2.1 Sample collection

Four ships using HFO and diesel and four excavators covering different emission standards and powers were chosen as representative off-road engines in China. Detailed information about the four ships and four excavators is presented in Table 1. Before conducting field sampling, the original fuel was obtained directly from the fuel tank and sent to the testing company for quality evaluation (Table 2). For excavator emission sampling, three operation modes (idling, moving, and working) were selected, and sampling time was approximately half an hour for each mode. The flow rate of the PM_2.5 sampler was 10 L min⁻¹. Commercial equipment (MFD25, produced by Shanghai Besser Environmental Protection Technology Co., Ltd.) was used for PM_2.5 sampling. The description of the particulate matter dilution and sampling system are presented in Xia's published study (Xia, 2017). In short, the exhaust plume was pumped into a retention chamber and PM_2.5 was intercepted by the four PM_2.5 samplers. One Teflon and three quartz filters (Φ=47 mm) were finally acquired, and one of the quartz filters was used to determine the chemical composition of PM_2.5 for each excavator in the three operation modes. For vessel emissions, an onboard measurement system was used, as previously described (Cui et al., 2017; Deng et al., 2017). The onboard measurement system contains one dilution tunnel connected to two particulate samplers. Finally, particulate matter was collected on two quartz filters (Φ=90 mm) for each vessel.

Table 1Technical parameters of test off-road engines.

Download Print Version | Download XLSX

Table 2Results of the fuel quality analysis.

Download Print Version | Download XLSX

2.2 Chemical analysis

Due to the limitations of organic matter load in filters and cost-prohibitive analysis, the filters sampled from off-road engines with the same operation modes or fuel quality were combined together to characterize the comprehensive molecular compositions of POCs for off-road engines in different operation modes and fuel quality. As shown in Table S1, five samples (1, 2, 3, 4, and 5) were selected to conduct FT-ICR MS analysis, which represented vessels using heavy fuel oil, vessels using diesel, excavators in idling, moving, and working modes, respectively. Sample 1 was combined with 25 % of the filter area from the two HFO-fueled vessels, namely YK and YF; sample 2 was combined with 25 % of filter area from two diesel-fueled vessels, namely GB1 and TB4; samples 3, 4, and 5 were combined with 50 % of the filter areas from four excavators in idling, moving, and working modes, respectively, namely CAT320, CAT330B, CAT307, and PC60. The portions of filters (Table S1 in the Supplement) were cut and combined for 40 min, subjected to ultrasonic extraction with 40 mL ultrapure water, and then filtered using a 0.22 µm PTFE membrane (Jinteng, China). The extraction solvent was then divided into three portions. Two portions were used for measuring the concentrations of organic carbon and optical absorbance, as described in the Supplement, Sects. S1 and S2. The third portion was processed to assess the chemical composition of POCs by FT-ICR MS. The remaining extraction filters were frozen and then dried to remove ultrapure water. The filters were then subjected to 40 min of ultrasonic extraction with 36 mL dichloromethane and 4 mL methanol. The extracted solvent was divided into three portions, and these were processed in the same way as previously described for the ultrapure water extract.

Both samples extracted with water or organic solvents were processed by a solid phase extraction (SPE) method to remove ions, which disturbed the results of FT-ICR MS. The majority of inorganic ions (e.g., ammonium, sulfate, and nitrate) and low-molecular-weight organic compounds such as isoprene-derived organosulfates and sugars could be removed during SPE treatment (Gao et al., 2006; Lin et al., 2012b; Surratt et al. 2007), which were not discussed in this research. The details of the solid phase extraction method were presented by Mo et al. (2018). Briefly, the pH value of water extracts was adjusted to 2.0 using HCl and then passed through an SPE cartridge (Oasis HLB, 30 µm, 60 mg/cartridge, Waters, USA). The adsorbed POCs were eluted with 6 mL 2 % (v∕v) ammonia∕methanol and dried under a gentle stream of N₂. Finally, the POCs were re-dissolved using 10 mL ultrapure water.

