Seasonal Control of Water-Soluble Inorganic Ions in PM_2.5 from Nanning, a Subtropical Monsoon Climate City in Southwestern China

Abstract

In this study, we measured the daily water-soluble inorganic ions (WSIIs) concentration (including SO₄²⁻, NO₃⁻, NH₄⁺, Ca²⁺, K⁺, Cl⁻, Na⁺, Mg²⁺, and F⁻) of PM_2.5 (particulate matter with a diameter smaller than 2.5 μm) throughout the year in Nanning (a typical subtropical monsoon climate city in southwestern China) to explore the influence of seasonal climate change on the properties of PM_2.5 pollution. This suggested that SO₄²⁻, NO₃⁻, and NH₄⁺ were the main component of WSIIs in Nanning. Secondary inorganic ions from fossil fuel combustion, agricultural activities, and automobile emissions were the main contributors to PM_2.5, contributing more than 60% to PM_2.5. Compared with the wet season, the contributions of different sources increased in the dry season (including pollution days); of these sources, automobile emissions and coal combustion emissions increased the most (about nine times and seven times, respectively). Seasonal weather and climate change affected the concentration level of WSIIs. During the wet season, higher temperatures and abundant rainfalls contributed to the volatilization and removal of WSIIs in PM_2.5, while in the dry season and on pollution days, lower temperatures and less precipitation, higher emissions, and poor diffusion conditions contributed to the accumulation of WSIIs in PM_2.5. NH₄HSO₄, (NH₄₎₂SO₄ and NH₄NO₃ were the main chemical forms of secondary inorganic ions. Sufficient NH₃, intense solar radiation, and moist particulate matter surface promoted the formation of secondary inorganic ions. The higher temperature contributed to the volatilization of secondary inorganic ions.

1. Introduction

PM_2.5 is known as fine particulate matter (PM) with an aerodynamic diameter smaller than 2.5μm, which can be emitted directly into the atmosphere from a source (primary particles) or produced via gas-to-particle conversion in the atmosphere (secondary particles) and transported regionally (regional transport of particles) [1,2,3,4]. PM_2.5 pollution causes much concern because it profoundly impacts human health [5,6], atmosphere visibility and air quality [7,8], weather and climate [9], and ecosystems [10]. Water-soluble inorganic ions (WSIIs) are vital chemical compositions of PM_2.5, which make up 30% to 70% (in some cases even >70%) of PM_2.5 mass in the atmosphere [11]. The characteristics of WSIIs affect the physical and chemical behaviors of PM_2.5, such as hygroscopicity, acidity, and alkalinity of aerosol [12]. It also plays an important part in accelerating the formation of particulate matter and degrading atmospheric visibility [2]. The characteristics of WSIIs are regarded as valuable indicators for evaluating PM_2.5 pollution [13,14].

In recent decades, with the increasing rate of industrialization and urbanization, the contributions of fossil fuel combustion, vehicle exhaust, and industry emissions to PM_2.5 has been increasing rapidly, and PM_2.5 pollution has been getting worse in China [2,3]. Many studies have been conducted in China trying to elucidate WSIIs pollution in PM_2.5 [12,15,16,17,18,19,20]. It has been suggested that emissions levels of primary particles, the conversion efficiency of secondary particles, special meteorological conditions, regional transport pathways, and their synergetic effects are the main factors regulating the WSIIs and PM_2.5 levels in heavily polluted Chinese cities [1,2,3,4]. However, few of these studies have focused on the impact of seasonal climate change on PM_2.5, especially in the southern monsoon climate zone. It has been suggested that seasonal climate changes, such as the differences in rainfalls and temperature, may have significant impacts on the characteristics of WSIIs and the formation of PM_2.5 [10,11,21].

