Subscriber access provided by Columbia University Libraries
Article
Emission Factors for Selected Semivolatile Organic Chemicals from Burning of Tropical Biomass Fuels and Estimation of Annual Australian Emissions Xianyu Wang, Carl P. Meyer, Fabienne Reisen, Melita Keywood, Phong K. Thai, Darryl W. Hawker, Jennifer Claire Powell, and Jochen F. Mueller Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b01392 • Publication Date (Web): 07 Aug 2017 Downloaded from http://pubs.acs.org on August 11, 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 26
Environmental Science & Technology
Emission Factors for Selected Semivolatile Organic Chemicals from Burning of Tropical Biomass Fuels and Estimation of Annual Australian Emissions Xianyu Wang,a,* Carl P. Meyer,b Fabienne Reisen,b Melita Keywood,b Phong K. Thai,a,c Darryl W. Hawker,d Jennifer Powell,b and Jochen F. Muellera
a
Queensland Alliance for Environmental Health Sciences, The University of Queensland, 39
Kessels Road, Coopers Plains, Queensland 4108, Australia b
CSIRO Oceans and Atmosphere Flagship, Aspendale Laboratories, 107-121 Station Street,
Aspendale, Victoria 3195, Australia c
International Laboratory for Air Quality and Health, Queensland University of Technology,
2 George St, Brisbane City, Queensland 4000, Australia d
Griffith School of Environment, Griffith University, 170 Kessels Road, Nathan, Queensland
4111, Australia
ACS Paragon Plus Environment
Environmental Science & Technology
1
ABSTRACT
2
This study reveals that open-field biomass burning can be an important source of various
3
semivolatile organic chemicals (SVOCs) to the atmosphere including polycyclic aromatic
4
hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), polybrominated diphenyl ethers
5
(PBDEs) and a range of pesticides. Emission factors (EFs) for 39 individual SVOCs are
6
determined from burning of various fuel types that are common in tropical Australia.
7
Emissions of PAHs are found to be sensitive to differences in combustion efficiencies rather
8
than fuel types reflecting a formation mechanism. In contrast, revolatilisation may be
9
important for other SVOCs such as PCBs. Based on the EFs determined in this work,
10
estimates of the annual emissions of these SVOCs from Australian bushfires/wildfires are
11
achieved, including for example ∑ PAHs (160 (min) – 1,100 (max) Mg), ∑ PCBs (14 – 300
12
kg), ∑ PBDEs (8.8 – 590 kg), α-endosulfan (6.5 – 200 kg) and chlorpyrifos (up to 1,400 kg),
13
as well as dioxin toxic equivalents (TEQ) of ∑ dl-PCBs (0.018 – 1.4 g). Emissions of SVOCs
14
that are predominantly revolatilised appear to be related to their use history, with higher
15
emissions estimated for chemicals that had a greater historical usage and were banned only
16
recently or are still in use.
ACS Paragon Plus Environment
Page 2 of 26
Page 3 of 26
17
Environmental Science & Technology
TABLE OF CONTENTS GRAPHIC
18
19
ACS Paragon Plus Environment
Environmental Science & Technology
20
INTRODUCTION
21
Open-field biomass burning including agricultural waste burning, peat fires and
22
forest/savannah fires is an important source of emissions for a wide range of organic
23
pollutants including semivolatile organic chemicals (SVOCs).1-4 Many SVOCs, including
24
various important air pollutants such as polycyclic aromatic hydrocarbons (PAHs) and
25
halogenated organic compounds, are hazardous to humans.5 Release of these compounds
26
from biomass combustion events involves processes of de novo formation (i.e. compounds
27
newly formed from precursors and dependent on combustion conditions) or revolatilisation
28
(i.e. thermally stable chemicals remobilised untransformed due to increased temperatures).
29
Amongst types of open-field biomass burning, forest/savannah fires are dominant on a global
30
basis, accounting for 95% of total carbon emissions from this source.6 Globally, tropical
31
regions comprise most of the open-field biomass burning area, with the largest contributions
32
being from (central and southern) Africa and (central and northern) Australia.7, 8 Satellite-
33
derived data suggest that, from 1996 – 2012, the annual mean area burned across Australia
34
was the highest of any individual country, accounting for 15% of the global burned area.7, 9
35
Most of these fire-affected areas are in Australia’s northern tropical savannah woodlands and
36
central and northern arid rangelands.10, 11 As such, the contribution of these fires in tropical
37
and arid Australia to the emission of harmful/toxic SVOCs is potentially significant. An
38
estimate of these emissions is essential to understand the contribution from open-field
39
biomass burning to the environmental burden of these chemicals.