2.3 FT-ICR MS analysis

The molecular characterization of POCs was undertaken using negative-ion electrospray ionization (ESI) FT-ICR MS (Bruker Daltonics GmbH, Bremen, Germany) with a 9.4 T refrigerated actively shielded superconducting magnet. Extracted solutions were injected at flow rate of 180 µL h⁻¹ through an Apollo II electrospray source. Emitter voltage, capillary column introduction voltage, and capillary column end voltage for negative-ion formation were 3.0 kV, 3.5 kV, and −320 V, respectively. The scan range was m∕z 100–900 with a resolution > 450 000 at $m/z=\mathrm{319}$ with < 0.4 ppm absolute mass error. During analysis, nitrogen-containing compounds were used as an internal calibration. Finally, the spectrum peaks with a ratio of signal to noise higher than 10 were exported.

CcHhOoNnSs was used as a general formula, since some criteria should be ascertained to assign the possible formula (Wang et al., 2017). Briefly, all of the mathematically possible formulas for each ion were calculated with a mass tolerance of ±2 ppm. The H-to-C, N-to-C, O-to-C, and S-to-C ratios were limited to 0.3–3.0, 0–0.5, 0–3.0, and 0–2.0, respectively, in the ESI⁻ mode. Peak magnitude is not indicative of a compound's concentration in a sample due to inherent biases of SPE extractions and electrospray ionization efficiencies (Wozniak et al., 2008). Therefore, the relative responses of detected peaks are discussed here. Due to the common occurrence of contamination during ESI analysis (Smit et al., 2015), the relative response of all peaks was calibrated by subtracting the response of peaks detected for blank filters. The double bond equivalents (DBEs) and aromaticity equivalents (Xn) were calculated as follows:

where c, h, o, n, and s are the number of C, H, O, N, and S atoms in the corresponding formulas. It should be noted that the formula with DBE < 0 or Xn < 0 has been excluded (Wang et al., 2017).

3.1 General characteristics of POCs for off-road engines

In general, the range of detected peaks for excavators and vessels had molecular weights between 150 and 900 Da, but most of the intensive peaks occurred in the molecular weight range of 200–400 Da. The mass spectra for excavators in different operational modes and vessels using different oils varied from each other. The number of peaks for POCs were 4734, 3097, 4731, 4554, and 2818 in excavator emissions in the idling, working, and moving modes and vessel emissions using HFO and diesel, respectively. The average molecular weight of excavator emissions in the working mode and vessels using HFO were the lowest (322.6±69.9 and 331.3±72.9 Da, respectively).

For excavators, CHO was the most abundant group of POCs in all three operation modes, accounting for 41 %, 46 %, and 48 % of all the formulas in terms of peak responses for the idling, working, and moving modes, respectively. S-containing compounds (CHOS and CHONS) were most abundant in the idling mode, while the relative peak response of the CHON group was highest in the working mode (Fig. 1). For vessels, CHO was the most abundant species group of POCs for both the vessel using diesel and the ones using HFO, accounting for 50 %–60 % of total peak intensity. However, CHOS accounted for almost 30 % of total ion intensity for vessels using HFO, higher than other off-road diesel engines. Furthermore, the chemical properties of POCs for vessels using HFO showed a larger degree of oxidation and unsaturation than other samples (Table S2). These differences in the composition of POCs might be attributable to variations in engine load, fuel supply, and air supply in different operation modes, which are discussed later.

As discussed in the Supplement (Sect. S3), the chemical properties of extractions derived from water or dichloromethane (DCM) or MeOH were significantly different (Supplement Figs. S1 and S2). And through comparing the optical properties between water and DCM or MeOH extractions, it was found that the average mass absorption efficiency of water extracts was significantly higher than those for 90 % DCM +10 % MeOH extracts (Fig. S3). Thus, it was necessary to extract PM_2.5 by water to explore the emission characteristics of POCs from off-road engines.