Nanning with a low latitude (22°48′ N) is located in southwestern China in Guangxi province, which is in the subtropical monsoon climate zone. Nanning is affected by the southeast and southwest monsoons, and the climate has conspicuous seasonal characteristics. Generally, from May to October is the wet season of the whole year, in which there are intense solar radiation, high temperature and abundant rainfalls (about 80% of the annual rainfall). From November to March is the dry season of the whole year, in which there is weak solar radiation, low temperatures, and little rainfall. It is expected that the geographic location and seasonal climatic conditions can significantly affect the properties of aerosol in Nanning. In recent years, the impact of human activities on air quality has intensified in Nanning [22]. Both the observed trend of pollutant levels and the deteriorating air have attracted increasing concerns of the people. There are some studies that have been conducted in Nanning trying to characterize aerosol pollution, and most of them have focused on organic compounds in the atmosphere [22,23], the single-particle chemical composition in fireworks pollution and ambient aerosols [24,25], and chemical composition of biofuel combustion particulate matter [26]. However, studies about WSIIs of PM_2.5 are relatively few in this region to date, especially studies about the variation of WSIIs in different seasons. To better understand the local PM_2.5 pollution and to effectively investigate their distribution characteristics and possible sources, we conducted this study in Nanning from 1 September 2017 to 31 August 2018. In this work, we reported the daily WSIIs characteristics of PM_2.5 from urban Nanning during different seasons. Combining with meteorological conditions, positive matrix factorization (PMF) models, and air parcel trajectories from regionally transported emissions, the influence factors and potential pollution sources of WSIIs in wet and dry seasons were identified. The unique datasets are expected to improve our knowledge of PM_2.5 pollution properties in monsoon climate cities and provide helpful information for future policy making.

2. Experiments

2.1. Experimental Site and Sampling

Nanning (107°45′ E to 108°51′ E, 22°13′ N to 23°32′ N) is the capital city of Guangxi Province. It is not only the political, economic, and cultural center of Guangxi province but also the largest city in Guangxi province. The air quality of Nanning has worsened because of rapid urbanization and industrialization [26]. In this study, we selected Guangxi University as the experimental site (108°17′ E, 22°50′ N), which is in the north of Nanning (Figure 1). The sampling site was located on the roof of the Comprehensive Experimental Building (an 11 stories high building, about 35 m above ground) on Guangxi University campus, which is the tallest building in Guangxi University; and there are no other taller buildings sheltering it. The sampling site is surrounded by roads with traffic, residential buildings, and business offices. The main street is 0.5 km away from the building. There were no obvious emission sources of atmospheric WSIIs in the vicinity of the sampling site during the study. The obtained PM_2.5 samples represented the air at the sampling location, which avoided single point emission sources.

2.2. Chemical Analyses

The PM_2.5 samples were collected on a quartz fiber filter (QFF, 8″ × 10″, Pall Tissuquartz, USA) by a high-volume air sampler (KC-1000, Qingdao Laoshan Electronic Instrument company, China) with an impactor cutoff of 2.5 μm (aerodynamic diameter) at a flow rate of 1.05 ± 0.03 m³ min⁻¹. We collected the samples (nominally one sample per 24 h) from 1 September 2017 to 31 August 2018 (359 samples). All samples were stored in a refrigerator at −20 °C until analysis in the laboratory. All the filters were weighed before and after sampling using a microelectronic balance with a reading precision of 10 μg after a 48-h equilibration inside a constant temperature (25 °C) and humidity (50%) chamber. Additionally, field blank filters were also collected by exposing filters in the sampler without drawing air through, to account for the introduction of any artifacts during the sampling procedure [27].

The collection of samples and analysis of water-soluble ions were described elsewhere [21,28]. Briefly, the quartz filters were cut into pieces and transferred to a Nalgene tube (50 mL), after which ultrapure water (50 mL) was added to the tube. After ultrasonic vibration, shaking, and centrifugation, the extract was filtered using pinhole filters. The extract was stored for WSIIs concentration analyses. All the procedures were strictly quality controlled to avoid any possible contamination of the samples. Ion chromatography (Dionex Aquion, USA) was used to detect WSIIs concentration. The anions (SO₄²⁻, NO₃⁻, and Cl⁻) were determined by the ion chromatography system with an AS23 4 × 250 mm analytical column, while the cations (NH₄⁺, Ca²⁺, Mg²⁺, Na⁺, K⁺) were determined using the same IC system with a CS12A 4 × 250 mm column. A sample volume of 100 μL was used for both anion and cation analyses. Before analysis of the samples to be tested, external standard solutions (Merck, Germany) were used for making the standard curve, and correlation coefficients greater than 0.999 were required. The limits of detection were 11.50 μg·L⁻¹ for SO₄²⁻, 21.60μg·L⁻¹ for NO₃⁻, 12.10 μg·L⁻¹ for NH₄⁺, 90.00 μg·L⁻¹ for Ca²⁺, 24.70 μg·L⁻¹ for Mg²⁺, 1.00 μg·L⁻¹ for Na⁺, 17.70 μg·L⁻¹ for K⁺, 5.10 μg·L⁻¹ for Cl⁻, and 3.80 μg·L⁻¹ for F⁻. The relative standard deviation was less than 5% for the reproducibility test [21,28].

