Asian Journal of atmospheric environment
[ Research Article ]
Asian Journal of Atmospheric Environment - Vol. 13, No. 2, pp.88-98
ISSN: 1976-6912 (Print) 2287-1160 (Online)
Print publication date 30 Jun 2019
Received 14 Nov 2018 Revised 19 Feb 2019 Accepted 01 Apr 2019
DOI: https://doi.org/10.5572/ajae.2019.13.2.088

Classification and Characterization of Organic Aerosols in the Atmosphere over Seoul Using Two Dimensional Gas Chromatography-time of Flight Mass Spectrometry (GC×GC/TOF-MS) Data

So Hyeon Jeon ; Hyung Bae Lim1) ; Na Rae Choi2) ; Ji Yi Lee2) ; Yun Kyong Ahn3) ; Yong Pyo Kim4), *
Department of Energy, Environmental and Chemical Engineering, Washington University in St.louis, Missouri
1)Air Quality Research Division, National Institute of Environmental Research, Seoul
2)Department of Environmental Science and Engineering, Ewha Womans University, Seoul
3)Western Seoul Center, Korea Basic Science Institute, Seoul
4)Department of Chemical Engineering & Materials Science, Ewha Womans University, Seoul

Correspondence to: *Tel: +82-2-3277-2832 E-mail: yong@ewha.ac.kr


Copyright © 2019 by Asian Journal of Atmospheric Environment
This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Two approaches were applied for characterization of organic compounds in the atmospheric aerosols. These were chemical and statistical grouping analysis. We applied chemical group analysis as a new approach using the GC×GC data. Based on the principle of GC×GC, analytes in two dimensional data were divided into 6 chemical group depending on their volatility and polarity. Using this classification standards, all analytes were classified as one of 6 chemical groups, HP-VOC, LP-VOC, HP-SVOC, LP-SVOC, HPNVOC and LP-NVOC. The relationship between the sum of the peak areas of each chemical group with other variables such as the concentrations of OC, EC, OCPOC, OC/EC and OCSOC were studied. However, no apparent relationship was observed. The principle component analysis (PCA) was carried out to find out the relationship among the analytes based on their sources. The OVOCs were dominant compounds in 3 out of 4 factors in summer. In winter, factor 1 was related to secondary organic aerosol and factor 3 was related to primary emission sources. The proposed approach looks promising in characterization of chemical groups but have limitations at current stage which are discussed.

Keywords:

GC×GC/TOF-MS, Chemical grouping, PCA analysis, Organic aerosols

1. INTRODUCTION

Organic compounds in the atmospheric aerosols occupy a large fraction (20- 80%) of the mass of the fine particulate matter (Kanakidou et al., 2005). Organic compounds are emitted from variety of sources such as combustion of organic matter and vegetation and also formed by atmospheric photo-oxidation reactions. Exposure to organic aerosols may cause several health effects such as respiratory, cardiovascular and neural effects. According to the U.S. EPA, out of the 177 hazardous air pollutants (HAPs) selected for the U.S. EPA 1999 National-Scale Air Toxics Assessment, 167 are either organic compounds or mixtures (Mauderly and Chow, 2008). Organic aerosols influence radiative forcing arising from the aerosol scattering and absorption of radiation directly or perturb the albedo of the Earth’s clouds through their important role as cloud condensation indirectly (Kanakidou et al., 2005; Ramanathan et al., 2001).

Only 10-40% of organic compounds were identified and quantified with unidentified compounds called ‘unknowns’, since organic aerosols are usually mixed with complex matrix and consist of extreme range of species (Goldstein and Galbally, 2007; Kanakidou et al., 2005; Zheng et al., 2002). Therefore organic compounds in aerosol has been studied incompletely such as classifying into two groups (organic carbon, OC and elemental carbon, EC) or selecting and analyzing target compounds such as n-alkanes, carboxylic acid, and polycyclic aromatic hydrocarbons (PAHs) (see, for example, Choi et al., 2016).

Thus, it is essential to identify organic compounds in the ambient aerosols to assess their effects reliably. Therefore, finding unknowns exhibiting their own characteristics according to season or chemical characteristics and identifying these unknowns are important.

To analyze carbonaceous compounds in the atmospheric aerosols, several technique have been used such as gas chromatography/mass spectrometer (GC/MS), liquid chromatography/MS (LC/MS), nuclear magnetic resonance (NMR) and Fourier transform infrared (FTIR) spectroscopy (Hallquist et al., 2009). Although GC is common technique to analyze organic compounds in the atmospheric aerosols, separation efficiency of a single chromatographic technique is still unsatisfactory.