Figure 1Mass spectra of POCs in water extractions for off-road diesel engine emissions. Panels (a), (b), (c), (d), and (e) were the mass spectra for excavators in idling, working, and moving modes and vessels using diesel and HFO, respectively.

Download

3.2 CHO compounds in POCs from off-road engine emissions

The number of peaks for CHO compounds were 1746, 1287, 1797, 1561, and 1318 for excavators in the idling, working, and moving modes and vessels using HFO and diesel, respectively. Considering the number of detected peaks for CHO compounds, the compositions of the CHO group emitted from off-road engines were more complicated than those from ambient samples while being relatively comparable to those from other sources of emission (e.g., biomass: 1514–2296; coal combustion: 918) (Lin et al., 2012a; Jiang et al., 2016; Song et al., 2018). The average molecular weight of detected ions for CHO compounds for excavators idling, working, and moving and vessels using HFO and diesel were 338±96.7, 316±84.6, 336±96.6, 331±72.9, and 357±123 Da, respectively, which were significantly higher than those emitted by coal and biomass burning ($m/z=\mathrm{227}$–337 Da) (Song et al., 2018). Excavators in working mode had higher engine loads and combustion temperatures than those in other operation modes. Thus, the lowest number of CHO group ions and the smallest average molecular weight were found during the working mode compared to the idling and moving modes. This indicated that long-chain aliphatic hydrocarbons were liable to crack under elevated temperatures and low air ∕ fuel ratio conditions. This trend was consistent with the results of a previous study (Wang et al., 2018), which found that low temperatures and oxygen-rich combustion would promote a chain propagation reaction. Although the lipid contents of fossil fuel might be an important precursor of CHO compounds from off-road engines, different fractions of heteroatom and isomers could lead to significantly different structures for the CHO group (Hellier et al., 2017). The highest intensity peaks for CHO compounds for off-road diesel engines were C₉H₅O₆ and C₁₃H₉O₄, while for vessels using HFO, the highest peaks were C₁₀H₅O₈, C₉H₅O₆, and C₁₀H₅O₉.

The Van Krevelen (VK) diagram (H∕C versus O∕C) was generally used to identify the structural properties of organic matter in FT-ICR MS research as only the molecular formula was given (Wozniak et al., 2008; Lin et al., 2012a). Upon comparing the ratios of H∕C and O∕C for CHO compounds for different off-road engines in three operational modes and using different fuel oils, we concluded that the CHO group for vessels using HFO had the highest degree of oxidation and unsaturation. Furthermore, the CHO group in the working mode had a higher degree of oxidation and unsaturation than in idling and moving modes. As shown in Fig. 2, region 1 possibly represented monocarboxylic acid, which was more abundant in both idling and moving modes than in the working mode (Wozniak et al., 2008; Lin et al., 2012a). Region 2 represented compounds with low ratios of H∕C and O∕C and DBE > 10 which were commonly regarded condensed hydrocarbons (Wozniak et al., 2008; Lin et al., 2012a). Most compounds detected in the CHO group for vessels using HFO were molecular species in region 2 with a high number of O atoms and a low ratio of H∕C. This was consistent with the original structure of combustion HFO, which was defined as bottom residue oil, containing fewer aliphatic hydrocarbons than those for diesel (Wikipedia, 2018). Furthermore, low engine speeds for vessels using HFO caused low-temperature combustion, which was prone to the addition of O₂ to alkyl radicals and the subsequent formation of six-member ring isomers (Sarathy et al., 2011; Ranzi et al., 2015). The ratios of DBE∕C can be used as an indicator of condensed aromatic ring structures (Hockaday et al., 2006; Lin et al., 2012b; Yassine et al., 2014). When DBE∕C was higher than 0.7, compounds were commonly identified as soot materials or oxidized polycyclic aromatic hydrocarbons (PAHs), an important class of light-absorption organic materials. The relative response of compounds with DBE/C > 0.7 accounted for 3.2 %, 6.5 %, 3.1 %, 26 %, and 8.3 % of total ions for excavators in the idling, working, and moving modes and vessels using HFO and diesel, respectively.

