31

Environmental Science & Technology

40

secondary sources for BrC have been proposed, and the majority of mechanistic laboratory

41

studies have focused on secondary organic aerosol (SOA) formation and processing8. Potential

42

BrC sources include various multi-phase polymerization mechanisms from gas-phase sources9–11,

43

SOA generated from processing of toluene12 and other precursors13, and production from

44

combustion sources14. Improved knowledge of BrC sources, composition, and optical properties

45

is necessary for accurate global models.

46

A few spectroscopic detection methods exist for field measurements of BrC.

First,

47

photoacoustic spectroscopy or aethalometry can directly measure total absorption by an aerosol

48

sample. BrC absorption can be inferred from measurements at multiple wavelengths, with the

49

assumption that absorption at visible wavelengths can be attributed to BC, while absorption in

50

the near-UV is due to both BrC and BC15–17. Second, because BrC is more volatile than BC, BrC

51

can be selectively removed by thermal denuding.

52

ringdown spectrometer, or broadband cavity enhanced spectrometer can be used to measure the

53

difference between a denuded and undenuded sample18,19. Third, a Particle-into-Liquid Sampler

54

(PILS) can be coupled to a liquid waveguide capillary cell (LWCC) and an absorbance

55

spectrometer20. A similar approach uses offline filter extracts where solubility differences

56

between BC and BrC are exploited, such that only soluble aerosol components are dissolved and

57

measured using absorbance spectrophotometry. The BrC content is inferred from the average

58

absorbance in the UV, such as between 360 and 370 nm (Abs365)20. None of these optical

59

methods provide any direct information about the chemical composition of BrC.

A photoacoustic spectrometer, cavity

60

Our group recently developed a method for offline determination of BrC absorption for

61

molecules from 250-75,000 Da using size exclusion chromatography and molecular absorbance

62

spectroscopy21. This method allows quantification of BrC absorbance as a function of molecular

ACS Paragon Plus Environment

3

Environmental Science & Technology

Page 4 of 31

63

weight, and provides new information about BrC composition. Our previous results showed BrC

64

in a biomass burning (BB) plume was primarily composed of large molecular weight species21.

65

This is consistent with observations that BrC from BB consists of extremely low volatility

66

organic compounds (ELVOCs)22. A recent field study also demonstrated that large molecular

67

weight organics that cannot be removed by thermal denuding at 250 °C are major contributors of

68

BrC in an aged biomass-burning plume23,24. The composition of BrC from BB is important

69

because field measurements have shown that water-soluble organic aerosol absorption is

70

predominantly correlated with BB in the southeastern U.S., despite the larger mass contribution

71

of biogenic SOA252518,25–29.

72

In this work, we apply our method to analyze filters from multiple locations with different

73

aerosol sources, thereby determining whether the molecular weight distribution of absorbing

74

species is consistent across BB plumes derived from wildfires, and whether these distributions

75

are consistent in regions not directly influenced by discrete BB events. Less aged BB aerosol was

76

collected during a wildfire event in Vancouver, British Columbia in July 2015, aged BB aerosol

77

was collected in St. John’s, Newfoundland and Labrador in July 2013, and background aerosol

78

was acquired during the Southern Oxidant and Aerosol Study (SOAS) in Alabama during June

79

and July of 2013 (Figure 1). We use these results together with the extensive instrumentation

80

available at the SOAS field site to gain information about the sources, the molecular species that

81

compose, and the mechanisms involved in the formation of BrC.

82

2. Methods

83

2.1 Sampling Sites

84

2.1.1 Less Aged Biomass Burning Aerosol in Vancouver, British Columbia

ACS Paragon Plus Environment

4

Page 5 of 31

Environmental Science & Technology

85

Less aged BB emissions were collected during summer 2015 as part of the Lower Fraser

86

Valley Air Quality Monitoring Network in British Columbia, Canada. These samples capture

87

early July wildfire events in nearby temperate coniferous forests, where PM2.5 concentrations

88

exceeded 200 µg/m3 (Figure S2). Samples were obtained from two sites: Burnaby/Kensington

89

Park (BKP) (49.2792°N, 122.9707°W, 110 m above sea level) and North Vancouver/Second

90

Narrows (NVSN) (49.3015°N, 123.0204°W). Vancouver BKP is in an urban neighbourhood near

91

residential, industrial, commercial and parkland areas. Vancouver NVSN is in a commercial and

92

industrial area, within 50 m of a chemical plant known to emit NOx. Both sites were

93

approximately 100 km east of the main burn location.

94

Aerosol samples were collected using real-time beta attenuation particle monitors (5030

95

SHARP Monitor at Vancouver BKP, 5030i SHARP Monitor at Vancouver NVSN, Thermo

96

Fisher Scientific, Waltham, MA, USA). Particles were collected at 16.67 L/min on glass

97

microfiber filter tape at 8 h intervals. Filter tape was stored in the dark at -20 °C until analysis.