2.3. Collecting Meteorological Parameters

Weather parameters including wind speed (WS), wind direction (WD), temperature (T), and relative humidity (RH) were collected from the Weather and Climate Information website [29]. Particulate matter and gaseous pollutant concentration data were collected from the Air Quality Study website [30]. These data were published by the ambient air quality monitoring stations (Beihu, Nanning) which are operated by the Guangxi Zhuang Autonomous Region Environmental Protection Agency. The station was about 2 km away from the sampling site, which was the nearest air quality monitoring station around the sampling site. The online monitoring device was located on the roof of the building, about 20 m up from the ground, and there were no other taller buildings sheltering it. The standard method used for gas pollutants monitoring data was based on Specification and Test Procedures for Ambient Air Quality Continuous Automated Monitoring System for SO₂, NO₂, O₃, and CO, which was published by the Ministry of Environmental Protection of the People’s Republic of China in 2013 [31].

2.4. Calculation of Non-Sea Salt (NSS) Aerosols

Since Nanning is near the sea, the impact of sea salt on WSIIs needs to be evaluated. The mass concentrations of non-sea salt SO₄²⁻ (NSS-SO₄²⁻), non-sea salt K⁺ (NSS-K⁺), non-sea salt Ca²⁺ (NSS-Ca²⁺), and non-sea salt Mg²⁺ (NSS-Mg²⁺) were estimated based on the hypothesis that the particles originating from sea salt have the same components as seawater, and the Na⁺ measured was assumed to be derived from sea salt [32,33]. The calculation formulae are as follows:

NSS-SO₄²⁻ = SO₄²⁻ − Na⁺ × 0.2516

(1)

NSS-K⁺ = K⁺ − Na⁺ × 0.0370

(2)

NSS-Ca²⁺ = Ca²⁺ − Na⁺ × 0.0385

(3)

NSS-Mg²⁺ = Mg²⁺ − Na⁺ × 0.12

(4)

2.5. Chemical Conversions of SO₂ and NO₂

The conversion efficiencies of SO₂ to SO₄²⁻ and NO₂ to NO₃⁻ are commonly represented by the Sulfur oxidation ratio (SOR) and nitrogen oxidation ratio (NOR) [34,35], respectively, which are defined, respectively, as

SOR = n-SO₄²⁻/(n-SO₄²⁻ + n-SO₂ )

(5)

NOR = n-NO₃⁻/( n-NO₃⁻ + n-NO₂

(6)

2.6. PMF Model and Back Trajectory Analyses

Positive matrix factorization (PMF) is a source apportionment receptor model. In this study, PMF 5.0 (EPA) was used to quantitatively estimate the source of WSIIs [36,37,38]. When running the PMF model, it is required to input sample concentration data and uncertainty files. The uncertainty file can be input based on observations and equations, as described by the PMF 5.0 user guide. The uncertainty based on observations reflects the errors caused in the process of sampling and measuring samples, which are usually provided by the laboratory or reporting agency. The uncertainty based on equations is usually calculated by concentration, error fraction, and method detection limit (MDL). At the same time, the uncertainty based on equations can let users adjust and control the deviation in PMF solution in a small range. In this study, the uncertainty based on equations was used for calculation, and the relative uncertainty of the sample was estimated by the laboratory. The uncertainty was input into the model as an error matrix, and the ion concentration and sampling time as an input matrix. The calculation method was as follows:

Uncertainty = 5/6 × MDL (xij ≤ MDL)

(7)

$$\mathrm{Uncertainty}=\sqrt{{(\mathrm{Error}\mathrm{Fraction}\times \mathrm{concentration})}^2+{(0.5\times \mathrm{MDL})}^2}(\mathrm{xij}>\mathrm{MDL})$$

(8)

where xij is the ion concentration, and the error fraction is the relative uncertainty (%) of the ion concentration [37].

