Introduction
Organic acids, including C6–C32 monocarboxylic acids (MCAs),
C2–C12 dicarboxylic acids (DCAs), ketocarboxylic acids, and
1-,2-, and 3-substituted aromatic acids (AMAs), are ubiquitous in aerosols
(Kawamura, 1993; Sempére and Kawamura, 1994; Fraser et al.,
2003; Kawamura and Yasui, 2005; Wang et al., 2006; Li, 2008; Jung et
al., 2010), and they can contribute major part of the organic matter in city
atmosphere (Satsumabayashi et al., 1989; Huang et al., 2006; Duan et al.,
2009). Organic acids can be primary or secondary. The C20–C30
monocarboxylic and terephthalic acids are primary, mainly from the
combustion of fossil fuels and biomass, vehicular exhausts, and cooking
(Kawamura and Gagasion, 1987a, b; Sempére and Kawamura, 1994; Reid and
Hobbs, 1998; Graham et al., 2002; Gao et al., 2003; Reid et al., 2005; Pio
et al., 2008; Kundu et al., 2010; Tan et al., 2013). C2–C6 DCAs
are mainly secondary, formed in the atmosphere as their anthropogenic and
biogenic precursors are oxidized (Kawamura et al., 1996; Claeys et al.,
2004; Ervens et al., 2004; Gelencsér, 2004; Kanakidou et al., 2005;
Kawamura and Yasui, 2005). Although some aromatic acids formed from
oxidation reactions of volatile and/or semivolatile polycyclic aromatic
hydrocarbons (PAHs) have been studied extensively, such as phthalic acid, which is
an oxidation product of naphthalene analogues with OH, nitrate radicals, and
ozone (Wang et al., 2007), their mechanistic features and other oxidation
products (e.g., 2-hydroxybenzoic acid and phthalic anhydride) of PAHs were
still poorly characterized (Lee and Lane, 2009; Kautzman et al., 2010). Thus, it
is necessary to develop novel analytical methods that can be used to
identify possible tracers and help understand the formation mechanisms of
secondary organic acid aerosol.
Generally, organic acids are extracted from aerosol samples along with other
organics by liquid–liquid or solid-phase extraction (SPE). SPE is the most
widely used preconcentration technique which mainly depends on the sort of
the packing and properties of the solvent. Due to the difference of selectivity,
affinity, and capacity of packing, efficient enrichment of target compounds
is provided through adsorption or ion-exchange separation mechanism.
Adsorption-type SPE, such as a polyurethane foam, silica, and
silica-alumina mixture, has been used to concentrate organic analytes from
organic extraction in fine aerosol (Fraser et al., 1997; Hou al., 2006;
Duan et al., 2006).
SPE also can separate target analytes and clean up interference with good
recoveries and enrichment (Rosenfeld, 1999; Ericsson and Colmsjö, 2003;
Hou et al., 2006; Zhao et al., 2014). Adler and Siren studied the enrichment
of α, ω-dicarboxylic acids (C2–C10) by
revised-phase polymeric Oasis HLB, Strata X, strong nonpolar Isolute 101 and
strong anion-exchange SAX, and J. T. Baker NH2 and Quaternary Amine SPE
materials in the water-soluble fraction of fine aerosols (Adler and Siren,
2014). Results showed that the nonpolar sorbents possessed higher
selectivity than ion-exchange SPE for C4–C10 DCAs under
inorganic matrix. However, an aminopropyl imidazole-modified silica sorbent
exhibited high extraction efficiency towards carboxylic acids and PAHs in
river water samples based on electrostatic, π–π, and hydrophobic
interactions (Wang et al., 2014). The anion-exchange SPE cartridges have
also been used to cleanup and/or concentrate organic acids in organic
matrices. Wang et al. (2013) demonstrated a novel NH2/Carb SPE procedure to
separate organic acids from indoxacarb/acetonitrile solution based on
ion-exchange interaction and electrostatic (Wang et al., 2013). Although the
interactions between organic compounds and silica bonded ion-exchange SPE
have been investigated, application of anion-exchange SPE to separate polar
compounds from complex organic matrices was relative few. Hence, the silica
anion-exchanged SPE shows potential for the separation and enrichment of
organic acid in fine aerosol.
Gas chromatography/mass spectrometry (GC-MS) as a specific ion detection
method has been widely used to study the detailed composition of atmospheric
aerosols. For GC-MS, organic acids need to be derivatized to more volatile
esters with BF3-methanol, BF3-butanol, or
N,O-bis-(trimethylsilyl)trifluoroacetamide (Kawamura and Kaplan,
1987; Kawamura, 1993; Fu et al., 2009). The butyl and trimethylsilyl (TMS)
derivatives are often used for low-molecular-weight organic acids, including
C2–C10 DCAs, C2–C9 ketocarboxylic acids (Boucharat et
al., 1998; Limbeck and Puxbaum, 1999; Nolte et al., 2002; Wang et al., 2006;
Oliveira et al., 2007). The methyl esters of C8–C32
monocarboxylic acids and AMAs are more volatile and convenient for GC
analysis than butyl and TMS esters (Kawamura and Gagosian, 1987a, b; Plewka
et al., 2003; Fraser et al., 2003). However, quantification of trace amounts
of oxygenated PAHs (O-PAHs), including aromatic acids, identified in urban
or diesel particles is still difficult because of the low levels and
interference of the complex background compounds (Walgraeve et al., 2010).
Moreover, during purification or fractionation which involves evaporation,
volatile and/or semivolatile compounds can be lost. To deal with this
problem, an advanced pretreatment technique which is capable of enriching and
separating target analytes from others and high-resolution analytical methods
deserve to be developed.
