Molecular distributions and compound-specific stable carbon isotopic compositions of lipids in wintertime aerosols from Beijing

Molecular distributions and stable carbon isotopic compositions (δ13C) of n-alkanes, fatty acids and n-alcohols were investigated in urban aerosols from Beijing, northern China to better understand the sources and long-range atmospheric transport of terrestrial organic matter during polluted and clear days in winter. n-Alkanes (C19–C36), fatty acids (C8–C32) and n-alcohols (C16–C32) detected in Beijing aerosols are characterized by the predominance of C23, C16 and C28, respectively. Carbon preference index (CPI) values of n-alkanes, the ratios of the sum of odd-numbered n-alkanes to the sum of even-numbered n-alkanes, are close to 1, indicating a heavy influence of fossil fuel combustion. Relatively higher ratios of C(18:0+16:0)/C(18:n+16:1) (fatty acids) on clear days than polluted days indicate that long-distance transport and/or photochemical aging are more significant during clear days. δ13C values of n-alkanes and low molecular weight fatty acids (C16:0, C18:0) ranged from –34.1 to −24.7% and −26.9 to −24.6%, respectively, which are generally heavier on polluted days than those on clear days. Such a wide range suggests that atmospheric lipids in Beijing aerosols originate from multiple sources and encounter complicated atmospheric processes during long-range transport in North China.


Results and Discussion
Molecular distributions of lipid compounds. A homologous series of n-alkanes (C 19 -C 36 ), fatty acids (C 8 -C 32 ) and n-alcohols (C 16 -C 32 ) were detected in Beijing aerosols. The temporal variations are shown in Fig. 1 and the summary of the concentrations is presented in Table 1. Concentrations of individual compounds are provided in Table S1 of the supporting information (SI).
n-Alkanes. Concentrations of n-alkanes (C 19 -C 36 ) range from 40.1 to 1720 ng m −3 (average 367 ng m −3 ) ( Table 1). These concentrations are comparable to those (573 ng m −3 ) reported in wintertime samples in fourteen Chinese megacities 34 and are higher than those (177 ng m −3 ) in summertime Mt. Tai aerosols 2 . Furthermore, the concentrations of n-alkanes in Beijing aerosols are 2 orders of magnitude higher than those (0.11-14.1 ng m −3 ) in marine aerosols 27,35 . The temporal variations of n-alkanes have been shown to exhibit a noticeable increase in nighttime (Fig. 1a). Concentrations of n-alkanes in nighttime (55.3-1720 ng m −3 , 472 ng m −3 ) were nearly two times higher than those (40.1-1440 ng m −3 , 262 ng m −3 ) in daytime (Table 1; Fig. 2a), although the diurnal variations are not significant (Table S2). Homologous n-alkanes are characterized by no odd/even predominance in the range of low molecular weight (LMW, < C 26 ) n-alkanes and weak odd carbon number predominance of HMW n-alkanes ( Fig. S1a) with the concentrations of LMW n-alkanes (242 ng m −3 ) being two times higher than HMW n-alkanes (111 ng m −3 ; Fig. 1a). In addition, the C max values are obtained at C 23 and C 25 (Fig. 2a; Table S1), while C 23 is the abundant homolog in most of the samples. CPI are used to identify pollution sources in aerosol [36][37][38] . CPI values obtained for C 19 to C 36 are 1.6 ± 0.35 in daytime and 1.4 ± 0.1 in nighttime (Table 1; Fig. 1a). They are in good agreement with those reported in Beijing and other urban aerosols from China (1.0 ± 0.43) 34 but lower than those reported in mountain aerosols (up to 8.0) 2 and marine aerosols (1.2-13) 13,27 . These results suggest that n-alkanes in Beijing were mainly derived from incomplete combustion of fossil fuels and petroleum residue.