The number of factors can be determined by the change of Q (robust) and Q (true). When Q (robust) and Q (true) change slowly, this indicates that too many factors are selected. In the range determined by Q (robot) and Q (true), different factor numbers are selected to run the model and test the feasibility of the results, which can determine the appropriate factor number in the PMF model. In this study, when the number of factors increased to 6 in PMF, the changes of Q (robust) and Q(true) were quite slow. Therefore, we ran the PMF model in the range of 1–6 factors.

Back trajectory cluster (BTC) analysis is a useful tool to evaluate the main origins and transport pathways of air pollutants at the receptor sites [28,39,40] and was used to provide a comprehensive view of the potential source regions for WSIIs in PM_2.5 in Nanning. For each day, 3-day (72 h) back trajectories of air masses arriving at Nanning were computed.

3. Results

3.1. Weather Condition and Air Pollutants Concentration in the Sampling Period

The wind speeds were 0.9 to 5.9 m s⁻¹ (mean 2.6 m s⁻¹), and dominant wind directions were east, northeast, and south (Figure 1 and Figure 2a) during the sampling period. The temperature range was 5.2 °C to 30.9 °C (mean 22.1 °C) and RH was 31.0% to 99.0% (mean 78.3%). The annual rainfall was 1491.4 mm, and precipitations mainly occurred in the month of May, June, July, August, September, and October, which were defined as the wet season; the other months were defined as the dry season (Figure 2b).

The average daily concentrations of SO₂, NO₂, CO, and O₃ (8h) were 11 μg·m⁻³, 36 μg·m⁻³, 1.0 mg·m⁻³, and 83 μg·m⁻³, respectively (Figure 2c,d). The PM_2.5 range was 6 to 130 μg·m⁻³ (mean 36 μg·m⁻³), and relatively high concentrations of PM_2.5 were observed during the following days: 25–27 October; 1, 2, 5, 6, and 13 November; 7, 10–12, and 22–28 December; 1–4, 14, 15, 19–23, and 30 January; and 9, 12, 13, 16 and 17 February. For these days, all the PM_2.5 concentrations exceeded the PM_2.5 24-h limitation value of 75 μg·m⁻³ of the Ministry of Ecology and Environment of the People’s Republic of China. We defined these days as pollution days and days with concentrations lower than 75 μg·m⁻³ were defined as normal days (Figure 2d).

During the sampling period, high particulate matter concentrations coincided with high NO₂ and SO₂ concentrations, while the variation of CO and O₃ concentrations were not synchronous with the change of PM concentrations.

3.2. Seasonal Variations of WSIIs

The average concentrations of WSIIs were 19.12 μg·m⁻³ (1.75 to 69.87 μg·m⁻³), which accounted for 51.65% (12.43 to 94.55%) of PM_2.5 concentrations. SO₄²⁻ (47.8%), NO₃⁻ (16.1%), NH₄⁺ (16.9%), and Ca²⁺ (12.1%) were the most important four ions in WSIIs. The ratios of other ions include K⁺ (3.2%), Cl⁻ (2.3%), Na⁺ (1.0%), Mg²⁺ (0.4%), and F⁻ (0.1%) were low. The SNA were the main components of WSIIs in Nanning which was consistent with the results from other regions around the world [41,42]. Seasonally, variations of WSIIs were in accordance with PM_2.5 and N₂O (Figure 2d,e). In the dry season, the WSIIs concentrations were relatively high, while they were low in the wet season (Figure 2e and Figure 3). The concentration of PM_2.5, polluted gas (SO₂, N₂O, and CO) and WSIIs in pollution days were also higher than those in the normal days. The K⁺, Mg²⁺, and Cl⁻ concentrations were relatively high on February 15 and 16, 2018 (Figure 3d,e).

3.3. Calculation Results by PMF Model

Five factors were identified and quantified based on the PMF model (

Table 1. Relative contribution (%) for different ions from potential five sources (factors) in Nanning, using the PMF model.

