Long-Term Observation of Mixing States and Sources of Vanadium-Containing Single Particles from 2020 to 2021 in Guangzhou, China

The distribution of vanadium (V) in aerosols is commonly used to track ship exhaust emissions, yet the atmospheric abundance of V has been greatly reduced due to the implementation of a clean fuel policy. Recent research mainly discussed the chemical compositions of ship-related particles during specific events, yet few studies focus on the long-term changes of V in the atmosphere. In this study, a single-particle aerosol mass spectrometer was used to measure V-containing particles from 2020 to 2021 in Huangpu Port in Guangzhou, China. The long-term trend of the particle counts of V-containing particles declined annually, but the relative abundance of V-containing particles in the total single particles increased in summer due to the influence of ship emissions. Positive matrix factorization revealed that in June and July 2020, 35.7% of the V-containing particles were from ship emissions, followed by dust and industrial emissions. Furthermore, more than 80% of the V-containing particles were found mixing with sulfate and 60% of the V-containing particles were found mixing with nitrate, suggesting that the majority of the V-containing particles were secondary particles processed during the transport of ship emissions to urban areas. Compared with the small changes in the relative abundance of sulfate in the V-containing particles, the relative abundance of nitrate exhibited clear seasonal variations, with a high abundance in winter. This may have been due to the increased production of nitrate from high concentrations of precursors and a suitable chemical environment. For the first time, the long-term trends of V-containing particles in two years are investigated to demonstrate changes in their mixing states and sources after the clean fuel policy, and to suggest the cautious application of V as an indicator of ship emissions.

In recent years, the distribution and characteristics of V in ship-emitted particles have attracted a great deal of attention. High loadings of V can be emitted from the combustion of heavy fuel oil in ship engines, and thus, the presence of V has generally been regarded as an effective tracer of ship emissions [20][21][22][23][24][25][26]. Since China designated three domestic emission control zones (DECAs) in 2015, i.e., the Yangtze River Delta (YRD), Pearl River Delta (PRD), and Bohai Rim Area (BRA), the quality of ship fuel oil has been forced to improve to meet the demand for cleaner ship emissions. All vessels inside the DECAs were required to use fuel oil with a sulfur content of less than 0.5% after 1 January 2020 [20], which also helped to reduce the content of V in ship fuel oil. After the implementation of this clean fuel policy, the V content of emissions decreased significantly. A recent study conducted in Pudong [16] revealed that the average concentration of V was 1.19 ng/m 3 in 2020, which was 74% lower than that in 2019 (4.60 ng/m 3 ).
In addition to the offline determination of the mass concentration of V in particles, the distribution and mixing states of V in ship emission-related single particles have also been studied. Single-particle aerosol mass spectrometry (SPAMS) has been widely used to identify the temporal and spatial distributions of V in single particles due to the precise ionization and detection of V of this method [27][28][29][30]. The presence and mixing states of V in coastal areas are mainly affected by ship emissions, industrial processes, and regional transportation [28,[31][32][33][34][35]. According to previous studies using SPAMS, shipemitted particles are dominated by elemental (EC) and organic carbon, V, sulfate, and nitrate, which have stronger signals than ambient particulates [22,28,30,[36][37][38][39]. Another study of single particles in ship plumes in Shanghai Port found that the freshly emitted ship particles were dominated by sulfate, EC, and V, but they exhibited a very low nitrate signal or no nitrate signal [28]. Sulfate is known to be formed rapidly in ship stacks from the gas-particle conversion of SO 2 , and Xiao et al. (2018) found that sulfate particles in the ambient air always remain synchronous with ship-source particles, which confirmed the relatively stable contribution of ship-source sulfate particles to the ambient sulfate particle concentration [30,39]. In addition, the abundance of sulfuric acid in the V-containing particles is higher than that in particles that do not contain V, suggesting that V may act as a catalyst during the oxidation of SO 2 to sulfate in these particles [28,36,40]. After the implementation of the clean fuel policy, field observations in Shanghai and Guangzhou showed that V was still prevalent in particles from ship plumes, but the V content was significantly reduced [16,28,29,41]. These studies found that V-containing particles were still closely related to ship emissions; however, after the implementation of the clean fuel policy and the outbreak of the COVID-19 epidemic, the distribution and sources of the V-containing particles may have changed. Hence, a comprehensive investigation of these particles is necessary to understand their sources and aging processes.
To investigate the long-term trend of the mixing state of the V-containing particles, V-containing single particles were continuously determined via SPAMS in 2020 and 2021 in Huangpu Port, Guangzhou. The chemical compositions and mixing states of these particles in the different seasons were compared. In the first part of this paper, the necessity to conduct the long-term study of V-containing single particles and the novelty of this work were described; in the second part, the single particle collection and data processing were presented, as well as the sources analysis via positive matrix factorization (PMF) and potential source contribution function (PSCF); in the third part, the annual variation and mixing states of V particles were discussed, and their sources and aging processes were investigated. In addition, the long-term trends of the mixing states of V with sulfate and nitrate were investigated to evaluate the changes in ship emission sources and the aging state of the V-containing particles. The main conclusions, limitations of this work, and future directions were summarized in part four. The results of this study provide insights into the impact of the clean fuel policy on the distribution of V and its role as a tracer for ship emissions.