Fatty acids. Fatty acids (C 8 -C 32 ) including three unsaturated fatty acids (palmitoleic acid (C 16:1 ), oleic acid (C 18:1 ), and linoleic acid (C 18:2 )) are identified and quantified in the Beijing aerosol samples (Table S1). The molecular distribution of fatty acids is characterized by a strong even carbon number predominance with C max at n-hexadecanoic acids (C 16:0 ), followed by n-octadecanoic acids (C 18:0 ) (Figs 2b and S1b). A similar pattern has been reported in the other urban aerosols in China 32 , India 17 and USA 39 , but is different from the bimodal distribution observed for mountain aerosols 2 or for marine aerosols, with maxima at C 16:0 and C 24:0 /C 28:0 35,40 .
The concentrations of fatty acids range from 137 to 3310 ng m −3 (871 ng m −3 ). Their averages were higher in nighttime (1010 ng m -3 ) than in daytime (733 ng m -3 ; Table 1; Fig. 1b) but not statistically significant (Table S2). There is no large difference between the concentrations of fatty acids in Beijing and other megacities in China (318-3240 ng m −3 , 1070 ng m −3 in winter) 34 . However, these values are higher than those (302 ng m −3 in winter and 504 ng m −3 in summer) in Chennai aerosols 17 , and are 1-2 orders of magnitude higher than those (2.46-60.2 ng m −3 , 13.8 ng m −3 ) in marine aerosols from the western north Pacific 35 .
Plant tissues consist of abundant fatty acids that can be emitted from biomass burning. While fossil fuel combustion, such as diesel truck exhaust and meat cooking, are also an important sources of fatty acids (C 16:0 , C 18:0 ) 41 .
Studies have demonstrated that the ratio of C 18:0 /C 16:0 fatty acid can be used as a qualitative tool for source assessment 42 . The C 18:0 /C 16:0 values are lower than 0.25 in aerosols from foliar vegetation combustion, waxy leaf surface abrasions, and wood smoke; values in the interval 0.25-0.5 are registered for car and diesel truck exhausts; values between 0.5 and 1.0 are achieved for cooking and paved and unpaved road dust 42 . The C 18:0 /C 16:0 ratios in our study showed average values of 0.84 in the daytime versus 0.83 in the nighttime (Table 1), implying a strong input from cooking, road dust, and/or vehicle emissions for both C16:0 and C18:0 in Beijing.
Both biogenic and anthropogenic emissions are main sources of fatty acids in the atmosphere. High molecular weight (HMW, ≥ C 20 ) fatty acids are mostly derived from higher plants, while low molecular weight (LMW, < C 20:0 ) fatty acids have multiple sources in urban aerosols, such as anthropogenic sources (such as kitchen emissions) or biomass combustion 8,32,[35][36][37][38][39][40][41][42][43] . The average ratio of LMW/HMW is 12 in daytime versus 7.7 in nighttime (Table 1), which is much higher than mountain aerosols in China (1.0 ± 0.8) 2 but similar to those (5.3 ± 1.8) reported in urban aerosols from Chennai, where the contribution of biomass burning is significant 17 . Furthermore, CPI values (2.4-7.6) of fatty acids imply that biogenic emission is also an important emitter of fatty acids in this region. Thus, these results demonstrate that fatty acids in Beijing aerosols can be explained by the mixed contributions of biogenic emission, biomass burning and/or anthropogenic source (such as cooking and/ or vehicle emissions) in the urban area.
Unsaturated fatty acids are reported to emit directly from many sources such as the leaf surfaces of plants 12 , wood combustion 11 , meat charbroiling 9 , and marine biota 13,44 . Once unsaturated n-fatty acids are emitted into the atmosphere, the double bond in the structures can be oxidized, resulting in aldehydes and dicarboxylic acids 44 . Therefore, the concentration ratios between C (18:0+16:0) and C (18:n+16:1) (C sat /C unsat ) can be used as a proxy to estimate the aging of organic aerosols in the atmosphere. The C sat /C unsat ratios were 1. 5-27 (14) in the daytime versus 1.6-20 (11) in the nighttime (Table 1). Although the difference between daytime and nighttime is insignificant (Table S2), slightly higher ratios of C sat /C unsat still suggest an enhanced photochemical degradation of unsaturated fatty acids in the daytime 17 .  (Tables 1 and S1), however the diurnal variations did not show statistical significance (Table S2). These results are slightly higher than those (6.1-527 ng m −3 ) reported in wintertime urban aerosols in fourteen Chinese megacities 34 . Compared with marine aerosols, the concentrations of n-alcohols in Beijing are one order of magnitude higher than those reported from remote Pacific and Atlantic Oceans aerosols (0.07-8.3 ng m −3 ) 40 . Again, higher concentrations of n-alcohols were observed in the nighttime, especially for C 21 -C 32 species (Figs 1 and 2c). Simoneit 45 reported that biomass burning processes can emit abundant fatty acids and n-alcohols into the air. The temporal patterns of n-alcohols are similar to that of fatty acids (Fig. 1b,c), suggesting that biomass-burning activities can enhance the emission of n-alcohols at a certain level.