Figure 2The Van Krevelen (VK) diagrams of CHO compounds for off-road engines. Panels (a), (b), (c), (d), and (e) were the mass spectra for excavators in idling, working, and moving modes and vessels using diesel and HFO, respectively.

Download

3.3 CHON compounds in POCs from off-road engine emissions

The peaks intensity percentage for the CHON group to total ions was the second largest in POCs emitted from off-road diesel engines, except for vessels using HFO (Fig. 1). The fraction of nitrogen oxide declined with increasing length of the straight-chain alkyl (Hellier et al., 2017), which was consistent with the relative response of the CHON group for diesel and HFO-fueled engine emissions. It was always considered that CHON mainly originated from biomass burning emission (18 %–41 %), while the percentage of peak responses for the CHON group to the total assigned ions measured from off-road diesel engines was comparable or slightly smaller than those emitted from biomass burning (Laskin et al., 2009; Wang et al., 2017; Song et al., 2018). As shown by the average ratios of O∕C and H∕C for the CHON group for vessels and excavators, the degree of oxidation and unsaturation of the CHON group for vessels was higher than those for excavators (Table S2).

C₁₀H₅O₅N₂, C₁₁H₇O₅N₂, and C₁₂H₉O₅N₂ were the most abundant peaks in terms of relative responses for the CHON group detected in diesel-fueled excavators and vessel emissions, while C₁₀H₄NO₆, C₉H₄NO₄, and C₁₀H₄NO₇ were highest for vessels using HFO. Diesel-fueled off-road equipment and vessels using HFO were high-rated speed and low-rated speed engines, respectively (Table 1). This might be the main reason for the formation of two nitrogen atoms in the CHON group for high-rated speed equipment, which results in higher combustion temperatures than those for low-rated speed engines, thereby promoting nitrogen atom attachment. The higher oxygen content in vessel emissions using HFO might be attributable to the higher oxygen content and kinematical viscosity of the HFO (Table 2). The large value of kinematical viscosity in HFO was a result of the presence of a certain number of aromatic hydrocarbons, fatty acids, etc., which were difficult to combust and resulted in incomplete combustion (Örs et al., 2018).

For further discussion of possible chemical structures, the CHON group was divided into 23 subgroups, including OxN1 ($\mathrm{1}\le x\le \mathrm{10}$) and OyN2 ($\mathrm{2}\le y\le \mathrm{14}$) (Fig. S4). The distribution patterns of CHON subgroups for excavators in three operational modes were similar, with the highest relative response of N₂O₅, while NO₄–NO₅ were the highest group for vessel emissions in terms of relative response. The ratio of O∕N higher than 3 is always indicative of the presence of nitro compounds (-NO₂) or some organic nitrates (with NO₃). Yassine et al. (2014) reported that DBE∕C was only valuable to the aromaticity properties of compounds with pure hydrocarbons.