Sources (%)	Factor	SO42−	NO3−	NH4+	Ca2+	Mg2+	Na+	K+	F−	Cl−	PM2.5
Automobile	1	13.1	79.1	38.5	0.0	0.0	0.0	29.7	1.7	16.4	20.8
Biomass	2	4.8	0.0	0.4	0.0	55.2	0.0	59.3	5.5	8.4	10.5
Fossil fuel/agriculture	3	69.0	0.0	55.4	22.3	21.6	44.2	3.8	0.0	1.3	41.8
Soil dust	4	9.0	20.9	0.0	77.7	20.9	45.9	6.2	86.0	11.6	17.7
Coal	5	4.1	0.0	5.7	0.0	2.3	9.9	1.0	6.8	62.3	9.2

). It could be seen that factor 1 contributed 79.1% to NO₃⁻ and 38.5% to NH₄⁺, factor 2 contributed 59.3% to K⁺ and 55.2% to Mg²⁺, factor 3 contributed 68.5% to SO₄²⁻ and 55.4% to NH₄⁺, factor 4 contributed 86.0% to F⁻, 77.7% to Ca²⁺ and 45.9% to Na⁺, and factor 5 contributed 62.3% to Cl⁻. Their percent contributions to PM_2.5 were 20.8%, 10.5%, 41.8%, 17.7%, and 9.2%, respectively. The variations of contributions from different factors in the different seasons were consistent with WSIIs concentrations (Figure 4), which suggested that normalized contributions were higher in the dry season and pollution days than that in the wet season and normal days.

3.4. Results of Back Trajectory Analyses

All the trajectories in different seasons can be classified into four main categories (clusters) based on their origins, paths, and latitudes (Figure 5). In these four main clusters, the high frequencies of clusters were from east, west, and southwest, and the higher frequencies of clusters corresponded with the higher mass concentrations of WSIIs (Figure 5). To be specific, in the wet season, the main trajectories with high mass concentrations of WSIIs were the short east airflow from eastern Guangdong province and the long southwest air parcel from Vietnam and Myanmar. In the dry season, the main clusters with high mass concentrations of WSIIs were western long transport paths from the north of India and west of Myanmar, and southwest short transport paths from east of Hainan province. On pollution days, the main four clusters were similar to those in the cold months, showing that high frequencies and high mass concentrations of WSIIs in the cluster were the west and southwest transport patterns from India, Myanmar, and the Northern Gulf near Guangxi province.

3.5. Correlations of WSIIs

Results suggested that NH₄⁺ had a strong linear correlation with SO₄²⁻, and the slopes were closed to 1 in the wet season and dry season. On pollution days, the linear correlation between NH₄⁺ and SO₄²⁻ were weak (R² = 0.25), and the slope of the linear equation was <1 (Figure 6a). Excess NH₄⁺ (excess [NH₄⁺] = ([NH₄⁺]/[SO₄²⁻] − 1.5) × [SO₄²⁻]) and NO₃⁻ concentrations were used to show characteristics of nitrate formed via the homogenous gas-phase reaction between ammonia and nitric in this study [43] (Figure 6b). The NO₃⁻ was positively and linearly correlated with excess NH₄⁺ when excess NH₄⁺ >0 (Figure 6b). There was also a strong positive linear correlation between [SO₄²⁻ + NO₃⁻] and [NH₄⁺] and the slope was <1.

3.6. Variations of SOR and NOR

The SOR (mostly >0.1) was more than NOR (mostly <0.1) on normal days and pollution days (Figure 7). Both on normal days and pollution days, SOR and NOR were related to NH₄⁺ (

Table 2. Correlation analyses among SOR, NOR, and weather factors and gaseous pollutants.

Sampling Period	Normal Days		Pollution Days
Parameter	SOR	NOR	SOR	NOR
SO2	0.01	0.41	−0.81	−0.26
NO2	0.2	0.4	−0.48	−0.35
NH4+	0.63	0.72	0.46	0.42
RH	0.04	−0.3	0.44	0.46
T	0.09	−0.6	0.41	−0.62
O3	0.32	0.05	0.21	−0.58

). However, the trend of SOR and NOR varying with NH₄⁺ was different on normal days (Figure 6a,b). It can be seen that when the NH₄⁺ concentrations were low, the SOR increased more quickly, while the NOR increased slowly; when the NH₄⁺ concentrations were high, the SOR increased slowly, while the NOR increased more quickly. In correlation analysis, the RH was positively correlated with SOR (R² = 0.44) and O₃ was negatively correlated with NOR (R² = −0.58) on pollution days (Table 2). On normal days, the relationship between RH and O₃ with SOR and NOR was not significant. In addition, the temperature was negatively correlated with NOR both on normal days and pollution days (Table 2).