Sampling Information and SPAMS Measurements
Ambient single particles were measured via SPAMS in Huangpu Port, Guangzhou, China, from 1 January 2020 to 31 December 2021. The sampling site was located on the northern coast of the Pearl River. Huangpu Port is the largest coastal and ocean transportation hub in southern China. The hourly concentrations of PM 2.5 , PM 10 , and gaseous pollutants, as well as meteorological data, were simultaneously collected. The sampling methods and data analysis used in this work are shown in Figure 1. investigated. In addition, the long-term trends of the mixing states of V with sulfate and nitrate were investigated to evaluate the changes in ship emission sources and the aging state of the V-containing particles. The main conclusions, limitations of this work, and future directions were summarized in part four. The results of this study provide insights into the impact of the clean fuel policy on the distribution of V and its role as a tracer for ship emissions.

Sampling Information and SPAMS Measurements
Ambient single particles were measured via SPAMS in Huangpu Port, Guangzhou, China, from 1 January 2020 to 31 December 2021. The sampling site was located on the northern coast of the Pearl River. Huangpu Port is the largest coastal and ocean transportation hub in southern China. The hourly concentrations of PM2.5, PM10, and gaseous pollutants, as well as meteorological data, were simultaneously collected. The sampling methods and data analysis used in this work are shown in Figure 1. The SPAMS was designed by the Guangzhou Hexin Analytical Instruments Limited Company based on the same principle as the aerosol time-of-flight mass spectrometer (ATOFMS) developed by Prather and co-workers [42][43][44]. The particle detection method of SPAMS has been described in detail by Li et al. (2011) [45]. Briefly, aerosol particles are introduced into the SPAMS through the orifice and are accelerated through an aerodynamic lens into the sizing region of the instrument, and are identified based on how fast they pass through two continuous diodes (Nd:YAG) and a 532 nm laser beam in the sizing region. The sized particles are then ionized by a 266 nm Nd:YAG laser (0.6 mJ), which is triggered at the precise time determined by the particle velocity. The positive and negative ions were analyzed using a Z-bipolar time-of-flight mass spectrometer [45][46][47].

Data Analysis
The chemical compositions and mixing states of the single particles measured via SPAMS were analyzed using the Computational Continuation Core (COCO) toolkit in MATLAB. In this study, a total of 37,867,836 particles were detected in two years. The Vcontaining particles were identified by the marker ions of 51 V + and 67 VO + with a relative peak area (RPA) threshold greater than 0.1% of the total signal in the mass spectrum [29,35,40]. According to this principle, the particle count of the V-containing particles was 373,805, which accounted for approximately 1% of the total particles in two years. In this paper, the definition of the mixing state based on SPAMS refers to the coexistence of two or more chemical components in the same particle, which can be used to track the origin and chemical evolution processes of specific components [48][49][50][51].
To implicitly track the chemical species that co-varied with V in the single particles, PMF was conducted to resolve the different types of V-containing particles. Similar The SPAMS was designed by the Guangzhou Hexin Analytical Instruments Limited Company based on the same principle as the aerosol time-of-flight mass spectrometer (ATOFMS) developed by Prather and co-workers [42][43][44]. The particle detection method of SPAMS has been described in detail by Li et al. (2011) [45]. Briefly, aerosol particles are introduced into the SPAMS through the orifice and are accelerated through an aerodynamic lens into the sizing region of the instrument, and are identified based on how fast they pass through two continuous diodes (Nd:YAG) and a 532 nm laser beam in the sizing region. The sized particles are then ionized by a 266 nm Nd:YAG laser (0.6 mJ), which is triggered at the precise time determined by the particle velocity. The positive and negative ions were analyzed using a Z-bipolar time-of-flight mass spectrometer [45][46][47].