The molecular distributions of n-alcohols were characterized by strong even-carbon numbered predominance (CPI = 1.9-10) with two maxima at C 26 and C 28 (Table 1 and Figs 2c and S1c). High molecular weight (HMW, ≥ C 20 ) n-alcohols are abundant in higher plant waxes and can also be emitted into the air by biomass burning 34 . Low molecular weight (LMW, < C 20 ) n-alcohols are limited to soil microbes and marine biota 2 . The ratios of HMW/ LMW ranged from 1.1 to 12 in the daytime and 0.97-16 in the nighttime (Table 1). These results suggest that higher plant waxes, biomass burning emissions and soil resuspension might be the sources of n-alcohols in this region, while their relative contributions varied with time.
n-Alkanes. The δ 13 C values of homologous C 20 to C 32 n-alkanes detected in Beijing aerosols range from − 34.1% to − 24.7% (Table S3). In average, δ 13 C values in the nighttime (− 27.9%) are slightly heavier than those in the daytime (− 28.7%; Fig. 3a). The average δ 13 C values of individual lipids show a saw-tooth pattern (Fig. 4). The average δ 13 C ranges of C 20-25 odd-carbon numbered n-alkanes (LMW odd : − 27.3%) are slightly heavier than C 20-25 even-carbon number homolog (LMW even : -27.7%); the average δ 13 C values of C 27-32 odd-carbon numbered n-alkanes (HMW odd : -29.6%) are slightly depleted in 13 C compared to C 27-32 even-carbon number homolog (HMW even : − 28.7%) (Fig. 4), exhibiting a similar pattern to those previously reported in urban aerosols from Japan 46 . Not all samples showed such saw tooth patterns, for example, samples with higher CPIs such as those collected in the daytime of 20 Jan, 24 Jan, and 1 Feb (Fig. S2). Meanwhile, the δ 13 C values of LMW odd n-alkanes are highly constant (− 25.6 ± 0.34%), suggesting a similar source for C 21 , C 23 and C 25 .
LMW n-alkanes are the dominant species for most of the samples, while HMW odd n-alkanes which are the typical lipid compounds derived from biogenic sources 47 contribute less than LMW n-alkanes in this study (Fig. 1a). Average δ 13 C values of LMW n-alkanes were heavier than HMW odd n-alkanes (Fig. 3a). The CPI LMW (1.3; Fig. 3b) further demonstrate LMW n-alkanes are mainly from anthropogenic emissions, while the higher CPI HMW (2.6) suggest that higher plants contribute more significantly for HMW odd n-alkanes. Parts of the δ 13 C values of HMW n-alkanes fall into the data from C 3 plants again demonstrate this point (Fig. 5a). In a previous study 46 , a positive correlation was found between δ 13 C values of HMW odd n-alkanes against CPIs in urban aerosols from Tokyo (R 2 = 0.5), while such a correlation was weak (R 2 < 0.1 in Beijing). The reason for that is because higher plants contribute less than anthropogenic sources in wintertime Beijing aerosols. The percentage of wax  [21][22][23][24][25][26][27][28][29][30][31] for n-alcohols. b %WaxC n = Σ (C n − 0.5(C n+1 + C n−1 ))/Σ C n × 100%. Negative values of C n were taken as zero.