Tandem mass spectrometry (MS-MS) is an approach which reduces the background
caused by the complex matrix by excluding all ions except the parent ion,
which can then be fragmented under specific collision energy and generate a
unique product ion mass spectrum. GC-MS-MS has been applied to detect and
quantify O-PAHs in airborne particulate (Nicol et al., 2001). For this work,
we developed a pretreatment technique to aerosol sampled in Beijing during
January 2013, when haze was particularly heavy and had spread over much of
eastern China. The air quality index in Beijing (Beijing Environment
Protection Bureau, http://www.zhb.gov.cn/) had reached nearly 500, and
visibility was sometimes below 100 m. The objective of this technique was to
separate organic acids from methyl esters with anion-SPE cartridge in the
raw extract of fine aerosol and to quantify them by GC-MS and GC-MS-MS.
Experiment and methodology
PM<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mn>2.5</mml:mn></mml:msub></mml:math></inline-formula> sample collection
Samples of PM2.5 were collected between 1 June 2012 and 30 April 2013.
The 30 samples discussed here came from 1 to 30 January, which contained the
periods of heaviest haze. The samples were collected on quartz-fiber filters
(20 cm × 25 cm Pallflex; Pall Corporation, Port Washington, NY,
USA), which were placed into a high-volume sampler (VFC-PM10; Thermo Fisher
Scientific Inc., Waltham, MA, USA) at 1.13 m3 min-1 from
21:00 local time for 24 h. The sampler was installed on the roof of a building at
Tsinghua University (40∘00′ N, 116∘32′ E; 52 m a.s.l.). Before sampling, unexposed filters and foils were
baked at 560 ∘C for 6 h to remove organic contaminants. Each
sample was wrapped in aluminum foil and stored in a freezer (at -25 ∘C) until it was extracted.
Sample pretreatment
For sample extraction, an area of 40 cm2 was punched out of the
quartz-fiber filter and extracted successively with 20 mL hexane,
dichloromethane (DCM), and acetonitrile in turn (all chromatographically
pure; Sigma-Aldrich, St. Louis, MO, USA). Acetonitrile was used as a polar
organic extraction solvent (rather than methanol) because it gives better
recoveries of fatty acids and avoids the risk of esterification between
organic acids and methanol (Polidori et al., 2008). Each extraction lasted
for 10 min in an ultrasonic instrument. The extracts were combined and
filtered through a 0.45 µm nylon syringe filter (Millex, Billerica, MA,
USA) to remove quartz fiber filter particles and insoluble suspended
particles.
For SPE, separation of organic acid from other species in solvent extract is
conducted by use of two anion-exchange SPE materials: NH2 cartridge (3 mL, containing 500 mg of a silica-based matrix with bonded aminopropyl
active groups; Supelco, Bellefonte, PA, USA) and SAX cartridge (3 mL,
containing 500 mg of a silica-based matrix with quaternary amine active
groups and Cl- counterions; Supelco) after filtration. Before sample
application, the cartridges were preconditioned with 3.0 mL hexane,
dichloromethane, and acetonitrile in turn. The 60 mL extract was percolated
through the preconditioned cartridge at a flow rate of 1–1.5 mL min-1
(controlled throughout the SPE procedure by adjusting the vacuum). After
sample application, the cartridge was washed with 8 mL hexane and then 4 mL
dichloromethane, and air was passed through the cartridge to dry it. After
washing in this way, the GC-MS chromatogram of the eluate exhibited no noise
at all (flat baseline). The retained organic acids were eluted with 2 mL of
5 % HCl–methanol into conical flasks. For enrichment, the eluate was
evaporated using a rotary evaporator at reduced pressure then under
ultra-pure nitrogen gas to about 100 µL, and then the organic acids in the
eluate were derivatized with 14 % BF3–methanol (500 µL,
Sigma-Aldrich) to obtain the corresponding methyl esters. The mixture was
placed in a water bath at 55 ∘C for 40 min. The reaction mixture
was washed with 3 mL of hexane and then 1 mL of a saturated Na2SO4 (aq) solution. The hexane layer (containing the methyl derivatives) was
transferred to a clean 2 mL vial (Millex, Billerica, MA, USA) and then
reduced in volume to 1 mL under ultra-pure nitrogen gas flow before analysis.
The effluate and wash solutions were merged and concentrated by rotary
evaporation under reduced pressure until nearly dry, then washed with 2 mL
of dichloromethane, and then reduced to 1 mL under ultra-pure nitrogen
before being analyzed.
GC-MS and GC-MS-MS analysis
The concentrated organic extracts were analyzed with an Agilent 6890N GC
system (Agilent Technologies, Santa Clara, CA, USA) equipped with an Agilent
7683 autosampler, a Quattro Micro GC triple quadrupole mass spectrometer
(Waters, Milford, MA, USA), and a SPB™-1 fused silica capillary column
(30 m long, 0.25 mm i.d., 0.25 µm film thickness; Supelco). Individual
analytes were identified by comparing the mass spectra to the standard or
searched by the National Institute of Standards and Technology (NIST08)
reference library. A 1 µL sample was injected in splitless injection
mode at 260 ∘C. The GC-MS-MS system was operated in multiple
reaction monitoring (MRM) mode using the parameters shown in Table 1. The
collision gas was argon (99.995 %), at 2.8 × 10-3 mbar.