Comprehensive two-dimensional gas chromatography (GC×GC), one of the multidimensional techniques, has been recently used in analysis of organic compounds in atmospheric aerosols (Dallüge et al., 2002). In GC ×GC, two different columns are connected through a modulator. Since GC×GC separates organic compounds in an aerosol sample based on different chemical mechanisms in two columns, it has high resolving power and provide enormous number of peaks than conventional GC. Commonly used columns in GC× GC are non-polar column in first column and polar column in second column. Separation takes place according to vapor pressure of analytes in the first column and polarity in the second column. Modulator improve sensitivity by cutting and refocusing narrow adjacent fractions of the first column eluate, focusing and releasing these rapidly into the second column (Dallüge et al., 2002). Xu et al. (2003) identified 235 volatile organic compounds (VOCs) in situ measurements using GC× GC coupled with a time-of-flight mass spectrometer (TOF-MS). Kallio et al. (2003) identified and quantified 23 PAHs and oxygenated PAHs (oxy-PAHs) in urban aerosols collected in Helsinki, Finland using GC×GC coupled with flame ionization detector (FID) and quadrupole mass spectrometer (QMS). Kallio et al. (2006) identified 50 organic compounds in atmospheric aerosols from coniferous forest using GC×GC-TOF-MS. Lee et al. (2009) identified about 300 organic compounds in PM2.5 collected in Seoul using thermal-desorption (TD) GC×GC-TOF-MS. GC×GC has been also applied to analyze organic aerosols from emission sources such as cigarette smoke, particulate matter from wood combustion and industrial chemicals (Weggler et al., 2016; Alam et al., 2013; Dallüge et al., 2002).

Application of GC×GC-TOFMS for analysis of organic aerosols in the atmosphere contributed to provide much more and specific information of organic aerosols. Still, researchers have had difficulty in data processing methods dealing with huge information.

Welthagen et al. (2003) identified 697 organic compounds (out of 1,060) in atmospheric aerosols and characterized unidentified chemical compounds depending on their volatility and polarity. According to their volatility, 1st dimension retention time was divided into half. Then chemical compounds detected in the front retention time were considered as ‘volatile’ compounds and those detected in the rear retention time were considered as ‘semi-volatile’ compounds. In the same way, 2nd dimension retention time reflecting their polarity divided the compounds into two parts, ‘polar’ for the compounds detected in the front retention time and ‘nonpolar’ compounds detected in the rear retention time, respectively. As a result, unidentified chemical compounds were classified 4 group; (i) polar volatile, (ii) non-polar volatile, (iii) polar semi-volatile and (iv) nonpolar semi-volatile. Based on this classification, their study tried to suggest detailed criterion for dividing 1st and 2nd retention time (Welthagen et al., 2003).

this study, we used two kinds of grouping methods for characterization of unknowns, since it is needed to extract unknowns exhibiting their own characteristics through group analysis before identification of unknowns. First, chemical grouping was carried out with volatility and polarity following the approach of Welthagen et al. (2003). We investigated seasonal and diurnal variation of each chemical group. Second method was statistical analysis by using the principle component analysis (PCA) method to group the compounds with similar sources or transformation characteristics.


2. DATA

The total suspended particle (TSP) samples were collected at Asan engineering building, Ewha Womans University, Seoul, Korea (37.56°N, 126.94°E, 20 m above ground level). The sampling site is located in the mixed resident area, commercial area, forest area and nearby roadside. A total of 67 samples were obtained during summer (August 12-30, 2013, n=32) and winter ( January 27-February 16, 2014, n=35) and day (9 a.m.-6 p.m., n=16 in summer, n=17 in winter) and night (8 p.m.-6 a.m., n=16 in summer, n=18 in winter) with a PUF sampler (Tisch, TE-1000) on quartz fiber filter (Quartz fiber filter, QFF, Ø10.16 cm, Whatman, UK). Prior to sampling, quartz fiber filters were baked for 8 hours in an electric oven at 550°C to remove possible organic contaminants. The sampled filters were wrapped in aluminium foils and stored in a freezer at -20°C until analysis.

Sampled filters were extracted twice using accelerated solvent extractor (ASE) (Dionex ASE-200) with a mixture of dichloromethane and methanol (3 : 1, v/v) at 40°C and 1,700 psi for 5 minutes. Prior to extraction 4 internal standards (naphthalene-d8, acnaphthene-d10, phenanthrene-d10, and fluoranthene-d10) were spiked in the sampled filters to estimate recovery of the extraction procedure. Extracts were blown down to 1 mL using a nitrogen evaporator (Turbo Vap II, caliper Life Sciences). Concentrated samples were transferred to 2 mL vial and blown down again to 0.5 mL using a nitrogen evaporator.