40
In order to achieve the above estimate for relevant SVOCs, it is first necessary to
41
measure/determine their emission factors (EFs), which are defined as mass of the compound
42
released to the atmosphere per unit mass of fuel consumed by combustion. This parameter is
43
important for quantitatively estimating the emissions of given chemicals from a regional or
44
global perspective, in combination with knowledge of the mass of relevant vegetation
45
combusted.6 It is also a key requirement in the construction of models of atmospheric
46
transport and chemistry, necessary for evaluating the atmospheric impact of biomass
47
burning.3 The typical approach to measure/determine EFs is through sampling the fire smoke
48
emissions from burning of a known amount of biomass. Measurement of SVOC EFs from
49
tropical biomass has been carried out to some extent for polychlorinated dibenzo-p-dioxins
50
and dibenzofurans (PCDD/Fs) and dioxin-like (dl) polychlorinated biphenyls (PCBs).12-15 It
51
has also been recognised that during biomass burning many other SVOCs such as PAHs and
52
pesticides can be released.16-19 However, relevant EF data for PAHs are mostly limited to
ACS Paragon Plus Environment
Page 4 of 26
Page 5 of 26
Environmental Science & Technology
53
extratropical fuels while data for tropical biomass fuels are scarce.1, 20 There are essentially
54
no relevant data for other SVOCs such as pesticides and polybrominated diphenyl ethers
55
(PBDEs).
56
The aim of this study was to determine the EFs for a wide range of SVOCs from burning
57
various fuel types that are common in tropical Australia. With these data, the present study
58
also provides a first estimate of the annual emissions of many SVOCs from tropical
59
Australian bushfires/wildfires.
60 61
METHODOLOGY
62
Sample collection. The study was conducted at the Mornington Sanctuary, a 3,500 km2
63
nature reserve in the Kimberly region of Western Australia (17°31′44″ S, 126°6′12″ E).21 The
64
region is typical of Australia’s open savannah woodlands and receives between 600 and
65
1,000 mm annual rainfall.21 There is a lack of current local anthropogenic sources for SVOCs
66
in the area contributing to the biomass-loading concentrations of these chemicals. Therefore
67
we expect the sampling site (and emissions of SVOCs of interest from combustion of fuels
68
naturally growing in the vicinity) to be representative of Australia’s most fire-prone areas, i.e.
69
relatively unpopulated northern and central Australia. The vegetation comprises sparsely
70
distributed trees less than 10 m in height (various Eucalyptus spp. and Corymbia spp.) with
71
an understorey of hummock grasses (spinifex, Triodia spp.) and annual and perennial tussock
72
grasses. Hummock grasses dominate the less fertile areas while tussock grasses tend to occur
73
mainly on the richer volcanic and alluvial soils. These biomass types are among the most
74
dominant in Australia in terms of areas, occupying at least 18%, 12% and 7% respectively of
75
the whole continental land area.22 Further, the fact that they are distributed mostly in northern
76
and central Australia means even higher proportions in these fire-affected areas. Test burns
77
were conducted at a location close to the sources of the fuels whose emissions we aimed to
78
investigate. The biomass fuels used in this study comprised eucalypt leaf litter, eucalypt
79
coarse woody debris, spinifex and tussock grasses.
80
Measurements were conducted in August 2013, using a high volume smoke sampler with a
81
sampling rate of approximately 1 m3 min-1. Details of the sampler design have been published
82
elsewhere14, 15 and a schematic diagram is provided as Figure S1 in the Supporting
83
Information (SI). To minimise dilution from background/ambient air, smoke samples were
84
collected directly above the fire plume (see Figure S2 in the SI as an example). Total
ACS Paragon Plus Environment
Environmental Science & Technology
85
suspended particles (TSP) and particle-associated chemicals were collected on a quartz fibre
86
filter (QFF, 203 × 254 mm) and gaseous chemicals separately collected on two subsequent
87
130 mm diameter polyurethane foam (PUF) plugs (51 and 25 mm thicknesses for the front
88
and back PUFs, respectively). A small bypass airflow was drawn into the associated carbon
89
monoxide (CO) and carbon dioxide (CO2) analyser (Gascard II, Edinburgh Instruments,
90
Edinburgh, UK) to determine their concentrations (Figure S1).
91
The fuels of interest were collected from the surrounding undisturbed vegetation class
92
immediately prior to each test and burned on an open hearth (within an area of 2 m2 and a
93
height of 0.2 m) in beds constructed to approximate their undisturbed state and density.