Scientific RepoRts | 6:27481 | DOI: 10.1038/srep27481 n-alkanes (%Wax C n ) is the contribution of biogenic n-alkane that is derived from higher plants waxes. There a linear relation (R 2 ≥ 0.95, p < 0.01) between %Wax C n and CPI (Fig. S3), which is consistent with that reported in other studies from China (R 2 = 0.91) 48 and Poland (R 2 = 0.85) 37 , demonstrating that %Wax C n here is the effective index to reflect the contribution of plant wax. The average contributions of biogenic n-alkanes to the total n-alkanes (%Wax C n ) are 19% in daytime and 15% in nighttime, respectively (Table 1; Fig. 3b) suggesting minor input from leaf epicuticular waxes.
n-Fatty acids. δ 13 C values of C 16:0 and C 18:0 fatty acids were measured in all the samples. δ 13 C values of even-number fatty acids from C 20:0 to C 28:0 were only available for a few samples collected on polluted days ( Fig. 3c; Table S4). δ 13 C values of C 16:0 and C 18:0 fatty acids range from − 26.9% to − 24.6% and are heavier than HMW fatty acids (− 31.6% to − 27.3%), suggesting the different sources. Fatty acids from plants with C 3 and C 4 photosynthetic pathways differ largely in the isotope fractionations 21 . Stable C isotope ratios of C 16:0 and C 18:0 fatty acids fall into biomass-burning aerosols (source material from C 3 and C 4 plant) that vary from -34.5% to − 25.1% 49,50 (Fig. 5b), further suggesting that biomass-burning emissions are the mainly source of LMW fatty acids. The δ 13 C values of HMW fatty acids have similar ranges compared with those from other urban aerosols that mainly influenced by higher plants; parts of δ 13 C values of HMW fatty acids also fall into C 3 plants and biomass-burning aerosols, implying mixed sources of HMW fatty acids.

Differences of atmospheric lipids between polluted and clear days.
To investigate the variations during the pollution evolution process, the sampling period was divided into two categories according to monitoring results of the Beijing municipal environmental monitoring center: polluted and clear days. The differences   where I and C are the API and the concentration of PM 10 , respectively. C high and C low are values close to C, while I high and I low are values close to I (Table S5). Monitoring of pollutants consists of SO 2 , NO 2 , PM 10 , CO and O 3 . Figure 6 shows the temporal variations of the meteorological parameters (temperature (T; Fig. 6a), relative humidity (RH; Fig. 6b), air pressure (P; Fig. 6c), wind speed (WS; Fig. 6d) and wind direction (WD; Fig. 6e) to further explore the differences between polluted and clear days.
Concentrations of lipids in the nighttime are higher than those in the daytime (Figs 1 and 2); their atmospheric levels on polluted days are approximately six times higher than those on clear days and the differences are statistically significant (p < 0.01; Table S2; Fig. S4). Relative humidity on polluted days (51%) or in the nighttime (36%) are much higher than those on clear days (21%) or in the daytime (25%), respectively. RH can affects the aerosol formation in many ways such as aqueous-phase reactions and gas-particle partitioning and favor the formation of severe haze pollution episodes in Beijing 51 . δ 13 C values of n-alkanes and fatty acids are heavier on polluted days than those on clear days. However, LMW lipids show a minor difference of δ 13 C values on polluted days, suggesting that they have a similar source during haze events (Fig. 3a,c). During the hazy days, air masses mainly originated from the south (Fig. S5), which were associated with relatively higher temperature, lower wind speed (0.5 m s −1 ), lower air pressure and lower heights of atmospheric boundary layer (Fig. 6). Such meteorological conditions made the ground convergence stronger, prevented the diffusion of atmospheric aerosols, leading to the narrow range of δ 13 C values of n-alkanes. Inversely, concentrations of lipids are very low and their δ 13 C values vary significantly on clear days (Figs 1-3). Results of back trajectories show that air masses are mainly from northwest Asian continent on clear days (Fig. S5). Clear air masses from northwestern regions via long-range atmospheric transport bring fresh aerosols and are likely to dilute and mix with local polluted aerosols, causing the complexity and mixed nature of aerosol particles in Beijing.