Data Analysis
The chemical compositions and mixing states of the single particles measured via SPAMS were analyzed using the Computational Continuation Core (COCO) toolkit in MATLAB. In this study, a total of 37,867,836 particles were detected in two years. The V-containing particles were identified by the marker ions of 51 V + and 67 VO + with a relative peak area (RPA) threshold greater than 0.1% of the total signal in the mass spectrum [29,35,40]. According to this principle, the particle count of the V-containing particles was 373,805, which accounted for approximately 1% of the total particles in two years. In this paper, the definition of the mixing state based on SPAMS refers to the coexistence of two or more chemical components in the same particle, which can be used to track the origin and chemical evolution processes of specific components [48][49][50][51].
To implicitly track the chemical species that co-varied with V in the single particles, PMF was conducted to resolve the different types of V-containing particles. Similar analytical methods have been used in studies of the evolution of biomass burning emissions using the single-particle mass spectrometry technique [52,53]. Twenty-two marker ions were selected to conduct the PMF analysis, comprising elemental carbon ( 24 35 Cl − . These components were abundant in the V-containing particles and were usually associated with specific sources and/or secondary processes. According to previous studies that used PMF to analyze single particles, the uncertainty was defined as 50% of the relative peak area of the marker ion [52,53], and two to seven factors were generated through PMF analysis of the V-containing particles. Based on their strong factor information and the moderate correlation of the RPAs under measurement and prediction, seven factors were selected as the best solution for the determination of the V-containing particles.

Potential Source Contribution Function Analysis
The PSCF analysis is an effective method for identifying source-contaminated areas in recipient sites [54,55]. In this study, the PSCF was used to calculate the 72 h backward trajectory 500 m above the sampling site [56,57]. Briefly, the ij th component of a PSCF field can be expressed as follows: where n ij is the total number of endpoints that fall in the ij th cell (divided into 0.5 • × 0.5 • latitude and longitude grids), and m ij is the number of endpoints of that parcel for which the measured values exceed a user-determined threshold criterion. Since the deviation of the PSCF results will increase with increasing cell-value receptor distance, to reduce the influence of the uncertainty of cells with small m ij values, the value of the PSCF is multiplied by the weighting factor W ij (hereinafter referred to as the WPSCF) [57,58]. The source distribution characteristics of V particles were similar between the three months of each season, and we have listed the PSCF results of March, April, and May in the Supplementary Materials ( Figure S1). Thus, we have chosen the results of January, April, July, and October to represent the annual results of the PSCF analysis of the V particles.