GC×GC-TOF-MS analysis was carried out using a cryo-jet Pegasus 4D (LECO, St. Joseph, MI, USA) equipped with an Agilent 7890B gas chromatography and a Pegasus time-of-flight mass spectrometer (TOF-MS). A 1 μL sample was injected on the splitless mode at 300°C. The chromatographic separation was executed by volatility in first dimension column and by polarity in second dimension column. The first dimension column was DB-5MS (30 m×0.25 mm I.D; 0.25-μm film thickness) from J&W Scientific capillary column (Santa Clara, CA, USA). The second dimension column was DB- 17MS (1.17 m×0.18 mm I.D; 0.18-μm film thickness) from J&W Scientific capillary column. The GC was held at 60°C for 1 minute and then raised at 6°C/min to 300°C for 15 minutes. The carrier gas was helium (99.999%) flowed at 1.3 mL/min. The secondary oven temperature was 5°C higher than GC oven temperature. The modulator was operated at 15°C higher than second oven temperature. The transfer line temperature was 250°C and the ion source temperature was 230°C. The TOF-MS was operated in electron impact (EI) mode at 70 eV electron energy. The TOF-MS mass range was set at 40-550 amu. ChromaTOF software was used for all acquisition control and data processing. To align analytes of each daily samples, “statistical compare” in ChromaTOF software was used.

Generally, the average concentrations of primary air pollutants in the winter sampling period were higher than the summer sampling period (Choi et al., 2018). For example, the average concentration of Polycyclic Aromatic Hydrocarbons (PAHs) in winter and summer were 17.23 and 39.94 ng/m3, respectively. Also the average concentrations of sulfur dioxide, nitrogen dioxide and carbon monoxide in winter were higher than summer by the factors of 1.4, 1.9 and 4.0, respectively.


3. RESULTS AND DISCUSSION

3. 1 Chemical Group Separation

While Welthagen et al. (2003) classified all analytes in the ambient aerosols sample to 4 chemical groups, we divided them to 6 according to volatility and polarity. In 2-dimensional (2D) chromatogram, 1st dimension retention time reflected volatility (vapor pressure) of analytes and 2nd dimension retention time reflected polarity of analytes. We established a clear criterion to separate analytes on the basis of retention time of specific analytes.

To separate analytes according to their volatility, retention time of naphthalene (Pv=11.332 Pa at 25°C) and benzo[a]pyrene (Pv=7.319×10-9 Pa at 25°C) were used as bases. Because naphthalene is typically observed in the gas phase, analytes detected in less than retention time of naphthalene were considered as a VOC. Meanwhile, benzo[a]pyrene has been typically observed in the particle phase and analytes detected in more than retention time of benzo[a]pyrene were considered nonvolatile organic compounds (NVOC). The other analytes detected between retention time of naphthalene and benzo[a]pyrene were grouped as semi-volatile organic compounds (SVOC).

For polarity, retention time of hentriacontane was used as a basis. Polarity of organic compounds are decided by the electronegativity difference between atoms in a bond. Since the functional group primarily affects the polarity of the chemical compounds, the polarity of chemical compounds can be ranked by a functional group. Amide is the most polar, followed by acid, alcohol, ketone, aldehyde, amine, ester, ether and alkane. Alkanes are generally considered as non-polar compounds, therefore, alkane was chosen as an indicator of polarity in this study. Among n-alkanes, hentriacontane was used as a separating compound because it is the last observed compounds in n-alkanes group on second dimension time. The analytes detected in less than retention time of hentriacontane were considered as similar or less polar than n-alkane (LP), the other as higher polar than nalkanes (HP). Fig. 1 shows the scheme of chemical group separation by volatility and polarity in 2D chromatogram. Each group was named as HP-VOC, LP-VOC, HP-SVOC, LP-SVOC, HP-NVOC and LP-NVOC, respectively.

Fig. 1.

Schematic diagram of chemical group separation by volatility and polarity in 2D chromatogram. Red letters and circles are correlation between partial analytes in chemical separation group with the PCA result.

There are two variables one can use to interpret the result, peak area and peak number. Ideally, both variables should be used but it is hard to use both since there is too much information to start with. The main aim of this study is to characterize major compounds in organic aerosols and if peak number is used as a variable, there would be too many compounds the concentrations of which are very low. Thus, we judged that peak number be not appropriate for the current study and used peak area as the main variable. To use both peak area and peak number, further studies are warranted.