C 23 n-alkane is a representative marker component of anthropogenic origin 5,17 , while C 29 n-alkane is a typical biomarker that originates from terrestrial higher plants 47 . We plot δ 13 C values of C 16:0 fatty acid and C 23 and C 29 of n-alkanes to further assess the sources of Beijing aerosols. Figure 7 shows that the stable C isotope ratios on polluted days differ obviously from those on clear days, indicating that lipid compounds have different sources. However, δ 13   Long-range atmospheric transport of aerosol particles. The reactivity of the unsaturated fatty acids can be used to determine the source and photochemical aging of atmospheric aerosols 52 . For example, compared with fatty acids derived from local emissions, aged aerosols suffer from photochemical degradation during long-range transport and thus have lower contents of unsaturated fatty acids, which result in a high ratio of C sat /C unsat . In our study, C sat /C unsat ratios on polluted and clear days were 7.2 and 14, respectively. These results provide evidence that aerosol particles were more aged through long-distance transport on clear days (Fig. S5). In contrast, fatty acids were mainly derived from local biological, anthropogenic sources such as cooking, transportation and/or biomass burning on polluted days. This is consistent with air mass back trajectories that they were most likely influenced by local emissions on polluted days and originated from the long-range atmospheric transport of terrestrial aerosols from the northern regions of the continent on clear days (Fig. S5).  Furthermore, we plot the relative abundances of n-alkanes (C 25-36 ), n-fatty acids (C 20-32 ) and n-alcohols (C 20-32 ), the typical terrestrial biomarkers, in a triangular diagram (Fig. 8). The results show that the aerosol samples on polluted days have higher n-alkanes (32-54%) and/or n-fatty acids (25-44%), while higher abundant n-alcohols (26-44%) occurred on clear days. The distribution of these plots in Beijing is different from other aerosol samples from mountain (category A) and marine regions (category B, C, D), which indicates that the main origins of terrestrial organic matter are probably different. Most of the aerosol samples fall in a relatively narrow range, being similar to those of the mountain aerosols collected at the summit of Mt. Tai in the North China Plain during a severe biomass-burning season (category A) 2 , which suggests that biomass-burning emission might be the significant source of terrestrial biomarkers in Beijing aerosols in wintertime. The data from Beijing aerosols are plotted in category B, which were collected under the condition of westerly winds from the Asian continent, indicating that the similarities in the distribution are more likely attributed to an identical source. The Asian continent should be one of the major source regions for terrestrial biomarkers transported over the western Pacific. Although the data from the summer and autumn seasons (category C) and the data from the western Pacific (category D) did not include the samples from Beijing aerosols, categories C and D are much closer to the urban aerosols, which again indicates that Asian aerosols are possibly one of the important sources of terrestrial organic matter in the western Pacific.
The δ 13 C values of n-alkanes and n-fatty acids again support such a point. LMW n-alkanes have different δ 13 C values with HMW odd n-alkanes on polluted days, which are consistent with fatty acids (Fig. 3a,c), suggesting although experienced similar atmospheric pathways and transport processes over the western Pacific, whereas the difference in δ 13 C reflects the different sources of the lipids. As mentioned above, LMW n-alkanes and n-acids on polluted days are mainly originated from anthropogenic source associated with southern air masses. While most of the δ 13 C values of LMW n-alkanes and n-acids detected in Beijing aerosols fall into the downwind urban and marine aerosols 26 (Fig. 5) which were influenced by the Asian continental aerosols through long-range atmospheric transport.

Materials and Methods
Aerosol sampling. Aerosol sampling was performed on the roof-top of a three-story building (10 m above ground level) at the Institute of Atmospheric Physics, Chinese Academy of Sciences (39°58′ 28″ N, 116°22′ 13″ E), Beijing, which is located in the northern part of the northern China plain. Total suspended particle (TSP) samples (n = 29) were collected on a 12-hour basis from 18 January to 1 February 2012. These samples were collected using pre-combusted (6 h in 450 °C in a muffle furnace) quartz filters (20 cm × 25 cm, Pallflex) and a high-volume air sampler at a flow rate of 1 m 3 min −1 . After the sampling, the filters were stored in a pre-combusted glass jar with a Teflon-lined screw cap in a freezer at − 20 °C until analysis.

Measurements of lipids.