Characteristics of V-Containing Single Particles
The monthly variation trends of the V-containing particles based on the average of the hourly data are shown in Figure 2. The particle count of the V-containing particles accounted for 1.4% and 0.6% of the total detected single particles in 2020 and 2021, respectively. The particle counts of the V-containing particles exhibited a decreasing trend from the beginning of 2020 to 2021, which may have been due to two reasons. First, the quality of the ship fuel oil was improved according to the clean fuel policy, so the fuel oil contained a smaller amount of V. This led to a reduction in the V emitted from the ship exhaust. The second possible reason is the influence of the COVID-19 epidemic since January 2020. Many field studies have evaluated the impact of the epidemic on anthropogenic activities [59,60], and they reported a sharp reduction in primary anthropogenic emissions. Although the number fraction (NF) of the V-containing particles (NFv) exhibited a general decreasing trend from 2020 to 2021, the NFv exhibited peaks in May, June, and July. Furthermore, the NFv accounted for 2.1% of the total particles in July 2020 and 1.0% in May 2021. This phenomenon was primarily due to the summer monsoon transporting the air pollutants emitted by ships to the sampling site, which was observed in our previous study [35]. Other studies have also reported the predominant contribution of V from ship exhaust in the sampling site. For example, the concentration of V was higher in the southeastern coastal area and the Qingdao area during the summer [54,61]. In Xiamen, the monthly pattern of V concentration increased from April to September and then decreased from October to January [17]. monthly pattern of V concentration increased from April to September and then decrea from October to January [17]. The monthly variations in SO2, O3, NO2, PM10, CO, and PM2.5 concentrations are sented in Figure 3. All of these compounds exhibited trends different from that of th containing particles. The PM2.5 and PM10 concentrations were both high in winter and in summer, which was mainly due to the different loadings of particulate matter (P from local anthropogenic emissions and transportation by the northerly wind in the ferent seasons. Figure S2 shows the temporal variations of meteorological parame (wind speed, wind direction, temperature, and relative humidity), and the north w prevailed in the winter, which brought large amounts of anthropogenic pollutant Guangzhou. The CO, SO2, and NO2 concentrations were all high in the winter seasons low in the summer seasons of 2020 and 2021, and they were mainly influenced by l emissions, including vehicle exhaust and coal burning. Although CO, SO2, and NO2 all be produced from ship exhaust, the total amount of their emissions was far less t that of fossil fuel and biomass burning sources in urban cities [24,62]. Thus, the gen trends of the CO, SO2, and NO2 concentrations were different from that of the V-cont ing particles from 2020 to 2021. As a secondary pollutant, O3 is produced from photoch ical reactions of NOx and volatile organic compounds (VOCs) [63,64]. The O3 concen tion peaks occurred in April, August, September, October, and November, which completely different from the trend of the PM and primary gas pollutants. The monthly variations in SO 2 , O 3 , NO 2 , PM 10 , CO, and PM 2.5 concentrations are presented in Figure 3. All of these compounds exhibited trends different from that of the Vcontaining particles. The PM 2.5 and PM 10 concentrations were both high in winter and low in summer, which was mainly due to the different loadings of particulate matter (PM) from local anthropogenic emissions and transportation by the northerly wind in the different seasons. Figure S2 shows the temporal variations of meteorological parameters (wind speed, wind direction, temperature, and relative humidity), and the north wind prevailed in the winter, which brought large amounts of anthropogenic pollutants to Guangzhou. The CO, SO 2 , and NO 2 concentrations were all high in the winter seasons and low in the summer seasons of 2020 and 2021, and they were mainly influenced by local emissions, including vehicle exhaust and coal burning. Although CO, SO 2 , and NO 2 can all be produced from ship exhaust, the total amount of their emissions was far less than that of fossil fuel and biomass burning sources in urban cities [24,62]. Thus, the general trends of the CO, SO 2 , and NO 2 concentrations were different from that of the V-containing particles from 2020 to 2021. As a secondary pollutant, O 3 is produced from photochemical reactions of NOx and volatile organic compounds (VOCs) [63,64]. The O 3 concentration peaks occurred in April, August, September, October, and November, which was completely different from the trend of the PM and primary gas pollutants.
The average mass spectra of the V-containing particles in the summers of 2020 and 2021 were similar, so we selected the spectra of the V-containing particles in June and July 2020 for a detailed description (Figure 4). The positive mass spectrum was characterized by carbon clusters such as 24 C + , 36   The average mass spectra of the V-containing particles in the summers of 2020 and 2021 were similar, so we selected the spectra of the V-containing particles in June and July 2020 for a detailed description ( Figure 4). The positive mass spectrum was characterized by carbon clusters such as 24 C + , 36 C + , 27 C2H3 + , 37 C3H7 + , 41 C3H3 + , and 43 C3H7 + , and metal ions such as 23 Na + , 51 V + , 52 Cr + , 55 Mn + , 58 Ni + , and 63 Cu + . Meanwhile, the negative spectrum mainly contained sulfate ( 97 HSO4 − ), nitrate ( 62 NO3 − ), and oxygenated organics ( 59 CH3CO2 − , 71 C3H3O2 − , and 73 C2HO3 − ). The mixing states of V with metals and secondary species suggest that these single particles were secondary particles.

Sources and Formation Processes of the V-Containing Particles
In this study, the PSCF analysis was conducted using the changes in the hourly particle counts of the V-containing particles. We chose the data from 2020 to conduct the investigation since the patterns of the 2020 and 2021 data for all four seasons were similar.