4. Discussion

4.1. Correlation of Weather Conditions and Air Quality

It has been suggested that meteorological conditions, including wind speed and wind direction, RH, temperature, and so on, have an important impact on the air quality [44,45]. In our results, there was no obvious correlativity between weather parameters with particulate matter and gaseous pollutants (SO₂, NO₂, CO, and O₃), which suggested that the influence of the wind speed, RH, and temperature on air quality may not be significant in this study. Specifically, wind speed and wind direction are important factors affecting the variability of the concentrations of air pollutants; in particular, the variations of wind speed impact the near-surface aerosol levels [10,21,45]. In our results, there was no significant correlation between wind speed and PM_2.5, these results suggested that the transport of pollutants related to wind speed might be weak. On pollution days, when the PM_2.5 concentration was relatively high (>75 μg·m⁻³), the wind speed (1.1 to 4.5 m·s⁻¹) did not show low values, which suggested that wind speed did not have a major impact on the formation of pollutants (Figure 2a,d). Previous research has shown that a high concentration of aerosol liquid water would dissolve more pollutants and accelerate chemical reactions, which increases the formation of secondary aerosols [21,44,45]. There was no significant correlation between RH and PM_2.5, suggesting that the effect of high RH dissolving more pollutants was not shown in this study. Previous research has also suggested that the relatively low air temperatures before haze episodes favor the partitioning of semi-volatile components and ammonium salts into the particle-phase, which further exacerbates air pollution [45]. In our results, the temperatures on pollution days and normal days did not show obvious changes (Figure 2b), which might indicate that temperature had little effect on air quality. SO₂ and NO₂ were the precursors of SO₄²⁻ and NO₃⁻, respectively, contained in PM_2.5, which could result in the variation of PM_2.5 coinciding with NO₂ and SO₂ [34].

4.2. Source Analysis of Major WSIIs

SO₄²⁻ (47.7%), NO₃⁻ (16.1%), and NH₄⁺ (16.9%) contributed about 80% to the total WSIIs. These ions (SNA) are considered to be secondary particulate components that are converted from gas pollutants [46]. In general, SO₄²⁻ is formed by the gas-phase photochemical conversion of SO₂, mainly from fossil fuel combustion. NO₃⁻ is mainly formed via the conversion of NO and NO₂, which are mainly from automobile and industrial emissions [47]. The high molar ratio of NO₃⁻ to SO₄²⁻ (NO₃⁻/SO₄²⁻ values>1) suggests that mobile sources over stationary sources [45,48]. In our results, NO₃⁻/NSS-SO₄²⁻ values were 0.03 to 3.20 (average 0.62) in Nanning, which were lower than 1 (Figure 3b), revealing that fossil fuel combustion was still the important contributor to WSIIs. Meanwhile, the NO₃⁻/NSS-SO₄²⁻ values were close to that of Beijing (0.58) [19], where the WSIIs were mainly influenced by vehicle emissions, which indicates that WSIIs were partly from vehicle exhaust. SO₂ was lower than NO₂ (Figure 2d), but the SOR and SO₄²⁻ were much higher than NOR and NO₃⁻, respectively (Figure 3a,b), which indicates that the conversion efficiency of SO₂ to SO₄²⁻ was higher than NO₂ to NO₃⁻ in Nanning. NH₄⁺ was formed through reactions between acidic species and NH₃, which was mainly from agricultural activities [49]. The SNA was correlated with the gaseous pollutants (SO₂, NO₂, and CO) which were mainly from coal combustion, biomass burning, and vehicle emissions as showed in

Table 3. Correlation analyses of different WSIIs and gaseous pollutants.

Species	SO42−	NO3−	NH4+	Ca2+	Mg2+	Na+	K+	F−	Cl−	SO2	NO2	CO
SO42−	1	0.63**	0.86**	0.34**	0.27**	0.06	0.54**	0.25**	0.48**	0.53**	0.59**	0.51**
NO3−		1	0.83**	0.32**	0.15*	0.12*	0.51**	0.32**	0.61**	0.68**	0.75**	0.59**
NH4+			1	0.29**	0.16*	0.10	0.50**	0.23**	0.52**	0.61**	0.70**	0.60**
Ca2+				1	0.08	0.26**	0.21**	0.79**	0.33**	0.68**	0.54**	0.09
Mg2+					1	0.18*	0.84**	0.27**	0.56**	0.12*	0.04	0.03
Na+						1	0.11	0.21**	0.17*	0.13*	0.14*	−0.02
K+							1	0.35	0.77	0.41	0.37	0.32**
F−								1	0.46**	0.60**	0.41**	0.08
Cl−									1	0.46**	0.5**	0.35**