94
Smoke samples were taken from above the fire using the high volume sampler. The height of
95
the sampling hood was adjusted throughout each test to ensure that surface temperatures of
96
the hood and the line were less than 200 °C to minimise the risk of formation artefacts on the
97
sampler surface, and that smoke levels (assessed by the CO2 and CO concentrations)
98
remained within measurement range. In total, 11 smoke samples were collected with the
99
sampling duration ranging from 18 to 80 min for each sample (Table S1). These experiments
100
included fires with short (0.3 – 1.2 m) or long (1.5 – 2 m) flames and smoldering and full-
101
course events in order to represent the range of complex burning conditions of actual
102
bushfires. For the flaming samples, fuel was fed into the hearth progressively at the rate
103
required to maintain the desired intensity of the flaming phase. Addition of the fuel was
104
carefully conducted from the side of the burning pile. The smoldering phase was defined as
105
the stage when little or no visible flame could be observed. For the full-course fires, sampling
106
was from ignition until the fuels were burnt out and no additional fuel was loaded during the
107
combustion. Collected QFF and PUF samples were stored at -25 °C until analysis.
108
Chemical analysis. Details of chemical analysis are provided in Section 2 in the SI. Briefly,
109
the mass of TSP within each sample was determined using a gravimetric method. The
110
collected QFFs and PUFs were spiked with a solution containing 7 deuterated PAHs, 18 13C-
111
PCB congeners, 7 13C-PBDE congeners and 14 13C-labelled pesticides at different levels as
112
internal standards for quantification purposes (Table S2). QFF and PUF samples (both plugs
113
combined) were then separately extracted in a Dionex ASE 350 Accelerated Solvent
114
Extractor (Thermo Fisher Scientific) using n-hexane and acetone (1:1, v/v). Each extract was
115
split 40%/40%/20% (v/v/v). The first aliquot (40%, F1) was cleaned up and analysed for non-
116
acid resistant compounds (i.e., the analytes that would not survive the cleanup procedures
117
involving concentrated sulfuric acid treatment) including 13 PAH compounds and 13
ACS Paragon Plus Environment
Page 6 of 26
Page 7 of 26
Environmental Science & Technology
118
pesticides. The second (40%, F2) was for acid resistant compounds targeting 18 PCB
119
congeners, 14 polychlorinated naphthalene (PCN) congeners, 14 other pesticides and 7 PBDE
120
congeners. The third (20%, F3) was analysed for the biomass burning tracer levoglucosan.
121
(See details of sample cleanup in Section 2 of the SI. The full chemical list is provided in
122
Table S2 of the SI.)
123
Aliquots F1, F2 and F3 were analysed separately for the respective target compounds using a
124
Thermo 1310 gas chromatograph coupled to a DFS Magnetic Sector high-resolution mass
125
spectrometer (GC-HRMS). The HRMS was operated in electron impact-multiple ion
126
detection (EI-MID) mode and resolution was set to ≥ 10,000 (10% valley definition).
127
Quality assurance and quality control (QA/QC). Details on QA/QC are provided as
128
Section 3 in the SI. Briefly, breakthrough effects were monitored for each sample. Solvent,
129
matrix and field blank samples were integrated within sample batches and accounted for
130
about 30% of the total sample numbers. Method detection limits (MDLs) were defined as the
131
average field blank plus three times the standard deviation. If the relevant compounds could
132
not be detected within the field blank samples, MDLs were determined based on half the
133
instrument detection limits. MDLs were typically < 1 ng m-3 for PAH analytes and < 10 pg
134
m-3 for other SVOCs as detailed in Table S3 for individual chemicals.
135
Derivation of emission factors. EFs are derived using the carbon-balance model, which
136
assumes that the total carbon in the fuels is a conserved quantity. Its principles are based on
137
the work of Andreae and Merlet3 and Meyer et al.15
138
The model or approach can be expressed by the following equation:
139
= ∆
∆
× =
×
(1)
140
where is the emission factor (mass analyte kg-1 fuel) for a specific compound or
141
compound group , represents the fuel carbon content and and are the
142
atmospheric concentrations (mass m-3) of the chemical or carbon under combustion
143
conditions and ambient (background) conditions respectively.
144
Typically, the carbon content of dry biomass fuel is close to 50% and varies only within a
145
limited range between different fuel types.3, 15 During the combustion process, more than
146
85% of the carbon is emitted as CO2.15 Therefore for simplicity we approximated the mass of
147
emitted carbon to be the mass of C in emitted CO2 (CO2-C). This will lead to a slight
148
overestimation of the EF but is well within the typical uncertainty of SVOC analysis (relative
ACS Paragon Plus Environment
Environmental Science & Technology
Page 8 of 26
149
standard deviation (RSD) of 20 – 50% for replicate QC samples fortified with analyte of
150
interest).23-26 The above equation is thus simplified to:
151
=
152
where '# and '# are concentrations of CO2-C (mass m-3) in the smoke and
153
ambient air respectively.