High-purity helium was used as the GC carrier gas at a flow rate of 1 mL min-1. The column oven was programmed as follows: the initial
temperature of the oven was set at 60 ∘C, held for 1 min,
increased at 9 ∘C min-1 to 160 ∘C, held for 1 min, increased at 3 ∘C min-1 to 250 ∘C, held
for 1 min, increased at 15 ∘C min-1 to 280 ∘C,
and held for 1 min. The ion source and interface transfer line temperature
were 230 and 270 ∘C, respectively. The analytes were ionized by
electron ionization (70 eV), and the emission current was 800 µA. The
trap current was 200 mA, the repeller was set at 7.2 V, and the multiplier
was set at 650.
Method validation
To investigate the recoveries of analytes and the repeatabilities, 240 cm2 quartz fibers were punched from one real fiber filter sample and
divided into six equal pieces. The quantitative standard mixture was spiked into
five of the six pieces, and then the samples were analyzed following the
above-mentioned procedures. The methyl esters of n-alkanoic acid standard
were used to determine the recoveries and to quantify both the n-alkanoic
and branched-chain isomer acid and corresponding methyl esters. For
9-oxo-9H-fluorene-1-carboxylic acid,
9-Hydroxy-9-fluorene carboxylic acid methyl ester was used. Hexamethylbenzene/hexane (50 µL,
1.0 ng µL-1; Sigma-Aldrich) was used as an internal standard to
correct losses due to evaporation and variations in injection volume before
each GC injection. The recoveries of organic acids were determined by
comparing the whole peak areas of the eluate (after derivatized) with the
peak areas from corresponding standard mixtures, and the recoveries of fatty acid methyl
esters (FAMES)
were determined by comparing the peak areas of the combined effluate and
washing solvent (without derivatization) with the corresponding standard
mixtures. The recoveries and relative standard deviations (RSD) of analytes
are given in Table 1 and 2.
Parameters used to determine organic acids, FAMEs, and the
concentrations found in samples collected in January 2013, Beijing.
Acids
MWa
RTb
SIM
Recovery
RSD
MDLc
Air concentration
(min)
(derivatization m/z)
(%)
%
ngµL-1
(ngm-3)
(n=5)
mean
range
Dicarboxylic acids (DCAs)
Succinic
118
5.52
115/87/59/55
86.4
4.2
0.14
78.5
6.4–225.5
Glutaric
132
7.22
129/100/59/55
93.1
3.9
0.26
38.3
6.3–108.6
3-Methylpentanedioic
146
7.76
143/114/101/69
89.6
0.7
0.27
20.7
6.0–62.6
Adipic
146
8.90
143/101/111/59
89.6
0.7
0.27
35.6
7.7–71.5
Heptanedioic
160
10.44
157/128/115/74
83.7
1.4
0.16
10.6
6.4-21.5
Octanedioic
174
11.95
171/138/129/125/74/55
93.5
6.2
0.18
24.2
10.9–64.3
Nonanedioic
188
13.52
185/152/143/111/74/55
92.3
5.4
0.12
123.7
20.9–398.1
Decanedioic
202
15.37
199/166/138/125/77/55
88.4
4.1
0.10
30.6
17.0–80.3
Undecanedioic
216
17.61
213/171/152/139/98/74
80.3
2.1
0.19
17.4
10.9–41.2
Monocarboxylic acids (MCAs)
Octanoic (C8 : 0)
144
7.03
144/115/127/87/74
97.2
3.7
0.61
24.2
6.1–58.5
Nonanoic (C9 : 0)
158
8.60
143/141/129/87/74
88.4
8.4
1.96
17.6
6.0–45.8
Decanoic (C10 : 0)
172
10.14
186/155/143/87/74
96.6
3.9
0.46
12.8
6.8–22.7
Undecanoic (C11 : 0)
186
11.60
200/169/157/143/87/74
88.5
7.2
1.27
12.6
7.1–23.5
11-Dodecenoic (C12 : 1)
198
11.79
212/196/166/124/77/55
72.5
9.1
0.54
16.3
6.6–41.0
Dodecanoic (C12 : 0)
200
13.10
214/183/171/143/87/74
92.4
6.4
0.77
41.4
9.4–99.6
Tridecanoic (C13 : 0)
214
14.93
228/197/185/143/87/74
91.2
8.7
0.24
22.9
8.1–53.3
Tetradecanoic (C14 : 0)
228
17.00
242/211/199/143/87/74
101.3
4.2
0.38
182.1
15.0–937.1
13-Methyltetradecanoic
242
18.49
256/225/213/143/87/74
89.6
9.2
0.60
12.0
7.56–22.3
12-Methyltetradecanoic
242
18.69
256/225/213/143/87/74
89.6
9.2
0.60