3. 2 Seasonal and Diurnal Variation

To understand the behavior of each group by season and day, we summed the peak area of all analytes detected in each chemical group including unidentified compounds. Fig. 2 shows the seasonal variation and contribution of each group. The sum of the peak areas in summer was larger than winter by a factor of ~2. The VOC group contributed less than 5% of the summed area irrespective of their polarity. Meanwhile, the SVOC group showed difference between summer and winter in contribution according to their polarity. The HP-SVOC contributed 43.6% in summer, while in winter it contributed less than half in summer, 18.9%. The LP-SVOC consisted of more than 50% of the whole organic part in both season. In winter, the LP-SVOC was 3 times larger than the HP-SVOC, but the variation of the LPSVOC and the HP-SVOC group in summer was less than 10%. The HP-VOC group contributed the least fraction among the chemical groups, 1.22% in summer and 0.11% in winter, respectively. The NVOC group had a small fraction in whole peaks regardless of their polarity in summer. However, in winter, the LP-NVOC contributed 16.4% which was almost 14 times higher than the HP-NVOC.

Fig. 2.

The average seasonal variation (a) and ratio of contribution (b) of each chemical group using the sum of the peak areas.

Fig. 3 shows the diurnal variation and contribution of each group in summer and winter. In summer, the sum of the peak areas showed clear difference between day and night. The summed areas at nighttime was almost 2 times larger than daytime. While in winter, it showed almost same total peak area whose variation between day and night is less than 1%. The composition of each group showed similar patterns at both day and night. In summer, the VOC and NVOC group contributed less than 5%, while the SVOC group accounted for about 40% in HP and 50% in LP. However, in winter, the NVOC group, particularly LP-NVOC, grew by 17%. In the SVOC group, the fraction of the HP-SVOC declined from summer to winter, from 41.1% to 14.0% during daytime and from 43.6% to 21.2% at nighttime, respectively. On the contrary, the fraction of the LP-SVOC increased from summer to winter, from 54.0% to 66.4% during daytime and from 52.3% to 59.5% at nighttime, respectively. As a result, the contribution of low polar group increased in winter.

Fig. 3.

Diurnal variation (a) and ratio of contribution (b) of each chemical group using the sum of the peak areas.

To find out relationship between each chemical group and emissions of atmospheric aerosols, correlation analysis was carried out between the sum of peak area of each chemical group and the concentrations of OC, elemental carbon (EC), primary organic carbon (OCPOC) which was a marker of primary organic aerosols, the OC/EC ratio and secondary organic carbon (OCSOC) which was a marker of secondary organic aerosols. OCPOC and OCSOC were calculated following the method in Lim and Turpin (2002). However, most chemical groups had no apparent correlation with OC, EC, OCPOC, OC/EC and OCSOC. One example of the relationship between the summed area and other variables is shown in Fig. 4.

Fig. 4.

Scattering plot between the daily sum of peak area in LP-NVOC group, organic carbon and elemental carbon.

3. 3 Principle Component Analysis (PCA)

To carry out a statistical analysis, the PCA was performed with the SPSS statistical package (version 20). To compare the data in summer and winter, only the peak appeared at least more than 50% the samples in each season was counted. The factor having more than 10% of variance and individual analytes having more than 0.6 of factor loadings in each factor was analyzed. Each analytes with their peak area were used as variables for PCA with the varimax rotation method. Factor loading is correlation coefficient between each variables and the factor. Then, individual analyte was classified for determining factor characteristics such as oxygenated volatile organic compounds (OVOCs), volatile organic compounds (VOCs), PAHs, organic nitrogen compounds (ON), organophsphates, organosulphates and phthalate as used in Alam et al. (2013). OVOCs and VOCs were further divided into linear and branched compounds, substituted aromatic compounds and cyclic compounds. PAHs included both PAHs and oxygenated PAHs (OPAHs).

In summer, 4 factors were extracted each of which explained more than 10% of variance each and analytes having more than 0.6 of factor loadings. The number of extracted analytes in factor 1 was 30. Table 1 shows chemical information and classification of each analytes in factor 1. 21 out of 30 analytes were classified as OVOCs and 2 analytes were unknowns. In factor 2, the number of analytes was 28. Among them, 12 analytes were classified as unknowns and other analytes were evenly distributed in OVOCs, VOCs and PAHs. In factor 3, 24 analytes were extracted. Unknowns and OVOCs were similarly contributed in factor 3, 10 and 9, respectively. In factor 4, 18 analytes showed strong correlation with factor 4. Of them, 13 analytes were OVOCs and 2 analytes were unknowns. In other words, OVOCs were dominant in 4 factors in summer except factor 2. It is not clear what were the major sources of each factor in summer.