Sources and Formation Processes of the V-Containing Particles
In this study, the PSCF analysis was conducted using the changes in the hourly particle counts of the V-containing particles. We chose the data from 2020 to conduct the investigation since the patterns of the 2020 and 2021 data for all four seasons were similar. The PSCF results ( Figure 5) clearly show that the V-containing particles were associated with the transportation of different air masses in the four seasons. Regional and long-range transport both contributed significantly to the distribution of the V-containing particles in Guangzhou in winter, spring, and summer, which was attributed to the influence of ship emissions and oil refineries [14,54]. However, the transmission routes of the V-containing particles to the sampling site were different in these three seasons. In summer, the air masses were solely from the South China Sea, indicating the direct contribution of ship emissions to the distribution of V in Guangzhou, whereas in spring, the air masses were also from the east and northeast coastline, except from the South China Sea. In winter, although the air masses were still from the marine areas to the east, due to the low transmission rate, the majority of the V was from the region of the sampling site. This suggests the contribution of industrial emissions of V, including oil refineries and metal factories. To further explore the sources and influencing factors of the V-containing particles, we analyzed the data from June and July 2020. The seven-factor PMF profiles are presented in Figure 6. Based on the abundant markers and significant features of each factor, we named these factors separately. Seven factors of the V-containing particles were identified: a ship emissions factor, a dust factor, an industrial emissions factor, a metal smelting factor, a secondary aerosols factor, an organic nitrogen factor, and a sea salt factor. The ship emissions factor was composed of abundant 24 C + , 36 C + , 51 V + , and 97 HSO4 − , which were significant in the particles emitted by ships [23,29,37,40]. Additionally, moderate amounts of 46 NO3 − occurred in this factor, suggesting the aging of the V-containing particles during transport [36,38,40]. In the dust factor, 40 Ca + was the most abundant, indicating the contribution of suspended dust [8]. The industrial emissions factor and the metal smelting factor were both composed of nitrate, indicating that the particles associated with these sources 2020 V 90ºE 95ºE 100ºE 105ºE 110ºE 115ºE 120ºE 125ºE 130ºE 135ºE 140ºE 145ºE 150ºE Figure 5. Potential source distribution of the particulate matter containing V in the four seasons in 2020. The hourly concentration of the particulate matter containing V was chosen as the standard threshold in the analysis. The colors in the legend represent the WPSCF values.
To further explore the sources and influencing factors of the V-containing particles, we analyzed the data from June and July 2020. The seven-factor PMF profiles are presented in Figure 6. Based on the abundant markers and significant features of each factor, we named these factors separately. Seven factors of the V-containing particles were identified: a ship emissions factor, a dust factor, an industrial emissions factor, a metal smelting factor, a secondary aerosols factor, an organic nitrogen factor, and a sea salt factor. The ship emissions factor was composed of abundant 24 C + , 36 C + , 51 V + , and 97 HSO 4 − , which were significant in the particles emitted by ships [23,29,37,40]. Additionally, moderate amounts of 46  particles during transport [36,38,40]. In the dust factor, 40 Ca + was the most abundant, indicating the contribution of suspended dust [8]. The industrial emissions factor and the metal smelting factor were both composed of nitrate, indicating that the particles associated with these sources were produced through atmospheric aging processes. According to previous studies, in addition to the combustion of fossil fuels, V-containing particles in Guangzhou could be from glass factories and ceramics factories [1]. In this study, the industrial emissions factor mainly consisted of 56 CaO + and 58 Ni + , suggesting the influence of factory emissions in addition to an oil combustion source [21,65]. Furthermore, Cu is a typical tracer of the steel industry. In this study, the metal smelting factor was composed of moderately abundant 65 Cu + , implying that a certain amount of the V was connected with metal manufacturing [65]. The secondary aerosols factor mainly comprised secondary ions, including 71 C 3 H 3 O 2 − and 73 C 2 HO 3 − . The organic nitrogen factor was composed of abundant 42 CNO − , 26 CN − , and 59 CH 3 CO 2 − , indicating the aged state of the organics in the V-containing particles. The sea salt factor was composed of a large abundance of 35 Cl − , representing the contribution of sea spray [8,65]. The fraction of each factor in the V-containing particles is shown in Figure 7. The V-containing particles were dominated by the ship emissions factor, which accounted for 35.7%, followed by the dust factor (19.6%) and the industrial emissions factor (17.9%). The ship emissions factor was characterized by EC ( 24 C + , 36 C + ), 51 V + , and sulfate (HSO 4 − ), suggesting that the freshly emitted ship exhaust particles were aged before arriving at the sampling site [1,2,24,66]. The characteristic ions of the dust factor mainly included 40 Ca + , 56 CaO + , and sulfate. arriving at the sampling site [1,2,24,66]. The characteristic ions of the dust factor mainly included 40 Ca + , 56 CaO + , and sulfate.   According to the above analysis, it can be concluded that the V-containing particles mainly originated from three factors (ship emissions factor, dust factor, and industrial emissions factor), and their formation and aging were primarily affected by the ship and industrial emissions factors. Although ship emissions were still a dominant source of the V-containing particles, the contribution of the ship emissions substantially decreased compared with that reported in a study conducted before the implementation of the clean fuel policy [67]. Thus, it should be cautious when using V as a tracer for ship emissions because V from industrial sources and dust sources may interfere with the judgment of the contribution of V from ship sources.