** Correlation significant at 0.01 level (two-tailed), * significant at 0.05 level (two-tailed).

, this indicated that the sources of SNA were impacted by these coal combustion, biomass burning, and vehicle emissions. Ca²⁺ (12.1%) is the fourth highest WSIIs in Nanning. In general, consistently high Ca²⁺ and Mg²⁺ concentrations may indicate that they are mainly from biomass burning and industry sources [15,17], and the soil dust has higher Ca²⁺ than Mg²⁺ concentrations [10]. A low NSS-Mg²⁺/NSS-Ca²⁺ value (average 0.08) in Nanning suggested a soil dust source of Ca²⁺. In the correlation analysis of WSIIs (Table 3), there was no significant correlation between Ca²⁺ and Mg²⁺, which indicated that Mg²⁺ has a different source than Ca²⁺. The F⁻ (0.1%) was lower than the other ions. Previous research suggested soil and glass manufacturing industries may contribute to F⁻ [17]. There are not significant emission sources of F⁻ in Nanning, so the concentration of F⁻ was low.

Cl⁻ (2.3%) and K⁺ (3.2%) were less than SNA and Ca²⁺. In PM_2.5, Cl⁻ and K⁺ were mainly from coal, biomass, sea salts, and so on [16]. Some studies indicated that biomass burning produced a high proportion of K⁺ and Cl⁻, and had a significant relationship between K⁺ and Cl⁻ [19]. The K⁺ was more than Cl⁻ (Figure 3d,e), and K⁺ was correlated with Cl⁻ (Table 3) in this study, suggesting that K⁺ was partially from biomass burning. Cl⁻ is also a major chemical composition of sea salt and coal combustion emission [49]. Generally, Cl⁻ was correlated with Na⁺, and Cl⁻/Na⁺ values are about 1.80 in seawater [50]. In our study, most Cl⁻/Na⁺ values were higher than 1.8 in the dry season, most Cl⁻/Na⁺ values were lower than 1.8 in the wet season, and there was no correlation between Na⁺ and Cl⁻ (Table 3). This suggested that there was little contribution of sea salt to the Cl⁻ values, and the Cl⁻ was likely mainly from coal combustion emissions.

4.3. Quantitative Estimation by PMF and Seasonal Variation of Different Sources

Five main factors (sources) were categorized by PMF (Table 1). Source 1 (factor 1) was related to secondary inorganic ions from automobile emission, indicated by high loadings of NO₃⁻ and NH₄⁺. Source 2 (factor 2) was identified to be sources of biomass burning with high loadings of K⁺ and Mg²⁺. Source 3 (factor 3) was associated with a mixture of secondary inorganic ions from fossil fuel combustion and agricultural activities, indicated by prominent loadings of SO₄²⁻ and NH₄⁺. Source 4 (factor 4) was identified to be sources of soil dust, indicated by high loadings of F⁻, Ca²⁺, and Na⁺. Source 5 (factor 5) was identified to be sources of coal combustion emissions, characterized by high loadings of Cl⁻. The contributions of sources (factors) 1–5 to PM_2.5 were 20.8%, 10.5%, 41.8%, 17.7%, and 9.2%, respectively (Table 1), which suggested that secondary inorganic ions from fossil fuel combustion, agricultural activities, and automobile emission contributed more than 60% to PM_2.5.