Chemical information of analytes in factor 1 in summer.

In winter, 3 factors were extracted with more than 10% of variance. In factor 1, 55 analytes indicated more than 0.6 of factor loadings. Number of unknowns was 22 and was the highest followed by OVOCs, 17. In factor 2, 30 analytes were extracted and 29 of them were unknowns except only one analytes. In factor 3, 19 analytes showed high factor loadings and 13 analaytes were identified in PAHs (Table 2). Its suggested that factor 3 in winter might be related to primary emission sources because PAHs are representative markers for primary organic aerosols. In addition to factor 3, factor 1 in winter appeared to be related to specific sources. Since factor 1 contained mostly unknowns and OVOCs, factor 1 might be related to secondary organic aerosols.

Chemical information of analytes in factor 3 in winter.

Through PCA, a large number of analytes in the atmospheric aerosols could be grouped according to presence of correlation between each other. In case of winter, each factors showed certain characteristics through chemical class. While, in case of summer, clear characteristics of each factor were not found though the presence of dominant OVOCs might suggest relationship with secondary organic aerosols.

3. 4 Limitations of the Results

There are a few limitations in the chemical grouping analysis. First, there might be inconsistency between theory and practice. We carried out the chemical grouping under the assumption that analytes would be separated in gas chromatography depending on their volatility and polarity sequentially. However, some analytes might not be separated as expected. Matrix effect is one of the reasons. Generally, most studies analyzing chemical compounds have been focused on specific target compounds and, thus, appropriate analytical method was adopted to eliminate matrix interference. However, in this study, some compounds were not separated as expected. For example, most PAHs which were nonpolar in the atmospheric samples were identified as high polarity compounds (HP-SVOC group). Also, many oxygenated volatile organic compounds which should be polar since an n-alkane compound was used as polarity standard (oxygenated compounds are theoretically more polar than n-alkanes) were classifies as low polarity compounds (LP-SVOC group).

Also, instrumental limitation might be also one of reasons. We used two columns filled with similar stationary phase used in Welthagen et al. (2003) to have similar analysis conditions. Among them, DB-17MS which was second column was known to be less reliable to separate chemical compounds depending on their polarity than DB-WAX. However, since DB-WAX has lower maximum temperature limit (240°C) than DB-17MS (320°C), DB-WAX could not be used in this study. It might influence on separation ability of analytes depending on polarity.

Other effect is certain patterns obtained from GC× GC data depending on their chemical class. As mentioned before, in GC×GC analytes separated in first column are focused in modulator and released to the second column. The difference with conventional 1-dimensional GC was that, since each analyte in each peak is modulated into several second-dimension peaks, same analytes can be observed in different retention time. Considering this property, we summed all peak areas of same analytes and chose the last retention time for chemical group separation. Therefore, this data processing could be influenced on chemical group separation. Furthermore, same chemical class (for example, n-alkanes and PAHs) are detected in a series of retention time in GC×GC data. In this case, analytes that should be classified in the same chemical class could be included in different chemical groups.

Finally, there are “too many” analytes to be separated according to several standards. Each analytes has its own characteristics such as emission source, chemical structure, etc. Since they were not separated clearly from their other properties, partial analytes in chemical group can already have relationship with other analytes in different chemical separation group. This was indicated when we applied chemical separation group to the PCA results. Fig. 1 also shows scheme of correlation between partial analytes in different chemical separation group with PCA results. Grouping by PCA means that each analyte has correlation each other. Then grouping of these correlated analytes by “chemical group separation” means that each analyte has their own correlation according to other parameters including chemical property.

Though there were several possible limitations in the approach used in this study, this approach is promising one and further studies are warranted such as more detailed and advanced library and dual oven system to accommodate each column.


4. CONLUSIONS

characterization of unknown peaks in the atmospheric aerosols one of which was chemical and the other was statistical analysis. We applied chemical group analysis as a new approach using GC×GC data referred as Welthagen et al. (2003). Based on principle of GC×GC, analytes in 2D data were divided into 6 chemical group depending on their volatility and polarity. Specific analytes were chosen for separation standards. Naphthalene and benzo[a]pyrene were used for separation depending on their volatility. Hentriacontane was used for separation depending on their polarity. Using these classification standards, all analytes were classified as one of 6 chemical groups, HP-VOC, LP-VOC, HP-SVOC, LPSVOC, HP-NVOC and LP-NVOC.

To investigate whether this chemical grouping is related to emission source or formation of secondary organic aerosol, relationship between the sum of the peak areas of the chemical groups with other variables such as the concentrations of OC, EC, OCPOC, OC/EC and OCSOC were studied. However, no apparent relationship was observed.