13.5
7.7–25.2
Pentadecanoic (C15 : 0)
242
19.32
256/225/213/143/87/74
89.6
9.2
0.60
59.7
8.2–154.2
Palmitoleic (C16 : 1)
254
20.39
268/236/194/152/87/74
78.4
10.4
0.56
22.8
8.8–54.3
14-Methylpentadecanoic
256
20.99
270/239/227/143/87/74
102.4
4.2
0.25
12.7
8.9–22.1
Palmitic (C16 : 0)
256
22.02
270/239/227/143/87/74
102.4
4.2
0.25
742.6
123.7–1487.4
15-Methylhexadecanoic
270
23.66
284/253/241/143/87/74
88.4
6.5
0.16
15.5
10.3–28.0
14-Methylhexadecanoic
270
23.88
284/253/241/143/87/74
88.4
6.5
0.16
22.8
10.9–46.5
Heptadecanoic (C17 : 0)
270
24.70
284/253/241/143/87/74
88.4
6.5
0.16
86.4
17.9–236.3
Linoleic (C18 : 2)
294
26.51
294/263/95/81/67/55
77.1
7.6
0.18
118.4
12.7–328.0
Oleic (C18 : 1)
296
26.68
296/264/222/191/69/55
90.2
7.1
0.22
74.7
10.7–208.6
Stearic (C18 : 0)
284
27.43
298/255/199/143/87/74
94.9
4.9
0.14
1490.5
269.8–3268.3
Nonadecanoic (C19 : 0)
298
30.13
312/281/269/143/87/74
87.7
7.3
0.29
34.9
10.8–90.3
11-Eicosenoic (C20 : 1)
324
32.06
324/292/250/67/55
79.9
9.2
0.37
22.5
10.6–142.1
10-Oxo-octadecanoic
312
32.13
281/214/156/55
78.9
7.7
0.18
24.5
16.8–49.5
Eicosanoic (C20 : 0)
312
32.82
326/295/283/143/87/74
94.2
5.3
0.26
130.9
24.6–370.1
Henicosanoic (C21 : 0)
326
35.44
340/297/241/143/87/74
90.7
8.1
0.36
80.9
14.2–239.3
Docosanoic (C22 : 0)
340
37.99
354/311/255/143/87/74
93.5
4.6
0.17
196.2
17.2–626.3
Tricosanoic (C23 : 0)
354
40.43
368/325/269/143/87/74
90.4
5.9
0.27
114.6
16.1–380.6
Tetracosanoic (C24 : 0)
368
42.86
382/339/199/143/87/74
92.7
3.3
0.19
322.5
19.2–1080.3
Pentadecanoic (C25 : 0)
382
45.10
396/353/297/143/87/74
91.4
9.0
0.34
74.7
15.6–244.2
Hexacosanoic (C26 : 0)
396
46.51
410/367/311/143/87/74
87.6
8.1
0.34
252.4
17.5–858.5
Heptacosanoic (C27 : 0)
410
47.92
424/381/355/143/87/74
88.4
8.3
0.41
46.6
22.4–129.3
Octacosanoic (C28 : 0)
424
49.50
438/395/199/143/87/74
84.0
9.1
0.46
143.2
15.1–538.2
Nonacosanoic (C29 : 0)
438
51.47
452/423/199/143/87/74
76.5
10.2
0.51
23.0
13.5–56.2
Triacontanoic (C30 : 0)
452
53.74
466/437/199/143/87/74
72.4
10.9
0.64
45.9
13.3–161.0
Fatty acid methyl esters (FAMEs)
Methyl dodecanoate
172
13.10
214/183/171/143/87/74
96.9
4.5
0.77
10.3
7.2–16.8
Methyl myristate
242
17.00
242/211/199/143/87/74
109.4
5.4
0.38
13.6
8.1–33.8
Methyl pentadecanoate
256
19.32
256/225/213/143/87/74
96.7
3.8
0.64
12.7
8.4–18.6
Methyl palmitate
270
22.00
270/239/227/143/87/74
111.3
9.2
0.25
62.9
17.8–279.6
Methyl heptadecanoate
284
24.70
284/253/241/143/87/74
95.9
2.7
0.14
15.4
10.8–27.8
Methyl stearate
298
27.42
298/255/199/143/87/74
97.9
2.9
0.26
85.3
18.8–654.0
Methyl arachidate
326
32.82
326/295/283/143/87/74
95.2
2.2
0.32
30.7
20.1–50.7
Methyl heneicosanoate
340
35.44
340/297/241/143/87/74
96.7
3.8
0.36
12.9
12.3–13.9
Methyl behenate
354
37.99
354/311/255/143/87/74
96.5
2.6
0.17
25.2
14.6–47.6
Methyl tricosanoate
368
40.43
368/325/269/143/87/74
93.5
2.9
0.27
11.6
8.3–24.4
Methyl lignocerate
382
42.86
382/339/199/143/87/74
94.8
7.3
0.19
69.9
26.4–148.3
Methyl pentacosanoate
396
45.10
396/353/297/143/87/74
97.2
6.5
0.34
17.1
14.6–23.5
Methyl hexacosanoate
410
46.51
410/367/311/143/87/74
89.5
5.6
0.34
32.6
17.7–60.2
Methyl heptacosanoate
424
47.92
424/381/355/143/87/74
91.6
6.4
0.41
20.9
18.4–35.0
Methyl octacosanoate
438
49.50
438/395/199/143/87/74
84.9
3.8
0.46
22.3
14.3–36.8
Methyl nonacosanoate
452
51.47
452/423/199/143/87/74
79.4
3.7
0.51
13.4
12.0–18.3
Methyl triacontanoate
466
53.74
466/437/199/143/87/74
73.6
2.7
0.64
12.7
12.2–13.9
a Molecular weight.
b Retention time.
c Method detection limit.
Parameters used to determine AMAs, dimethyl phthalate, and the
concentrations found in samples collected in January 2013, Beijing.