Mixing States of V with Sulfate and Nitrate
The presence of sulfate and nitrate in particles is associated with the aging process and the secondary formation of particles [27,28,40]. In recent years, sulfate and V have been considered to be special tracers for studying ship emissions before the clean fuel policy was implemented [27,28,37,68]. Nitrate is considered to be a marker of particulate aging, and fresh ship emission particles produce a very low nitrate mass spectrometry signal or no nitrate signal, while the nitrate signal in environmental particulate matter is stronger than that of fresh particles, indicating that particulate matter is prone to aging during emission and transport [38,69]. In addition, the increasing abundance of nitrate in fine particles is closely related to haze and O3 pollution events in Guangzhou [70]. Thus, the mixing states of V with sulfate and nitrate were investigated in this study to track the aging process of the V-containing particles. Before investigating the sulfate and nitrate in these particles, the general distributions of the sulfate and nitrate in total single particles (particles with and without V) were analyzed to determine the overall characteristics.
The variation trends of the NFs of sulfate and nitrate in the total particles are shown in Figure 8. The particle counts of both sulfate and nitrate were high in the winter and decreased in summer, which was due to changes in their precursors (SO2 and NOx) (Figure 3). The particle counts of sulfate were generally higher than those of nitrate in 2020, but they exhibited similar particle counts in 2021. However, their precursors (SO2 and NOx) did not exhibit similar trends. Thus, the secondary formation processes of sulfate and nitrate from SO2 and NOx may have been different in 2020 and 2021. The NF of sulfate V-containing particles 7  According to the above analysis, it can be concluded that the V-containing particles mainly originated from three factors (ship emissions factor, dust factor, and industrial emissions factor), and their formation and aging were primarily affected by the ship and industrial emissions factors. Although ship emissions were still a dominant source of the V-containing particles, the contribution of the ship emissions substantially decreased compared with that reported in a study conducted before the implementation of the clean fuel policy [67]. Thus, it should be cautious when using V as a tracer for ship emissions because V from industrial sources and dust sources may interfere with the judgment of the contribution of V from ship sources.