98

2.1.2 Aged Biomass Burning Aerosol in Newfoundland, Canada

99

Aged BB aerosol samples were collected in St. John’s, Newfoundland (47.572°N, 52.722°W,

100

42 m above sea level) two days after a large-scale forest fire erupted in northern Quebec and

101

Labrador, Canada on 4 July 2013. Sample collection is described in detail in Di Lorenzo and

102

Young21. Briefly, PM2.5 samples (n=6) were collected in parallel on 47 mm quartz fibre filters

103

(Pall Life Sciences, Washington, NY, USA) using a multi-channel, medium volume air sampler

104

(URG-3000ABC, URG Corp., Chapel Hill, NC, USA) for 25.5 h during the BB plume intrusion.

105

Filters were stored in the dark at -20 °C until analysis.

106

2.1.3 Regional Background of Aged Biomass Burning Aerosol in the southeastern U.S.

ACS Paragon Plus Environment

5

Environmental Science & Technology

Page 6 of 31

107

Background air samples were collected over six weeks during summer 2013. PM2.5 samples

108

were collected as part of SOAS in the Talladega National Forest in Central Alabama

109

(32.90328°N, 87.24994°W, 76.5 m above sea level). Samples represent typical rural air masses

110

characterized by biogenic emissions of surrounding forests and occasional intrusion from

111

neighboring urban areas of Birmingham and Tuscaloosa26. Forests are primarily composed of

112

pine and hardwoods with isoprene emissions representative of the southeastern US30.

113

PM2.5 samples were collected on the back filter assembly of a custom-built broadband cavity

114

enhanced spectrometer (BBCES)31 on 47 mm PTFE filters (Pall Life Sciences, Washington, NY,

115

USA) housed in a molded PFA 47 mm filter assembly (Savillex Corp., Eden Prairie, MN, USA)

116

across two channels. Each channel was sampled at 2035 standard cm3 min-1 for 3, 12, or 24 h

117

through a 6m copper inlet and conductive silicone tubing within the instrument. The earliest

118

samples (10 - 13 June 2013) may have been affected by contamination from the off-gassing of

119

siloxanes from the black silicone tubing used in the instruments32,33, based on the absorbance of

120

these filters at 300 nm34, and these three filter samples were excluded from analysis.

121

These measurements were co-located with the Electric Power Research Institute Southeastern

122

Aerosol Research and Characterization (SEARCH) Network, where a suite of gas- and particle-

123

phase measurements were performed: CO by non-dispersive infrared (NDIR) detection, K+ by

124

the Monitor for Aerosols and Gases in Ambient Air (MARGA)35, and OA fractions by high

125

resolution time-of-flight aerosol mass spectrometer (HR-ToF-AMS)36. Particle absorption was

126

also measured in real time using PILS-LWCC for water soluble OA absorption20 and multi-angle

127

absorption photometer (MAAP) for BC absorption37.

128

2.2 Fire identification and transport time

ACS Paragon Plus Environment

6

Page 7 of 31

Environmental Science & Technology

129

Fire locations and burn areas (Figure 1) were obtained from the British Columbia Wildfire

130

Service38 and the Canadian Wildland Fire Information System39 for Vancouver and St. John’s

131

samples, respectively. Fires occurring within seven days prior to sampling time were included.

132

Airmass back trajectories computed using the Hybrid Single Particle Lagrangian Integrated

133

Trajectory Model (HYSPLIT) are shown as grey traces of Figure 140. Back trajectories were

134

calculated from the ground-level sampling location, using archived Global Data Assimilation

135

System (GDAS1) meteorology. Vertical motion was computed using the vertical velocity field

136

from the meteorological data. New trajectories were computed every 1 hour for 48 and 72 hours

137

for Vancouver and St. John’s locations, respectively.

138

2.3 Sample Extraction and Analysis by Size Exclusion Chromatography Absorption

139

Spectrophotometry (SEC-UV)

140

Each filter was sub-sampled using a 10 mm arch punch and placed in a glass vial sealed with

141

an acid washed cap containing a Teflon-lined septum. 750 µL of ultrapure deionized water

142

(Barnstead Infinity Ultrapure Water System, Thermo Scientific, Waltham, MA, USA) was added

143

and samples sonicated for 45 minutes. For SOAS samples, seven punches from each filter were

144

combined per extraction. For Vancouver samples, one punch from each filter spot was used per

145

extraction. Due to the hydrophobicity of the SOAS PTFE filters, a 38 mm, 20-gauge needle (BD

146

Precision Glide, Fisher Scientific, Waltham, MA, USA) was inserted through the septa to keep

147

filter media below water surface during sonication, and the exposed needle opening was wrapped

148

in parafilm to prevent contamination. After sonication, all extracts were filtered with 0.2 µm

149

PTFE syringe filters (Iso-Disc, Supelco, Bellefonte, PA, USA), transferred to pre-muffled glass

150

sample vials and stored at 4 °C until analysis.