Acids
MWa
RTb
SIM
Recovery
RSD
MDLc
Air concentration
(min)
(derivatization m/z)
(%)
%
ngµL-1
(ngm-3)
(n=5)
mean
range
Benzoic
122
6.61
136 > 105(10); 136 > 77(15)
86.4
4.3
0.24
23.3
6.5–40.8
2-Hydroxybenzoic
136
8.21
152 > 1120(10); 152 > 92(15)
81.4
5.1
0.16
15.0
6.1–36.5
2-Hydroxy-5-methylbenzoic
152
10.00
166 > 134(10); 166 > 77(15)
79.6
2.7
0.37
8.5
5.6–12.0
3,5-Dimethylbenzoic
150
10.59
164 > 133(10); 166 > 105(15)
83.2
4.4
0.46
16.6
7.3–22.4
Phthalic
166
12.12
194 > 163(15); 163 > 135(20)
91.5
5.2
0.38
54.1
13.3–187.2
Terephthalic
166
12.74
194 > 163(15); 163 > 135(20)
92.3
4.6
0.22
130.6
31.3–286.4
Isophthalic
166
12.98
194 > 163(15); 163 > 135(20)
88.4
4.3
0.40
12.8
7.6–23.5
4-Methylphthalic
180
14.00
208 > 163(10); 193 > 149(20)
80.8
5.1
0.39
15.2
6.8–48.5
4-(ethoxycarbonyl)Benzoic
180
14.07
208 > 163(15); 179 > 149(25)
67.5
3.2
0.57
7.8
7.0–10.5
2-Naphthoic
172
15.38
186 > 155(20); 155 > 127(25)
84.2
7.4
0.49
2.5
2.8–8.1
Benzene-1,2,4-tricarboxylic
210
19.81
252 > 221(20); 221 > 193(25)
80.4
7.6
0.46
3.4
2.6–10.8
Biphenyl-4-carboxylic
198
19.99
212 > 181(10); 181 > 152(20)
92.6
5.8
0.36
11.5
4.5–22.2
1,8-Naphthalic anhydride
198
24.67
198 > 154(15); 154 > 126(25)
82.5
4.2
0.58
11.0
4.8–26.9
9-Oxo-9H-fluorene-1-carboxylic
224
27.76
238 > 207(10); 180 > 151(25)
68.5
3.4
0.37
11.6
3.7–19.0
Biphenyl-4,4′-dicarboxylic
242
29.37
270 > 239(15); 270 > 152(25)
76.9
4.5
0.49
10.3
2.7–15.8
Phenanthrene-1-carboxylic
222
30.87
236 > 205(20); 205 > 177(25)
87.3
5.8
0.44
16.4
3.5–32.2
Dehydroabietic
300
32.97
314 > 299(20); 299 > 239(25)
79.4
5.7
0.68
25.4
16.7–40.2
Dimethyl phthalate
194
12.12
194 > 163(15); 163 > 135(20)
87.5
2.7
0.54
11.4
8.1–18.8
The limits of detection (LODs) were calculated based on a signal-to-noise
ratio of 3 (S/N=3) by analyzing a series of standard mixtures. The
liquid-phase LODs were transformed into the corresponding atmospheric method
detection limit (MDL) of the compounds. The MDL as atmospheric concentration
furthermore depends on the sampling rate of the filter sampler (1.13 m3 min-1), the sampling time (24 h), the extracted filter area (40 cm2),
and the solvent volume for filter extraction (60 mL), for elution (2 mL), and for injection (1 mL). These sampling and sample preparation
conditions result in a theoretical sampled air volume of 130.17 m3
being enriched, and the enrichment factor was 30 in each concentrated SPE
extract. The LODs (ng µL-1) in solvent were translated to MDL in
air concentrations by multiplying 7.68 [= (LOD × concentrated
sample solution volume)/sampled air volume]. Seven blank filters were
spiked, extracted, and analyzed to monitor the extraction procedures and
detect possible contamination. The few significant contaminants identified
via GC-MS as solvent byproducts were excluded from the data set.
Results and discussion
Efficiency of SPE
The extraction efficiency of SPE, expressed as percentage recovery, was
calculated as ratio of the concentration obtained by subtracting the
measured quantity before adding standard from after adding standard to the
theoretical adding standard. The results are shown in Table 1 for DCAs, MCAs,
and FAMEs and in Table 2 for AMAs. Extraction efficiencies for DCAs ranged
from 80.3 to 93.5 %, for MCAs ranged from 72.4 to 102.4 %, and for FAMEs
ranged from 73.6 to 111.3 %. The results were similar with the DCM/menthol
extract of organic acids (70–110 %) by ultrasonic without SPE (Huang et al., 2006; He et al.,2006) and hexane/DCM extract of organic acids
(75–96 %) by ultrasonic and silica-based or silica mixed alumina SPE
(Duan, 2006; Hou et al.,2006), which suggested the adsorption properties
were suitable for the extract of organic acids from methyl esters. Method
precisions, expressed as the RSD, were
calculated across replicate measurements (n=5). As can be seen from Tables 1 and 2, the RSD of DCAs ranged from 0.7 to 6.2 %, MCAs ranged from 3.3 to
10.9 %, FAMES ranged from 2.2 to 9.2 %, and AMAs ranged from 2.7 to 7.6 %,
the repeatability values of which we considered to be acceptable.
Adsorption-type SPE, such as silica-based materials, should be preferred for
concentration of organic acids in organic matrices but provides little
cleanup efficiency of polar functionalized acids, which are likely to be
produced in oxidation reactions in the atmosphere. One advantage capacity of
anion-exchange SPE was the cleanup efficiency of organic acids from
acetonitrile matrices. Although the fatty acids could be separated from
their corresponding methyl esters through butyl and TMS derivatization
instead of methylation, the low volatility makes the detection by GC of long-chain products
difficult. To improve the sensitivities, the extraction
efficiency of the SAX and NH2 SPE cartridges for organic acids were
investigated with 1 January 2013 provided as an example. Results for the
extraction of analytes (DCAs, MCAs, and AMAs in eluate and FAMEs in
effluate) are shown in Fig. 1. The total concentrations with the NH2
SPE cartridge were 430 ng m-3 for nine DCAs, 250 ng m-3 for 17 AMAs, and 6020 ng m-3 for 34 MCAs. They contrasted with 320 ng m-3
for eight DCAs, 160 ng m-3 for 13 AMAs, and 3950 ng m-3 for 32 MCAs with the SAX cartridge, which indicated that the NH2 SPE cartridge
gave higher extraction efficiency than the SAX cartridge.