Mixing States of V with Sulfate and Nitrate
The presence of sulfate and nitrate in particles is associated with the aging process and the secondary formation of particles [27,28,40]. In recent years, sulfate and V have been considered to be special tracers for studying ship emissions before the clean fuel policy was implemented [27,28,37,68]. Nitrate is considered to be a marker of particulate aging, and fresh ship emission particles produce a very low nitrate mass spectrometry signal or no nitrate signal, while the nitrate signal in environmental particulate matter is stronger than that of fresh particles, indicating that particulate matter is prone to aging during emission and transport [38,69]. In addition, the increasing abundance of nitrate in fine particles is closely related to haze and O 3 pollution events in Guangzhou [70]. Thus, the mixing states of V with sulfate and nitrate were investigated in this study to track the aging process of the V-containing particles. Before investigating the sulfate and nitrate in these particles, the general distributions of the sulfate and nitrate in total single particles (particles with and without V) were analyzed to determine the overall characteristics.
The variation trends of the NFs of sulfate and nitrate in the total particles are shown in Figure 8. The particle counts of both sulfate and nitrate were high in the winter and decreased in summer, which was due to changes in their precursors (SO 2 and NOx) (Figure 3). The particle counts of sulfate were generally higher than those of nitrate in 2020, but they exhibited similar particle counts in 2021. However, their precursors (SO 2 and NOx) did not exhibit similar trends. Thus, the secondary formation processes of sulfate and nitrate from SO 2 and NOx may have been different in 2020 and 2021. The NF of sulfate during these two years was 70.3-66.1%, while the NFs of nitrate were 49.7% and 64.2% in 2020 and 2021, respectively. The variation trends of the NFs of sulfate and nitrate were opposite in 2020, but they exhibited similar trends in 2021. Given the identical variation trends of SO 2 and NOx, the opposite trends of the NFs of sulfate and nitrate may have resulted from their different formation processes. Similar results were also found in the Pearl River Delta region, Hong Kong, and Xiamen [17,18,71].
Toxics 2023, 11,339 11 of 17 during these two years was 70.3-66.1%, while the NFs of nitrate were 49.7% and 64.2% in 2020 and 2021, respectively. The variation trends of the NFs of sulfate and nitrate were opposite in 2020, but they exhibited similar trends in 2021. Given the identical variation trends of SO2 and NOx, the opposite trends of the NFs of sulfate and nitrate may have resulted from their different formation processes. Similar results were also found in the Pearl River Delta region, Hong Kong, and Xiamen [17,18,71]. The mixing states of V with sulfate and nitrate were investigated through the variation trends of their NFs in the V-containing particles ( Figure 9). The particle counts of the V-sulfate (V-S) and V-nitrate (V-N) particles both decreased from 2020 to 2021, which were different from the particle counts of sulfate and nitrate in the total particles ( Figure  8). However, the abundances of sulfate and nitrate in the V-containing particles exhibited trends different from those of their particle counts. The NF of sulfate in the V-containing particles exhibited small peaks in summer and autumn in both years. The increase in sulfate in the V-containing particles was possibly due to the same influence of ship emissions. Zhou et al. (2020) observed that the fraction of sulfate-containing particles exhibited a moderate decrease from 95% in high-sulfur fuel (SF) oil (HS) emissions to 78% in low (SF) oil (LS) emissions. In this study, the annual average proportions of the sulfate content in the V-containing particles in 2020 and 2021 were 90% and 86%, respectively, which were much higher than that in LS. This can be explained by the fact that the secondary processes in the atmosphere had a strong effect on the V-containing particles. In contrast to sulfate, the NF of the nitrate in the V-containing particles exhibited peaks in winter due to the aged state of the V-containing particles. Since nitrate was not directly emitted from the ship exhaust, the high nitrate content was due to the aging of the V-containing particles. This suggests that the freshly emitted V-containing particles were aged during their transportation from the port to the urban area. The mixing states of V with sulfate and nitrate were investigated through the variation trends of their NFs in the V-containing particles ( Figure 9). The particle counts of the V-sulfate (V-S) and V-nitrate (V-N) particles both decreased from 2020 to 2021, which were different from the particle counts of sulfate and nitrate in the total particles ( Figure 8). However, the abundances of sulfate and nitrate in the V-containing particles exhibited trends different from those of their particle counts. The NF of sulfate in the V-containing particles exhibited small peaks in summer and autumn in both years. The increase in sulfate in the V-containing particles was possibly due to the same influence of ship emissions. Zhou et al. (2020) observed that the fraction of sulfate-containing particles exhibited a moderate decrease from 95% in high-sulfur fuel (S F ) oil (HS) emissions to 78% in low (S F ) oil (LS) emissions. In this study, the annual average proportions of the sulfate content in the V-containing particles in 2020 and 2021 were 90% and 86%, respectively, which were much higher than that in LS. This can be explained by the fact that the secondary processes in the atmosphere had a strong effect on the V-containing particles. In contrast to sulfate, the NF of the nitrate in the V-containing particles exhibited peaks in winter due to the aged state of the V-containing particles. Since nitrate was not directly emitted from the ship exhaust, the high nitrate content was due to the aging of the V-containing particles. This suggests that the freshly emitted V-containing particles were aged during their transportation from the port to the urban area.