For statistical analysis, the principle component analysis (PCA) was carried out. The factors explaining more than 10% of variance in summer and winter were separated. The OVOCs were dominant compounds in 3 out of 4 factors in summer. In winter, factor 1 was related to secondary organic aerosol and factor 3 was related to primary emission sources. Through PCA, information of unknowns about source of atmospheric aerosols could be obtained, though partially.

The proposed approach looks promising in characterization of chemical groups but have limitations at current stage. Further studies are warranted to fully utilize the proposed approach.

Acknowledgments

This work was supported by National Research Foundation of Korea (NRF-2017R1A2B4006760).

References

  • Alam, M.S., West, C.E., Scarlett, A.G., Rowland, S.J., Harrison, R.M., (2013), Application of 2D-GCMS reveals many industrial chemicals in airborne particulate matter, Atmospheric Environment, 65, p101-111. [https://doi.org/10.1016/j.atmosenv.2012.10.014]
  • Choi, N.R., Ahn, Y.G., Lim, H.B., Lee, J.Y., Jung, C.H., Kim, Y.P., (2018), Particulate nitrosamines in the atmosphere at Seoul and their major sources, Air Quality, Atmosphere and Health, 11(7), p855-865. [https://doi.org/10.1007/s11869-018-0594]
  • Choi, N.R., Lee, S.P., Lee, J.Y., Jung, C.H., Kim, Y.P., (2016), Speciation and source identification of organic compounds in PM10 over Seoul, South Korea, Chemosphere, 144, p1589-1596. [https://doi.org/10.1016/j.chemosphere.2015.10.041]
  • Dallüge, J., Vreuls, R.J., Beens, J., Brinkman, U.A.T., (2002), Optimization and characterization of comprehensive two-dimensional gas chromatography with time-of-flight mass spectrometric detection (GC×GC-TOF MS), Journal of Separation Science, 25(4), p201-214. [https://doi.org/10.1002/1615-9314(20020301)25:4<201::AID-JSSC201>3.0.CO;2-B]
  • Goldstein, A.H., Galbally, I.E., (2007), Known and unexplored organic constituents in the earth’s atmosphere, Environmental Science & Technology, 41(5), p1514-1521. [https://doi.org/10.1021/es072476p]
  • Hallquist, M., Wenger, J., Baltensperger, U., Rudich, Y., Simpson, D., Claeys, M., Dommen, J., Donahue, N., George, C., Goldstein, A., (2009), The formation, properties and impact of secondary organic aerosol: current and emerging issues, Atmospheric Chemistry and Physics, 9, p5155-5236. [https://doi.org/10.5194/acp-9-5155-2009]
  • Kallio, M., Hyötyläinen, T., Lehtonen, M., Jussila, M., Hartonen, K., Shimmo, M., Riekkola, M.-L., (2003), Comprehensive two-dimensional gas chromatography in the analysis of urban aerosols, Journal of Chromatography A, 1019, p251-260. [https://doi.org/10.1016/S0021-9673(03)01238-X]
  • Kallio, M., Jussila, M., Rissanen, T., Anttila, P., Hartonen, K., Reissell, A., Vreuls, R., Adahchour, M., Hyotylainen, T., (2006), Comprehensive two-dimensional gas chromatography coupled to time-of-flight mass spectrometry in the identification of organic compounds in atmospheric aerosols from coniferous forest, Journal of Chromatography A, 1125(2), p234-243. [https://doi.org/10.1016/j.chroma.2006.05.050]
  • Kanakidou, M., Seinfeld, J., Pandis, S., Barnes, I., Dentener, F., Facchini, M., Dingenen, R.V., Ervens, B., Nenes, A., Nielsen, C., (2005), Organic aerosol and global climate modelling: a review, Atmospheric Chemistry and Physics, 5, p1053-1123. [https://doi.org/10.5194/acp-5-1053-2005]
  • Lee, J.-Y., Lane, D.A., Huh, J.-B., Yi, S.-M., Kim, Y.P., (2009), Analysis of organic compounds in ambient PM2.5 over Seoul using thermal desorption-comprehensive two dimensional gas chromatography-time of flight mass spectrometry (TDGC×GC- TOFMS), Journal of Korean Society for Atmospheric Environment, 25(5), p420-431. [https://doi.org/10.5572/KOSAE.2009.25.5.420]
  • Lim, H.-J., Turpin, B.J., (2002), Origins of primary and secondary organic aerosol in Atlanta: Results of time-resolved measurements during the Atlanta supersite experiment, Environmental Science & Technology, 36(21), p4489-4496. [https://doi.org/10.1021/es0206487]
  • Mauderly, J.L., Chow, J.C., (2008), Health effects of organic aerosols, Inhalation Toxicology, 20, p257-288. [https://doi.org/10.1080/08958370701866008]
  • Ramanathan, V., Crutzen, P., Kiehl, J., Rosenfeld, D., (2001), Aerosols, climate, and the hydrological cycle, Science, 294(5549), p2119-2124. [https://doi.org/10.1126/science.1064034]
  • Weggler, B.A., Ly-Verdu, S., Jennerwein, M., Sippula, O., Reda, A.A., Orasche, J.r., Gröger, T., Jokiniemi, J., Zimmermann, R., (2016), Untargeted identification of wood type-specific markers in particulate matter from wood combustion, Environmental Science & Technology, 50(18), p10073-10081. [https://doi.org/10.1021/acs.est.6b01571]
  • Welthagen, W., Schnelle-Kreis, J., Zimmermann, R., (2003), Search criteria and rules for comprehensive two-dimensional gas chromatography-time-of-flight mass spectrometry analysis of airborne particulate matter, Journal of Chromatography A, 1019, p233-249. [https://doi.org/10.1016/j.chroma.2003.08.053]
  • Xu, X., Stee, L., Williams, J., Beens, J., Adahchour, M., Vreuls, R., Brinkman, U., Lelieveld, J., (2003), Comprehensive twodimensional gas chromatography (GC×GC) measurements of volatile organic compounds in the atmosphere, Atmospheric Chemistry and Physics, 3, p665-682. [https://doi.org/10.5194/acp-3-665-2003]
  • Zheng, M., Cass, G.R., Schauer, J.J., Edgerton, E.S., (2002), Source apportionment of PM2.5 in the southeastern United States using solvent-extractable organic compounds as tracers, Environmental Science & Technology, 36(11), p2361-2371. [https://doi.org/10.1021/es011275x]