Nevertheless, of the two silica-based anion-exchange materials, NH2 SPE
gave higher extraction efficiencies to organic acids with long hydrophobic
alkyl chain and polycyclic compounds, especially the C14–C30 MCAs,
1,8-naphthalic anhydride, 4-(ethoxycarbonyl)benzoic, 2-naphthoic,
phenanthrene-1-carboxylic, and undecanedioic acid. In addition, NH2 and
SAX extracted equal amount of 17 FAMEs (251 and 256 ng m-3,
respectively), which indicated that π–π interactions make an
important contribution to the extraction. These results indicate that the
active group bonded to the silica-based matrix in the SPE cartridge had a
marked influence on the extraction efficiency of the analytes from the
PM2.5. The active quaternary amine groups based on the SAX cartridges
might have exclusively electrostatic interactions with weak acids, which
will lead some analytes to be retained by the cartridge rather than eluted.
The aminopropyl active groups in the NH2 cartridges offer only weakly
selective retention. The results showed that the NH2 cartridges were
the most suitable for isolating organic acids but that the SAX cartridges
were most suitable for purifying nonpolar compounds.
Identification of organic acids and FAMEs
Chromatograms of the organic acids in a sample that had been pretreated with
an NH2 SPE cartridge are shown in Fig. 2. The GC-MS chromatogram
(Fig. 2a) shows that the organic acids were isolated and concentrated from the
extract and the polar and weakly polar compounds had been cleaned (meaning
that the background noise level was low). It is clear that there is a
complex suite of organic compounds in PM2.5 extracts, but the selective
separation of organic acids improves our ability to detect species that are
difficult to determine and increases the sensitivity of the method.
Even separated from the extracts, the signals of the trace multi-substituted
AMAs and PAH acids were still under the LODs of GC-MS in SIM mode. The
selectivity and sensitivity of the assay was improved by using GC-MS-MS in
MRM mode, because only the daughter ions of the selected precursors were
detected (Fig. 2b), decreasing the disturbances in the signals and improving
the signal strengths.
Concentrations of analytes found when an extract of PM2.5
collected on 1 January 2013 in Beijing was analyzed using the SAX and
NH2 SPE cartridges. The left-hand y axis (in black) is for the DCAs,
AMAs, and fatty acid methyl esters. The right-hand y axis (in red) is for
the MCAs. The T-shaped marks indicate the numbers of compounds (cpds.) that
were found.
![]()
Chromatograms of the organic compounds of the PM2.5 sample
collected on 22 January 2013 in Beijing after pretreatment using an
NH2-SPE cartridge. (a) GC-MS chromatograms of organic acids in
SPE eluate. (b) GC-MS-MS chromatograms of PAH acids in SPE eluate.
(c) GC-MS chromatogram of compounds from combined effluate and
washing solution. (d) The integrated areas of palmitic acid in the
elute (top) and methyl hexadecanoate in the combined solution (bottom).
![]()
Ambient concentrations of organic acids in January
Table 1 presents the ambient concentrations of the aliphatic acids in
PM2.5 during January 2013, along with the GC-MS validation parameters.
Table 2 presents the ambient concentrations of the aromatic acids and
GC-MS-MS validation parameters. The results suggest that the method could
also be used on samples from other seasons. Nonanedioic acid and succinic
acid were the most abundant of the DCAs that were measured, with
concentrations of 21–398 ng m-3 (mean 124 ng m-3) and 6–225 ng m-3 (mean 78 ng m-3), respectively. The aromatic DCAs were
dominated by 1,2-benzenedicarboxylic (phthalic) acid and
1,4-benzenedicarboxylic (terephthalic) acid, with concentrations of 31–286 ng m-3 (mean 131 ng m-3) and 13–187 ng m-3 (mean 54 ng m-3), respectively.
The mean concentrations of the n-alkanoic acids, n-alkenoic acids, and FAMEs
in the samples from January 2013 are shown in Fig. 3. The alkanoic acids
were strongly dominated by those with even numbers of carbon atoms, and the
pattern reached a maximum at stearic acid (C18: 0), concentration ranged
from 270 to 3268 ng m-3 (mean of 1491 ng m-3), followed by
palmitic acid (C16: 0), ranged from 124 to 1487 ng m-3 (mean of 743 ng m-3). The C18: 0 and C16: 0 acids contributed 20–65 %
(mean of 43 %) by mass of the quantified organic acids.
Seventeen FAMEs in the combined effluate and washing solvent were measured, from
which the acids had been removed by passing them through the SPE cartridges.
Of those, methyl stearate had the highest concentrations, 19–332 ng m-3
(mean of 65 ng m-3), and contributed 2–23 % (mean of 6 %) of
the total FAME. The next higher one was methyl palmitate, which ranged from 18 to 278 ng m-3 (mean of 63 ng m-3) and contributed 2–28 % (mean
of 10 %) of the total FAME. The integrated area of the palmitic acid peak
in the initial extract of the PM2.5 sample from 22 January and the area
of the methyl palmitate peak in the combined solution (effluate and washing
solvent) are shown in Fig. 2d, the methyl palmitate represented 6 % of
the palmitic acid, but the highest result can account for 13 % in this
period.