Conclusions
The annual characteristics and mixing states of V-containing single particles were investigated using SPAMS from 2020 to 2021 in Huangpu Port in Guangzhou. The particle count of the V-containing particles accounted for 1.4% and 0.6% of the total detected single particles in 2020 and 2021, respectively. The decreasing trend of the V-containing particles from 2020 to 2021 may have been due to the implementation of the clean fuel policy and the reduced anthropogenic emissions during the COVID-19 epidemic since 2020. The increase in the NFv of the V-containing particles in summer was attributed to the summer monsoon carrying the air pollutants emitted by ships to the sampling site. The positive matrix factorization analysis revealed that 35.7% of the V-containing particles were from ship emissions in June and July 2020, followed by dust and industrial emissions. Although ship emissions were still a dominant source of the V-containing particles, the contribution of ship emissions substantially decreased compared with the results of studies conducted before the implementation of the clean fuel policy. Thus, care should be taken when using V as a tracer for ship emissions because V from industrial sources and dust sources may interfere with the judgment of V from ship sources. Furthermore, more than 80% of the V-containing particles were found mixing with sulfate and 60% were found mixing with nitrate, suggesting that the majority of the V-containing particles were secondary particles processed during the transport of the ship emissions to the urban area. These results emphasize the significant changes in the distribution of the V-containing particles after the implementation of the clean fuel policy and suggest the reduced role of V as an indicator of ship emissions. It is worth noting that single-particle analysis based on SPAMS only provided semi-quantitative results, which still needs comprehensive field measurements to acquire quantitative results to fully study the annual characteristics of V in atmospheric fine particles. Besides, the detailed source analysis of freshly ship-emitted particles is necessary to explore the authentic distributions of V and other metals to evaluate their changes in the atmosphere after the implementation of the clean fuel policy.

Conclusions
The annual characteristics and mixing states of V-containing single particles were investigated using SPAMS from 2020 to 2021 in Huangpu Port in Guangzhou. The particle count of the V-containing particles accounted for 1.4% and 0.6% of the total detected single particles in 2020 and 2021, respectively. The decreasing trend of the V-containing particles from 2020 to 2021 may have been due to the implementation of the clean fuel policy and the reduced anthropogenic emissions during the COVID-19 epidemic since 2020. The increase in the NFv of the V-containing particles in summer was attributed to the summer monsoon carrying the air pollutants emitted by ships to the sampling site. The positive matrix factorization analysis revealed that 35.7% of the V-containing particles were from ship emissions in June and July 2020, followed by dust and industrial emissions. Although ship emissions were still a dominant source of the V-containing particles, the contribution of ship emissions substantially decreased compared with the results of studies conducted before the implementation of the clean fuel policy. Thus, care should be taken when using V as a tracer for ship emissions because V from industrial sources and dust sources may interfere with the judgment of V from ship sources. Furthermore, more than 80% of the V-containing particles were found mixing with sulfate and 60% were found mixing with nitrate, suggesting that the majority of the V-containing particles were secondary particles processed during the transport of the ship emissions to the urban area. These results emphasize the significant changes in the distribution of the V-containing particles after the implementation of the clean fuel policy and suggest the reduced role of V as an indicator of ship emissions. It is worth noting that single-particle analysis based on SPAMS only provided semi-quantitative results, which still needs comprehensive field measurements to acquire quantitative results to fully study the annual characteristics of V in atmospheric fine particles. Besides, the detailed source analysis of freshly ship-emitted particles is necessary to explore the authentic distributions of V and other metals to evaluate their changes in the atmosphere after the implementation of the clean fuel policy.

Supplementary Materials:
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/toxics11040339/s1, Figure S1: Potential source distributions of the V-containing single particles in March, April and May, 2020. Figure S2: Temporal variations of wind speed (WS), wind direction (WD), relative humidity (RH), temperature (T), NO 2 concentration, SO 2 concentration, PM 2.5 concentration, PM 10 concentration, total ambient particles, V-containing particles and its number fraction (NFv) in total ambient particles from 2020 to 2021 in Guangzhou, China.
Author Contributions: Conceptualization, C.C. and L.M.; methodology, C.C. and L.M.; sampling and data analysis, X.X., Z.W., L.Y., S.L. and M.L.; writing-original draft preparation, X.X. and C.C.; writing-review and editing, X.X., Z.W., C.C., L.M. and Z.Z. All authors have read and agreed to the published version of the manuscript. Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: The observational data, including SPAMS and the meteorological parameters obtained in this study, are available from the corresponding authors upon request (limei2007@163.com).

Conflicts of Interest:
The authors declare no conflict of interest.