154
The Mornington Sanctuary sampling site is considered remote as mentioned, which means a
155
potentially low level of SVOCs in the ambient air. Taking some other remote sites in northern
156
Australia as examples, PCB concentrations in air are typically reported as some hundred
157
femtograms per cubic metre.27 Obtaining reliable results for ambient levels of these SVOCs
158
at the Mornington Sanctuary sampling site would then be expected to require a sampling
159
duration of approximately 12 to 24 hours. Several logistical challenges were recognised in the
160
operation of the sampler for such an extended time at this site such as a lack of power supply,
161
maintenance of the sampling equipment and ongoing sampling parameter monitoring and
162
adjustment. We therefore decided not to collect background samples directly within this
163
sampling campaign. Instead, background atmospheric concentration data for SVOCs and CO2
164
refer to those from a study based on another remote site in the Northern Territory, Australia,
165
namely the Australian Tropical Atmospheric Research Station (ATARS, 12°14'56.6"S,
166
131°02'40.8"E) which provides better access to power supply and shelter for both personnel
167
and equipment. These data were obtained in the year of 2014, using high volume air samplers
168
(for SVOCs) and a high precision Fourier Transform Infrared trace gas and isotope analyser
169
(for CO and CO2, Spectronus, Ecotech Pty. Ltd., Knoxfield, Australia). Samples identified as
170
not being impacted by fire events were used. These were analysed in the same laboratory and
171
using the same methods as those for the smoke plumes in the current work. (See details in
172
Wang et al.1).
173
RESULTS AND DISCUSSION
174
Detection and concentrations of SVOCs in smoke samples. Overall, 47 out of the 79
175
targeted chemicals were detected in over half of the samples, including all the PAH analytes,
176
most PCB and PBDE congeners, some of the pesticides such as α-endosulfan, chlorpyrifos
177
and hexachlorobenzene (HCB), and some PCN congeners. Concentrations of most of these
178
SVOC analytes (39 out of 47), as well as TSP and the cellulose combustion product
179
levoglucosan in the smoke samples were considerably higher (i.e. having a mean value at
180
least 10 times higher) than background levels (Table S4 in the SI) with the factor ranging
!"# !"#
× 0.5
ACS Paragon Plus Environment
(2)
Page 9 of 26
Environmental Science & Technology
181
from 16 to 11,000 for different analytes. This suggested that combustion of these fuels
182
represents an important source of these SVOCs to their ambient atmospheric environment.
ACS Paragon Plus Environment
Environmental Science & Technology
183
Table 1. Emission factors of TSP (g kg-1 fuel burnt), gaseous + particle-associated levoglucosan (g kg-1 fuel burnt) and selected target SVOCs
184
(µg kg-1 fuel burnt) from burning of different fuels. For dioxin-like PCBs, the emission factor is expressed on the basis of dioxin toxic
185
equivalents (TEQ) of ∑ dl-PCBs (pg kg-1 fuel burnt). Also shown is the modified combustion efficiency (MCE)
186 187 188 189 190
Page 10 of 26
Spinifex Tussock grasses Eucalypt leaf litter Eucalypt coarse woody debris Short Long Long flaming Short Long flaming Flaming + Full-course Flaming Smoldering Flaming Smoldering flaming flaming + smoldering flaming + smoldering smoldering TSP 8.6 24 11 2.7 7.4 7.2 4.2 24 12 3.2 31 Levoglucosan 0.082 0.21 0.090 0.0086 0.047 0.029 0.034 0.17 0.045 0.012 0.24 ∑ PAHs(a) 3,800 3,500 3,700 560 640 2,700 880 2,500 780 680 2,600 ∑ PCBs(b) 0.33 1.1 0.39 0.085 0.14 0.12 0.10 0.21 0.050 0.059 0.16 ∑ PCNs(c) 0.011 0.0088 0.0059 0.0047 0.0022 0.011 0.0012 0.0070 0.0011 0.00066 0.0025 ∑ PBDEs(d) 0.58 2.1 0.42 0.092 0.15 0.094 0.057 0.19 0.031 0.081 0.14 HCB 0.045 0.089 0.029 0.011 0.015 0.023 0.013 0.049 0.024 0.022 0.042 γ-HCH 0.040 0.10 0.093 0.014 0.013 0.012 0.024 0.015 0.0015 0.0084 0.016 p,p’-DDE 0.0067 0.021 0.0056 0.0016 0.0019 0.0038 0.00115 0.0021 0.00050 0.00068 0.00059 α-endosulfan 0.67 0.46 0.73 0.10 0.091 0.19 0.067 0.52 0.12 