Through SPE-GC-MS-MS method, we quantified 1,8-naphthalic anhydride and five
PAH acids (2-naphthoic, biphenyl-4-carboxylic,
9-oxo-9H-fluorene-1-carboxylic, biphenyl-4,4′-dicarboxylic, and
phenanthrene-1-carboxylic acids) in the tested fine aerosol samples.
Phenanthrene-1-carboxylic acid and 9-oxo-9H-fluorene-1-carboxylic acid were
the most abundant of the PAH acids measured and had
concentrations of 10–32 ng m-3 (mean of 16 ng m-3) and 9–19 ng m-3 (mean 12 ng m-3), respectively.
Mean monocarboxylic acids (MCAs) and fatty acid methyl esters
(FAMEs) concentrations in PM2.5 samples collected in January 2013 in
Beijing.
![]()
Correlation matrix of C4–C11 DCAs and AMAs concentrations measured
in January 2013, Beijing.
C5b
C5c
C6d
C7e
C8f
C9g
C10h
C11i
AMAj
AMAk
AMAl
AMAm
AMAn
AMAo
AMAp
AMAq
AMAr
AMAs
AMAt
AMAu
AMAv
AMAw
AMAx
AMAy
AMAz
C4a
0.97
0.87
0.75
0.74
0.62
0.50
0.80
0.77
0.42
0.23
0.67
0.26
0.79
0.68
0.78
0.77
0.43
0.54
0.63
0.31
0.70
0.68
0.63
0.46
0.31
C5b
0.93
0.86
0.91
0.90
0.85
0.92
0.88
0.40
0.21
0.61
0.21
0.76
0.61
0.78
0.73
0.36
0.50
0.71
0.31
0.68
0.68
0.63
0.39
0.27
C5c
0.91
0.61
0.59
0.76
0.64
0.82
0.27
0.20
0.61
0.20
0.93
0.66
0.87
0.90
0.37
0.46
0.75
0.28
0.76
0.70
0.62
0.44
0.20
C6d
0.82
0.76
0.66
0.77
0.77
0.76
0.35
0.71
0.34
0.45
0.61
0.68
0.44
0.45
0.28
0.43
0.56
0.59
0.49
0.68
0.55
0.47
C7e
0.96
0.85
0.95
0.93
0.54
0.25
0.74
0.34
0.55
0.69
0.82
0.53
0.46
0.34
0.66
0.58
0.74
0.61
0.78
0.64
0.62
C8f
0.94
0.95
0.97
0.51
0.25
0.68
0.29
0.54
0.70
0.82
0.50
0.41
0.25
0.64
0.59
0.77
0.54
0.80
0.66
0.56
C9g
0.86
0.90
0.46
0.20
0.59
0.30
0.59
0.59
0.73
0.58
0.37
0.16
0.50
0.64
0.66
0.45
0.77
0.66
0.56
C10h
0.91
0.55
0.30
0.75
0.29
0.58
0.78
0.87
0.65
0.46
0.26
0.61
0.61
0.81
0.57
0.82
0.74
0.60
C11i
0.55
0.20
0.62
0.26
0.42
0.79
0.74
0.78
0.34
0.21
0.51
0.60
0.81
0.46
0.73
0.55
0.56
AMAj
0.72
0.67
0.43
0.21
0.50
0.49
0.26
0.43
0.18
0.22
0.73
0.43
0.28
0.57
0.69
0.55
AMAk
0.63
0.54
0.19
0.45
0.52
0.45
0.30
0.56
0.12
0.13
0.52
0.36
0.43
0.46
0.57
AMAl
0.55
0.57
0.80
0.79
0.62
0.83
0.57
0.51
0.64
0.75
0.72
0.66
0.83
0.77
AMAm
0.20
0.25
0.42
0.23
0.41
0.25
0.18
0.36
0.22
0.44
0.46
0.36
0.50
AMAn
0.84
0.90
0.97
0.34
0.45
0.72
0.30
0.93
0.70
0.66
0.44
0.18
AMAo
0.93
0.89
0.60
0.34
0.53
0.46
0.91
0.62
0.78
0.84
0.57
AMAp
0.40
0.73
0.65
0.81
0.56
0.87
0.84
0.81
0.83
0.60
AMAq
0.38
0.53
0.65
0.34
0.95
0.74
0.62
0.93
0.21
AMAr
0.48
0.29
0.39
0.52
0.75
0.61
0.71
0.60
AMAs
0.86
0.13
0.90
0.78
0.41
0.31
0.39
AMAt
0.32
0.86
0.70
0.64
0.37
0.29
AMAu
0.37
0.84
0.62
0.37
0.61
AMAv
0.66
0.69
0.86
0.48
AMAw
0.67
0.61
0.58
AMAx
0.83
0.63
AMAy
0.79
a Succinic acid; b glutaric acid; c 3-methylpentanedioic
acid;
d adipic acid; e heptanedioic acid; f octanedioic acid; g nonanedioic acid; h decanedioic acid;
i undecanedioic acid;
j benzoic acid; k 2-hydrobenzoic acid; l 2-hydroxy-5-methylbenzoic acid; m 3,5-dimethylbenzoic
acid;
n phthalic acid;
o terephthalic acid; p isophthalic acid; q 4-methylphthalic acid; r 4-(ethoxycarbonyl)benzoic
acid;
s 2-naphthoic acid; t benzene-1,2,4-tricarboxylic
acid;
u biphenyl-4-carboxylic acid; v 1,8-naphthalic
anhydride;
w 9-oxo-9H-fluorene-1-carboxylic acid; x biphenyl-4,4′-dicarboxylic
acid; y phenanthrene-1-carboxylic acid; z dehydroabietic
acid.