The aromaticity equivalent (Xn) has been proposed to evaluate the aromaticity of organic material with heteroatoms (e.g., N, S). When the value of Xn exceeds 2.5, aromatic structures are most likely to be present within the compounds, while a value of Xn higher than 2.7, indicates the presence of condensed aromatic compounds (e.g., benzene core structure with Xn =2.5; pyrene core structure with Xn=2.83; ovalene core structure with Xn=2.92). As mentioned, the formulas in the CHON group with the most abundant relative responses for diesel-fueled excavators and vessels were C₁₀H₅N₂O₅, C₁₁H₇N₂O₅, and C₁₂H₉N₂O₅, which compose the largest green ball in Fig. 3 with Xn =2.5; it most likely indicated the presence of a benzene core structure in the compounds. Thus, it was could be dinitrophenol and methyl dinitrophenol compounds. Likewise, C₁₀H₄NO₆, C₉H₄NO₄, and C₁₀H₄NO₇ comprise the largest yellow ball in Fig. 3 for HFO-fueled vessels, most of which have Xn > 2.7, indicating the presence of condensed aromatic compounds. Nitronaphthol and methyl nitronaphthol were potentially the most significant compounds arising from HFO-fueled vessel emissions, which have previously been reported in vehicle emissions (Yassine et al., 2014; Tong et al., 2016). Furthermore, almost 55 % of the CHON group had an O∕N ratio higher than 5, and half of the CHON group had aromatic rings higher than 3 for HFO-fueled vessels. This was significantly higher than those for excavators and vessels using diesel. The reactivity and life span of these compounds should be considered to determine whether these chemicals could be used as tracers for off-road engine combustion. The same chemical structure discovered in the atmosphere indicated that secondary organic aerosol (SOA) could provide valid evidence that CHON with nitrophenol or nitronaphthol could exist in the atmosphere for long enough to be detected (Zhang et al., 2010).

Figure 3Molecular composition and possible structure of CHON for excavators in three modes and vessels using HFO and diesel.

Download

3.4 S-containing compounds in POCs from off-road engine emissions

The last group of POCs was S-containing organic compounds, including CHOS and CHONS. As shown in Fig. 1 and Table S2, the percentage of peak responses for S-containing species to total assigned peaks from vessels using HFO (35 %) was higher than those from other vehicles, with 1, 3, 2, and 3 times more than those for excavators idling, working, and moving and vessels using diesel, respectively. However, the CHONS group for excavators was significantly higher than those for vessel emissions in terms of relative ion intensity. The high fraction of peak responses for S-containing species from vessels using HFO might be attributed to the high sulfur content in HFO. The maximum sulfur content in HFO detected in this study was 2.46 %, which was significantly higher than those in diesel (Table 2). In addition, for excavators in the idling mode, the fraction of the relative response of S-containing compounds was 32.5 %, while for the working and moving modes, they were 11.6 % and 17.1 %, respectively.

To facilitate further discussion, three subgroups for CHONS (N₁O₅S₁, N₁O₆S₁, and N₁O₇S₁) and eight subgroups for CHOS (O₃S₁, O₄S₁, O₅S₁, O₆S₁, O₆S₂, O₇S₁, O₇S₂, and other) were characterized (Fig. 4). Generally, O₄S₁ and O₅S₁ were the most abundant subgroups for all off-road engines. For example, N₁O₅S₁ was the most abundant subgroup for the working mode, accounting for 36.7 % of S-containing compounds. When O∕S ≥4, this indicates that a sulfate group exists within the organic compounds. Thus, O₅S₁ and O₄S₁ may be organosulfates or sulfonates (Riva et al., 2015). Riva et al. (2015) found that sulfur-containing products from PAHs were possible and might not be solely sulfates but also sulfonates, especially with O∕S values of 4–5. Upon comparing the sulfur-containing products observed in this study and the study by Riva et al. (2015), it was interesting to conclude that some PAH-derived OS products generated in the lab also have a significant response in field measurements. As shown as Fig. S5, three of the most abundant peaks (C₈H₇O₅S⁻, C₁₁H₅O₆S⁻, and C₁₈H₂₉O₄S⁻) of S-containing compounds emitted from HFO-fueled vessels were also observed in the lab from PAH oxidations in the presence of sulfate. Although the fraction of S-containing compounds for vessels using HFO was similar to those for excavators idling (Fig. 1), different structures of compounds might exist between these two engines. The relative response of O₆S₂ and O₇S₂ for excavators idling was considerably higher than those for vessels using HFO.