0.064 0.023 Chlorpyrifos NA 3.9 0.21 1.1 NA 0.34 NA 5.1 1.8 NA NA ∑ dl-PCBs TEQ(e) 1.2 5.0 1.6 0.18 0.33 0.16 0.15 0.28 0.068 0.064 0.22 MCE(f) 0.954 0.935 0.927 0.980 0.972 0.923 0.973 0.936 0.961 0.986 0.917 (a) Refers to sum of phenanthrene (Phe), anthracene (Ant), fluoranthene (Flu), pyrene (Pyr), benzo[a]anthrancene (BaA), chrysene (Chr), benzo[b]fluoranthene (BbF), benzo[k]fluoranthene (BkF), benzo[e]pyrene (BeP), benzo[a]pyrene (BaP), indeno[1,2,3-cd]pyrene (I123cdP), dibenzo[a,h]anthracene (DahA) and benzo[g,h,i]perylene (BghiP) data; (b) Refers to sum of data for congeners 28, 52, 101, 138, 153, 180, 77, 105, 114, 118, 156, 157 and 167; (c) Refers to sum of data for congeners 13, 27 and 28+36; (d) Refers to sum of data for congeners 28, 47, 99, 100 and 154; (e) Refers to sum of TEQ data for congeners 77, 105, 114, 118, 156, 157 and 167; (f) Expressed to three significant figures
ACS Paragon Plus Environment
Page 11 of 26
Environmental Science & Technology
191
Group- and compound-specific emission factors for SVOCs. EF values were calculated
192
for each chemical group and the aforementioned 39 individual chemicals that were detected
193
in over half of the samples and had concentrations considerably higher than background
194
levels (Tables 1 and S5; values shown are gaseous (from PUFs) + particle-associated (from
195
QFFs)). Also derived were the EF values for TSP and levoglucosan. The EF for TSP from
196
this study ranged from 2.7 to 31 g kg-1, agreeing very well with the literature values from
197
savannah, grassland and tropical forest (6.5 – 11 g kg-1).3 Levoglucosan EF (0.0086 – 0.24 g
198
kg-1) was also within the range from the few relevant reports for tropical biomass
199
combustions (up to 0.50 g kg-1).3, 28 As a group, PAHs had the highest emission factors,
200
ranging from 560 to 3,800 µg kg-1 fuel burnt for ∑ PAHs depending on the fuel and
201
combustion conditions. The individual compound with the highest EF was phenanthrene
202
(Phe, 200 – 1,300 µg kg-1 (Table S5)). Other SVOCs/SVOC groups with relatively high EFs
203
(µg kg-1 fuel burnt) were ∑ PCBs (0.050 – 1.1), ∑ PBDEs (0.031 – 2.1), and amongst
204
pesticides, α-endosulfan (0.023 – 0.73) and chlorpyrifos (up to 5.1). Overall, the variation in
205
EF data from all samples was less with PAHs than other SVOC groups. For example, the
206
ratio of the highest to the lowest EF for ∑ PAHs is approximately 7, compared to the one of
207
22 for ∑ PCBs.
208
Variability in SVOC emissions can be discussed within two contexts: variability arising from
209
combustion chemistry (i.e. inherent fuel chemical composition and the characteristics of the
210
combustion event) and variability that arises from the revolatilisation of relatively stable
211
SVOCs that have previously been deposited/uptaken on/by the fuel from other sources. With
212
the former context, we expect to see variability between combustion types and fuel types
213
rather than within these types; with the latter we would expect to see variability within both
214
combustion and fuel types.
215
Emissions of PAHs from biomass burning are mainly through de novo formation processes
216
(for example from aliphatic precursors such as propargyl moieties forming intermediate
217
cyclopentadienyl radicals).16 Given there should be limited variation in the carbon content of
218
the fuels of interest,3, 15 we should expect to see variation in PAH emissions associated with
219
combustion types rather than with fuel types. The emission profile of PAHs produced during
220
combustion is related to the relative completeness of the oxidation process, commonly
221
expressed as the modified combustion efficiency (MCE, shown for each sample in Table 1):29 ( =
)*+ , -*. + -*+ .
ACS Paragon Plus Environment
Environmental Science & Technology
222
where -*. and -*+ . are the number of moles of each measured during the collection of a
223
sample. Increased levels of the former are associated with reduced combustion efficiency.
224
Both EF data for PAHs and levoglucosan are negatively correlated with MCE (Figure 1, p