Correlations between FAME and n-alkanoic acid
The correlation coefficients between the C16: 0, C14:0, and
C18: 0 acids, and the other C13–C30 n-alkanoic acids were
significant (R2 > 0.88, p < 0.01), which implied
common primary sources from biomass burning, vehicular exhausts, cooking, and
vegetation detritus (Simoneit, 1986; Rogge et al., 1991, 1993a, b; Schauer et
al., 1999, 2001; Fraser et al., 2003; Hyder et al., 2012; Pietrogrande et
al., 2014). The C18: 0 / C16: 0 ratio has been used as a qualitative
tool for assessing organic acid sources (Ho et al., 2010), and we found
C18: 0 / C16: 0 > 1 during our sampling period. Taking into
account a recent study of organic acids (Zhao et al., 2014), we suggest the
strong contribution from cooking and vehicular emissions in Beijing in
January 2013. Concentrations of high molecular-weight n-alkanoic acids (≥ C20) are high in this period, which suggested the release of plant
waxes from dead leaves and wind abrasion of other sources (Simoneit et al.,
1981, 2004; Rogge et al., 1993b; Guo et al., 2003; Kawamura
and Yasui, 2005; Feng et al., 2006).
Like n-alkanoic acids, the methyl esters of the C20–C30 waxy acids
also have even carbon preference. The pattern had a maximum at methyl
lignocerate (C24), with concentrations ranging from 26 to 148 ng m-3 (mean of 70 ng m-3), followed by methyl hexacosanoate
(C26). The methyl esters of the C13 and C19 acids were lower
than the GC-MS (in SIM mode) detection limits, and they also could not be
measured by GC-MS-MS-MRM because of their low molecular-ion signal
strengths. The correlations between C14, C16, and C18 methyl
esters and among the other C13–C30 methyl esters were high
(R2 > 0.85), which was quite similar to the fatty acids.
From the above information and the molecular structures, we suggest that
FAMEs could have similar sources to carboxylic acids, be formed by
esterifying carboxylic acids, or that a fraction of the fatty acids were
formed by hydrolyzing FAMEs. Also, the uncertainty in the measurements of
phthalic acid and dimethyl phthalate have been avoided by the use of the
NH2 SPE technique, like fatty acids separated from their corresponding
methyl esters.
Correlations between DCA and AMA concentrations
The correlation coefficients between the C4–C11 DCAs and AMAs are
shown in Table 3. The DCAs were strongly correlated (r2 > 0.85, p < 0.01). The di-C4–C5 DCAs (succinic, glutaric, and
3-methylpentanedioic acids) were strongly correlated with phthalic,
4-methylphthalic, benzene-1,2,4-tricarboxylic acids, and 1,8-naphthalic
anhydride, and the newly identified 9-oxo-9H-fluorene-1-carboxylic and
2-naphthoic acids. Previous studies have suggested that C4–C5 DCAs
are mainly secondary products of volatile organic compound precursors
(Kawamura and Kaplan, 1987; Kawamura and Kasukabe, 1996; Graham et al.,
2002; Claeys, et al., 2004; Ervens et al., 2004; Gelencsér, 2004;
Kanakidou et al., 2005; Kawamura and Yasui, 2005; Oliveira et al., 2007).
Phthalic acid has been proposed as a surrogate for the contribution of
secondary organic aerosols (Fine et al., 2002; Ho et al., 2010), since it
mainly derives from the oxidation of PAHs (e.g., naphthalene) or phthalates
in vehicular exhaust and biomass burning (Guillard et al., 1993; Kawamura
and Ikushima, 1993; Kawamura and Kaplan, 1987; Kawamura and Yasui, 2005). In
light of the above-mentioned facts, we suggest that
9-oxo-9H-fluorene-1-carboxylic acid, 2-naphthoic acid, and 1,8-naphthalic
anhydride are mainly secondary.
Phenanthrene-1-carboxylic acid correlated strongly with the
di-C7–C11 acids terephthalic, isophthalic,
2-hydroxy-5-methylbenzoic, and 4-(ethoxycarbonyl) benzoic. Among these
acids, terephthalic acid and isophthalic acid are used as tracers of primary
emissions from motor vehicles and domestic heating with fossil fuels
(Kawamura and Kasukabe, 1996; Fine et al., 2002; Fraser et al., 2003; Wang
et al., 2006), which indicated phenanthrene-1-carboxylic acid was primary
during January.
The correlation between biphenyl-4-carboxylic acid and benzoic acid (0.73)
is higher than for other acids. Benzoic acid may be primary from vehicular
exhausts (Kawamura and Kaplan, 1987; Rogge et al., 1993a) and secondary from
photochemical degradation of aromatic hydrocarbons, such as toluene, emitted
by automobiles (Ho et al., 2006; Sun et al., 2006; Li et al., 2009). The
aromatic biphenyl is a widely distributed pollutant (Selesi and Meckenstock,
2009), and it is found in coal tar at concentrations of 0.2 to 0.4 %.
The oxidation of toluene in the atmosphere is more complicated than the
breaking of the single bond that is found in biphenyl, so we suggest that
biphenyl-4-carboxylic acid is mainly a secondary product of the
photochemical degradation of biphenyl. The biphenyl-4,4′-dicarboxylic acid
correlated relatively strongly with di-C8 acids (0.80), di-C10
acids (0.82), and terephthalic acid (0.81), which are all mainly primary, as
noted above. In view of this, we suggest that biphenyl-4,4′-dicarboxylic
acid is mainly primary.