ACS Paragon Plus Environment

7

Environmental Science & Technology

Page 8 of 31

151

Samples were analyzed according to the procedure outlined in Di Lorenzo and Young21.

152

Briefly, samples were separated using high performance liquid chromatography (HPLC) (1260

153

Infinity, Agilent Technologies, Santa Clara, CA, USA) on an aqueous gel filtration column

154

(Polysep GFC P-3000, Phenomenex, Torrance, Ca, USA) at 1 mL min-1 with 25 mM aqueous

155

ammonium acetate eluent. Compounds were detected using a diode array detector (1260, Agilent

156

Technologies, Santa Clara, CA, USA). Integrations of total absorbance were performed at 300,

157

365 and 405 nm. Conversion of absorption peak areas (units of AU·min) to absorption

158

coefficients (units of Mm-1), as well as blanks and quality control, can be found in the supporting

159

information (Text S1).

160

3. Results and Discussion

161

3.1 Calculated Ages for Biomass Burning Aerosol

162

Although the back trajectories and transport can be uncertain, they give a useful indication of

163

transport age for the BB samples (Figure 1). Plume ages were estimated by determining the

164

HYSPLIT-calculated age at which a plume traversed over a likely fire source. Back trajectories

165

for Vancouver NVSN/BKP sites show aerosol transport times between 8 and 30 h. Airmasses

166

sampled at the St. John’s site were more aged, with transport times between 48 and 72 h.

167

Finally, the aerosol at the SOAS field site was significantly aged, and was generally not

168

associated

169

Spectroradiometer (MODIS) fire locations and back trajectories showed the SOAS site was not

170

influenced by large, discrete fires, and more likely was affected by a regional background of

171

small agricultural fires41. In this case, aerosol age is difficult to quantify, but we estimate a value

172

of up to 5 days from determining the aerosol lifetime over terrestrial southeastern US with

173

HYSPLIT back trajectories.

with

identifiable

fires.

Examination

of

Moderate

Resolution

Imaging

ACS Paragon Plus Environment

8

Page 9 of 31

174

Environmental Science & Technology

3.2 BrC Absorbance as a Function of Molecular Weight

175

Absorbance density plots as a function of molecular weight of BrC for less aged

176

subalpine/montane wildfire, more aged boreal plume, and typical background rural aerosol

177

sample are shown in Figure 2. Absorption density plots show marked similarity in molecular

178

weight profile distribution across sample locations. Molecular weights of absorbing species were

179

estimated using the calibration method outlined previously21 (Text S2). Samples from each site

180

contain three apparent populations with consistent absorbance maxima, all occurring at 2400 Da,

181

1400 Da and 250 Da. Each sample shows a maximum observable molecular weight of 10,000

182

Da. Further, this molecular weight absorbance profile distribution remained constant throughout

183

all days of the SOAS campaign with measurable absorbance (Figure S3). The only day in which

184

the molecular weight profile was different was 11 June 2013, which may be a result of siloxane

185

contamination, as described above.

186

Previous studies reported the amount of BrC emitted from BB depends on burn conditions,

187

with a smaller influence from fuel type22. Although the two wildfire (Vancouver and St John’s)

188

samples are of different origin, age and fuel source (boreal vs. temperate coniferous forests), they

189

have similar absorption characteristics. This suggests that although the quantity of BrC produced

190

is variable depending on the burning event, molecular characteristics responsible for absorption

191

of BB-derived BrC may be constant.

192

To assess the relative absorbance and aging of BB aerosol from each fire, the ratio of BrC to a

193

conserved CO tracer (∆BrC/∆CO) was plotted as a function of aerosol transport time in Figure

194

3A42. Background CO values were calculated from average ground-level CO measurements

195

made at the same site 8 h before plume intrusion. These background levels were consistent

196

within 0.2 ppb across days with little anthropogenic input. Background CO concentrations were

ACS Paragon Plus Environment

9

Environmental Science & Technology

Page 10 of 31

197

calculated as 0.15, 0.11 and 0.14 parts per million by volume (ppmv) for the Vancouver BKP,

198

Vancouver NVSN and St. John’s sites, respectively. Background BrC values were assumed to be

199

zero, determined from the negligible BrC absorbance (10 h old. Furthermore, total

208

BrC absorption at the Vancouver and St. John’s sites is the same after correcting for plume

209

dilution. Differences between values observed in this study and those of Forrister et al.27 may be

210

because of uncertainty in determining plume ages from back trajectories, difficulty in

211

determining background CO levels in comparison to aircraft flights in and out of BB plumes, or

212

because Forrister et al. reported the sequential sum of methanol and water extracts.

213

Although total water-soluble absorption of measured BB-derived samples appears constant

214

with age, the relative contributions of higher molecular weight components (>500 Da) increase

215

in comparison to smaller molecular weight components (