Fig. 1.

Fig. 1.
Schematic diagram of chemical group separation by volatility and polarity in 2D chromatogram. Red letters and circles are correlation between partial analytes in chemical separation group with the PCA result.

Fig. 2.

Fig. 2.
The average seasonal variation (a) and ratio of contribution (b) of each chemical group using the sum of the peak areas.

Fig. 3.

Fig. 3.
Diurnal variation (a) and ratio of contribution (b) of each chemical group using the sum of the peak areas.

Fig. 4.

Fig. 4.
Scattering plot between the daily sum of peak area in LP-NVOC group, organic carbon and elemental carbon.

Table 1.

Chemical information of analytes in factor 1 in summer.

Compound CAS no Molecular formula Class 1st RT 2nd RT Chemical group separation
7,9-Di-tert-butyl-1-oxaspiro(4,5)deca-6,9-diene-2,8-dione 82304-66-3 C17H24O3 Cyclic OVOCs 27 : 49.3 00 : 01.6 HP-SVOC
cis-9,10-Epoxyoctadecanamide 172995-07-2 C18H35NO2 Linear and branched ON 37 : 09.0 00 : 02.0
9,12,15-Octadecatrienal 26537-71-3 C18H30O Linear and branched OVOC 35 : 48.2 00 : 01.8
9,12-Octadecadienal 26537-70-2 C18H32O Linear and branched OVOC 35 : 33.6 00 : 01.8
9,12-Octadecadienoic acid, methyl ester, (E,E)- 2566-97-4 C19H34O2 Linear and branched OVOC 36 : 14.1 00 : 01.8
2,5-Cyclohexadiene-1,4-dione, 2,6-bis(1,1-dimethylethyl)- 719-22-2 C14H20O2 Cyclic OVOCs 19 : 33.9 00 : 01.2
Furan, 2-[(2-ethoxy-3,4-dimethyl-2-cyclohexen-1-ylidene)methyl]- 55162-49-7 C15H20O2 Cyclic OVOCs 22 : 45.4 00 : 01.3 LP-SVOC
2-Undecenal 2463-77-6 C11H20O Linear and branched OVOC 17 : 24.0 00 : 01.2
2-Decenal, (E)- 3913-81-3 C10H18O Linear and branched OVOC 15 : 04.2 00 : 01.1
Nonanoic acid, methyl ester 1731-84-6 C10H20O2 Linear and branched OVOC 14 : 04.8 00 : 01.1
Decanal 112-31-2 C10H20O Linear and branched OVOC 13 : 43.2 00 : 01.1
Eicosanoic acid, 2-hydroxyethyl ester - C22H44O3 Linear and branched OVOC 35 : 43.7 00 : 01.4
Nonadecane, 2-methyl- 1560-86-7 C20H42 Linear and branched VOCs 24 : 38.5 00 : 00.9
Eicosane 112-95-8 C20H42 Linear and branched VOCs 26 : 45.0 00 : 00.9
Decane, 2,3,5,8-tetramethyl- 192823-15-7 C14H30 Linear and branched VOCs 17 : 28.9 00 : 00.9
Hexadecane 544-76-3 C16H34 Linear and branched VOCs 21 : 37.2 00 : 00.9
Hordenine 539-15-1 C10H15NO Substituted aromatic ON 20 : 50.4 00 : 01.0