S-containing compounds for vessels were highly unsaturated with 8.03 for the average DBE value which was higher than those for excavators (6.77; Table S2). Furthermore, the fraction of compounds with Xn ≥2.5 accounted for 9.3 %, 3.7 % , 2.5 %, 1.5 %, and 3.4 % of the total S-containing compounds for vessels using HFO, diesel, and excavators in idling, working, and moving modes, respectively. Through a comparison of the average DBE value and fraction of compounds with Xn ≥2.5 between excavators and vessels, it was proposed that different structures were present in S-containing compounds. The most abundant S-containing compounds emitted by off-road diesel engines were possibly aliphatic with long chains and a sulfate fraction, which was consistent with the results of Tao et al. (2014), who found that most of the CHOS group contained long aliphatic carbon chains and low degrees of unsaturation and oxidation in ambient air in Shanghai. They suggested that most of these compounds were derived from diesel emission. It was interesting to infer that the most abundant peaks of CHOS compounds observed in this study were also identified through the laboratory simulation study (Riva et al., 2016). The conclusions reported from Riva's study could provide a possible chemical reaction path to explain the chemical formula detected from off-road engine combustion. The formulas marked in bold red in Table S3 were the homologous compounds with C₁₂H₂₃O₅S⁻, which were reported to have been generated from dodecane oxidation by Riva et al.'s (2016) research, while the formulas in bold blue were likely formed from cycloalkanes. In contrast, the structures of S-containing compounds emitted from HFO-fueled engines were liable to have condensed aromatic rings.

Figure 4The distribution of subgroups of S-containing compounds for off-road engines.

Download

On an average, 88.5 % ±9.1 % and 98.1 % ±0.22 % of CHOS compounds for vessels and excavators, respectively, were with O∕S ≥4, which might indicate that organic sulfates or sulfonates were the most impactful S-containing compounds emitted from off-road engines. For most CHOS compounds containing O∕S ≥4, O*(O*=O-3) was used to identify the oxidation of organic compounds by removing the effect of sulfate or sulfonates. The modified VK diagram (H∕C and ${\mathrm{O}}^{*}/\mathrm{C}$) was used to characterize the structure of CHOS compounds (Fig. 5). Some organosulfates or sulfonates with an aromatic ring (subgroup A) could be produced by SOA precursors (e.g., α-pinene, β-pinene, monoterpenes) (Surratt et al., 2008). The most important precursors generating the subgroup A in this study might be the PAH-derived OS products in the presence of sulfate (Riva et al., 2015). However, there was another subgroup of organosulfates (subgroup B) emitted by diesel vehicles with long aliphatic carbon chains (Tao et al., 2014). Subgroup B of organosulfates was defined as C > 8, DBE < 3, and 3 < O < 7 (red circles in Fig. 5) (Tao et al., 2014), while the remaining CHOS compounds could be regarded as subgroup A (blue circles represent subgroup A with Xn ≥2.5, and the rest is contained within the black circle). The fractions of subgroup B of CHOS compounds in off-road diesel engine emissions (average: 33.9 % ±6.64 %) were significantly higher than those in HFO-fueled vessel emissions (19.9 %). For vessels using HFO, almost 10 % of CHOS compounds (blue in Fig. 5) were possibly organosulfates with one or more aromatic rings, which was consistent with the quality of the HFO. The structures detected for CHON compounds for HFO-fueled vessels and the organosulfates with one or more aromatic rings were similarly detectable in the atmosphere (Surratt et al., 2008).

Figure 5The ratios of $\mathrm{O}*/\mathrm{C}$ vs. H∕C of CHOS and proportion of the subgroup B and Xn > 2.5 for off-road engines. Panels (a), (b), (c), (d), and (e) are the mass spectra for excavators in the idling, working, and moving modes and vessels using diesel and HFO, respectively.