An aliquot of filter (ca. 25 cm 2 ) sample was extracted with dichloromethane (DCM)/methanol (MeOH) (2:1, v/v) for three times under ultrasonication for 10 min each. After filtering through Pasteur pipette packing with quartz wool, the supernatant was concentrated with a rotary evaporator and then blown down to dryness with pure nitrogen gas. The extracts were reacted with 50 μ l of N,O-bis-(trimethylsilyl) trifluoroacetamide (BSTFA) with 1% trimethylsilyl chloride and 10 μ l of pyridine at 70 °C for 3 h to derive trimethylsilyl derivatives. After the derivation, the solution was diluted with 140 μ l of n-hexane containing the internal standard (C 13 n-alkane, 1.43 ng μl -1 ) and preserved in a freezer (− 20 °C) until GC/MS analysis.
Lipid class compounds (n-alkanes, fatty acids and n-alcohols) were measured using a Hewlett-Packard model 6890 GC coupled with a Hewlett-Packard model 5973 mass-selective detector (MSD). The GC is equipped with an on-column injector and a DB-5 MS fused silica capillary column (30 m × 0.25 mm, 0.25 μm film thickness). The GC oven temperature was held at 50 °C for 2 min, programmed from 50 to 120 °C at 30 °C min -1 followed by 120 to 300 °C at a rate of 6 °C min -1 and then held isothermally at 300 °C for 16 min. GC/MS data were processed by the Chemstation software. Recoveries for the authentic standards of lipid compounds that were spiked into pre-combusted quartz filters were better than 80%. The relative standard deviation of the concentrations based on duplicate analysis was < 15%.

Determination of compound-specific stable carbon isotopic compositions.
Prior to the measurement of the 13 C/ 12 C ratios of n-alkanes and fatty acids, filter samples were extracted with DCM)/MeOH (2:1, v/v) for three times. The extracts were then passed through LC-NH 2 solid phase extraction (SPE) tubes. The neutral and acid fractions of the lipid extract adsorbed on to the LC-NH 2 SPE tube were eluted with DCM/isopropyl alcohol (2:1, v/v) and ether with 4% acetic acid, respectively, and collected into separate sample vials and dried under nitrogen gas blow. The fraction of n-alkanes was then separated from the neutral fraction of the extract on a silica gel column with hexane and preserved in a freezer (− 20 °C) until analysis. To separate the fatty acids from the acid fraction of the extract, 0.3 ml of toluene and 1 ml of acidified methanol were added to the sample vial and blow down with nitrogen for 5 s and then tightly capped in the vial. The sample vial was then heated at 60 °C for 12 hrs followed by cooling to room temperature. Then, 5 ml of organic free NaCl (5%) water, 1 ml of hexane were added to the sample vial, and the solvent fraction was extracted into another sample vial with a pipette. The solvent extracts were concentrated, and then the fatty acids were separated using a small silica gel column with hexane and preserved in a freezer (− 20 °C) until analysis. δ 13 C of individual n-alkanes (and fatty acids) were determined using a Thermo Trace GC Ultra coupled with a Gas Isotope Ratio MS (MAT 253) via a combustion furnace maintained at a temperature of 1000 °C. The GC was equipped with an on-column injector and a DB-5 MS fused silica capillary column (30 m × 0.25 mm, 0.25 μm film thickness). The GC oven temperature was held at 50 °C for 2 min, programmed from 50 to 120 °C at 30 °C min -1 followed by 120 to 300 °C at a rate of 6 °C min -1 and held isothermally at 300 °C for 16 min -1 .
δ 13 C of individual fatty acid was estimated from the isotopic signature of the methyl esters using the mass balance equation: 13 13 where C Methanol , δ 13 C FA , δ 13 C FAME are the carbon isotope signatures of the methanol, the underivatized fatty acid, and the fatty acid methyl esters, respectively. f Methanol and f FA are the fractions of carbon in the fatty acid methyl esters due to the methanol and the underivatized fatty acid, respectively 26 . The isotope ratios of individual fatty acids were calculated using the δ 13 C Methanol value of − 21.1‰. The analytical error of the δ 13 C values of each compound based on duplicate analysis was < 0.4‰.