1-Octanone, 1-phenyl- 1674-37-9 C14H20O Substituted aromatic OVOC 23 : 43.2 00 : 01.4
Phenol, 2,4-bis(1,1-dimethylethyl)- 96-76-4 C14H22O Substituted aromatic OVOC 20 : 15.2 00 : 01.3
Analyte 504 Unknown 34 : 16.8 00 : 01.3
Analyte 271 Unknown 21 : 24.0 00 : 01.4
Analyte 420 Unknown 30 : 12.0 00 : 01.2
Furan, 2-pentyl- 3777-69-3 C9H14O Cyclic OVOCs 08 : 36.0 00 : 00.9 LP-VOC
2-Nonanone 821-55-6 C9H18O Linear and branched OVOC 10 : 55.2 00 : 01.1
2-Octanone 111-13-7 C8H16O Linear and branched OVOC 11 : 30.7 00 : 01.1
Octanal 124-13-0 C8H16O Linear and branched OVOC 08 : 51.6 00 : 01.0
Nonanal 124-19-6 C9H18O Linear and branched OVOC 11 : 01.2 00 : 01.1
Heptanoic acid, methyl ester 106-73-0 C8H16O2 Linear and branched OVOC 09 : 18.8 00 : 01.0
2-Octenal, (E)- 2548-87-0 C8H14O Linear and branched OVOC 10 : 12.0 00 : 01.1
Octanoic acid, methyl ester 111-11-5 C9H18O2 Linear and branched OVOC 11 : 41.7 00 : 01.1

Table 2.

Chemical information of analytes in factor 3 in winter.

Compound CAS no Molecular formula Class 1st RT 2nd RT Chemical group separation
Benzo[a]pyprene 207-08-9 C20H12 PAHs 40 : 10.6 00 : 02.9 HP-NVOC
(E)-2-formylcinnamaldehyde 61650-52-0 C10H8O2 OPAHs 21.88 2.02 HP-SVOC
9H-Fluoren-9-one 486-25-9 C13H8O OPAHs 25 : 10.1 00 : 02.0
1,2-Naphthalic anhydride 5343-99-7 C12H6O3 OPAHs 26.75 1.87
Phthalic anhydride 85-44-9 C8H4O3 OPAHs 30 : 07.2 00 : 02.6
1-Phenanthrenecarboxaldehyde, 1,2,3,4,4a,9,10,10a-octahydro- 1,4a-dimethyl-7-(1-methylethyl)-, [1S-(1,4a,10a]- 13601-88-2 C20H28O OPAHs 29 : 52.8 00 : 01.5
Phenanthrene 85-01-08 C14H10 PAHs 25.94 1.73
Fluoranthene 206-44-0 C16H10 PAHs 35 : 28.8 00 : 02.3
Naphthalene, 2-phenyl- 612-94-2 C16H12 PAHs 29 : 09.6 00 : 01.9
Pyrene 129-00-0 C16H10 PAHs 31 : 06.0 00 : 02.1
Benzo[ghi]fluoranthene 203-12-3 C18H10 PAHs 35 : 28.8 00 : 02.3
Chrysene 218-01-9 C18H12 PAHs 36.37 2.26
1-Penten-3-one, 1-phenyl- 3152-68-9 C11H12O Substituted aromatic OVOCs 21 : 57.6 00 : 02.2
Analyte 709 43 : 50.4 00 : 00.5 LP-NVOC
n-Hexadecanoic acid 00 : 00.0 C16H32O2 Linear and branched OVOCs 28 : 32.5 00 : 01.3 LP-SVOC
Octadecanoic acid 00 : 00.0 C18H36O2 Linear and branched OVOCs 31 : 44.3 00 : 01.3
Pentadecane 629-62-9 C15H32 Linear and branched >VOCs 30 : 25.8 00 : 01.0
Eicosane 112-95-8 C20H42 Linear and branched VOCs 26 : 45.0 00 : 00.9
Heneicosane 629-94-7 C21H44 Linear and branched VOCs 28 : 53.6 00 : 01.0