Download

3.5 Comparison of POCs in fresh smoke particles emitted from different sources

CHO compounds were the most abundant species across all sources (biomass, coal, on-road vehicles, and off-road vehicles) in terms of ion intensity, while the fractions of CHON and S-containing compounds were different from anthropogenic source emissions. Furthermore, the possible chemical structures of these compounds for diverse sources varied sharply.

For CHO compounds, the average DBE values from excavator and vessel emissions were 8.38±3.84 and 8.55±3.77, which was higher than those emitted from crop burning (6.9) and coal combustion (7.48) (Song et al., 2018). The methoxyphenols (C₉H₁₄O₄, C₁₈H₂₈O₈, C₁₁H₂₀O₅, C₁₈H₂₂O₇, C₉H₁₂O₄) and fatty acids (C₁₆H₃₂O₂), derived from limonene and palmitic acid, were frequently observed in crop burning emissions. CHO compounds with high C numbers (C₂₀H₂₈O₂, C₂₀H₂₆O₂, C₂₀H₃₀O₂) were detected from wood burning. One benzene ring substituted with O-containing groups such as hydroxyl, methoxyl, and carboxyl (C₈H₆O₄, C₁₃H₁₈O₄, and C₁₂H₁₆O₄) was dominant in coal combustion. Overall, mono-aromatics dominated tunnel samples (Tong et al., 2016) and off-road diesel vessels. However, abundant condensed aromatic ring structures with high O numbers (C₁₀H₅O₈, C₉H₅O₆ and C₁₀H₅O₉) were proposed for HFO-fueled vessels.

For CHON compounds, almost all sources were reported to emit nitrophenol compounds, while the substituted groups were slightly different due to different numbers of N and O atoms. The fraction of relative peak response of CHON compounds, an important light-absorbing substance, could reach half of the POCs from biomass burning emission. Methyl-nitrocatechols produced from the oxidation of cresol and N-bases composed of C, H, and N elements were regarded as the biomarkers for biomass burning (Laskin et al., 2009; Wang et al., 2017). However, on comparing the signal intensity of nitroaromatics in ambient aerosol and fresh biomass burning smoke, Wang et al. (2017) found evidence to the contrary. Signal intensity was stronger in ambient aerosols than that in fresh biomass burning smoke, which indicated the existence of other sources or aging processes. Recently, nitrophenol was also detected in tunnel samples indicating traffic sources. In this study, we inferred that dinitrophenol were abundant in off-road diesel vehicle emissions, while nitronaphthol with one or more methyl groups was dominant for HFO-fueled vessel emissions.

Except for biomass burning, S-containing compounds were still an important group of organic matter for coal combustion, on-road vehicles, off-road diesel vehicles, and HFO-fueled vessels and at background sites, accounting for 48 %, 17 %, 8.9 %, 33 %, and 32 %, respectively, of total detected organic matters. Organosulfates and sulfonates were one of the most important HULIS, which were reported as the prominent S-containing compounds at background sites due to aging reactions of organics with H₂SO₄. For coal combustion, S-containing compounds had low DBE and AI_mod,w values, which were probably regarded as alkylbenzene rings substituted with one sulfate group. The specific structure of S-containing compounds was organosulfates or sulfonates with condensed aromatic rings for HFO-fueled vessels, while more abundant organosulfates with aliphatic chains were observed in emissions from off-road diesel equipment. Likewise, the organosulfates with aliphatic long-chain alkanes were observed in on-road traffic emissions due to their original oil structure (Jiang et al., 2016; Riva et al., 2016; Tao et al., 2014). These S-containing compounds with high aromaticity, or long-chain alkanes, were frequently formed from secondary photochemical reactions between oxidation products of volatile and intermediate-volatility organic compounds and acidified sulfate particles (Riva et al., 2015, 2016). Therefore, a high-abundance of S-containing compounds in the atmosphere might come from secondary photochemical reactions and also be emitted directly from the combustion of off-road engines.