A previous study on air pollutant emissions inventory in Nanning have shown that the air pollutant emissions of Nanning were arranged in order of magnitude as PM₁₀ (200 thousand tons) > NO_x (120 thousand tons) > CO (84 thousand tons) > SO₂ (41 thousand tons) [51]. The emissions inventory results showed that soil dust sources were the major contributors to PM₁₀, which accounted for 87% of PM₁₀. Industrial combustion and vehicle exhaust emissions were major contributors to NO_x, which accounted for 55% and 27% of NO_x. The industrial combustion emissions were the major contributors to CO, which accounted for 76% of CO. The stationary combustion sources (including thermal power plants, industrial combustion, and civil combustion) were the major contributors to SO₂, which accounted for 82% of SO₂. In these pollutants, SO₂ and NO₂ were the precursors of SO₄²⁻ and NO₃⁻ which were the main components in PM_2.5. This result further suggested that combustion sources and vehicle exhaust emissions were the major contributors to secondary inorganic ions of PM_2.5 in Nanning from the perspective of emissions inventory. This was consistent with the PMF results in our study that secondary inorganic ions from fossil fuel combustion, agricultural activities, and automobile emission contributed more than 60% to PM_2.5, as high SO₄²⁻, NH₄⁺, and NO₃⁻ concentrations in PM_2.5 in our results. Similarly to the results of the emissions inventory, the soil dust sources were identified in PMF results as high Ca²⁺ concentrations in PM_2.5 (which contributed 17.7% to PM_2.5). One difference from the results of the emissions inventory was that we identified agricultural activity sources in the PMF results as high NH₄⁺ concentrations in PM_2.5. Although, the NH₄⁺ may also come from the burning of fossil fuels [10,15].

Results suggested that variation of contributions from different sources were in agreement with ions concentrations (Figure 3; Figure 4), and contributions were higher in the dry season and on pollution days, but lower in the wet season. Compared with the wet season, the contribution of automobile emissions (factor 1) and coal combustion emissions (factor 5) increased the most in the dry season (including pollution days) by about nine times and seven times, respectively (Figure 4). Compared with normal days, the contributions of automobile emissions (factor 1) and coal emissions (factor 5) also increased the most on pollution days, which increased about five times and two times, respectively (Figure 4). This phenomenon may be associated with seasonal weather variation. During the wet season, T and rainfall were higher (Figure 2b), which contributed to the volatilization and removal of WSIIs in PM_2.5 [21,33]; thus, the concentrations and contributions were high in this season. Conversely, in the dry season, as the T and rainfall decreased (Figure 2b), the volatilization of the SNA and ion removal rate also decreased, which contributed to the increase of WSIIs in PM_2.5; thus, the concentrations and contributions were low in this season [2,21]. On pollution days, besides the decrease of T and rainfall, the WS was low (Figure 2a,b) and the concentrations of gaseous pollutants (SO₂, NO₂, O₃, and CO) (Figure 2c,d) and the transformation efficiency of NO₂ (NOR) and SO₂ (SOR) were high on these days (Figure 3b). These specific weather conditions promote the accumulation of pollutants.

The normalized contributions of biomass burning (factor 2) and coal combustion emissions (factor 5) were the highest on February 15 and 16, 2018, which was consistent with the highest K⁺, Mg²⁺ and Cl⁻ concentrations on the two days (Figure 3d,e). This result may be related to the display of fireworks and firecrackers. February 15 and 16, 2018 were the Chinese New Year’s Eve and the first day of Chinese New Year, respectively. A great number of fireworks and firecrackers were displayed to celebrate the festival on these two days, which could increase the concentration of Mg²⁺, K⁺ and Cl⁻ [11,19].

4.4. Main Origins and Transport Pathways of Air Pollutants

It has been suggested that WSIIs of PM_2.5 in Nanning were easily enriched in the east and southwest trajectories in the wet season, and the WSIIs of PM_2.5 were apt to be enriched in the west and southwest trajectories in the dry season and during pollution days (Figure 5). The air masses in the dry season and on pollution days brought more WSIIs pollution, while in the wet season they brought less WSIIs pollution.

4.5. Potential Formation Mechanisms of Major WSIIs

In our results, NH₄⁺ was linearly related to SO₄²⁻ on the normal days and pollution days (the linear correlation was relatively weak), and the slopes of the linear equations between NH₄⁺ and SO₄²⁻ were close to 1 and 0.5 on the normal days and pollution days, respectively (Figure 6a). The equivalent ratios of NH₄⁺ to SO₄²⁻ in (NH4)₂SO₄ and NH₄HSO₄ were 1.0 and 0.5 [16,45], respectively. Therefore, (NH₄)₂SO₄ might be the main chemical form of SO₄²⁻ and NH₄⁺ on normal days, while NH₄HSO₄ might be the main chemical form of SO₄²⁻ and NH₄⁺ on pollution days. The NH₄HSO₄ and (NH₄)₂SO₄ were mainly produced by the homogeneous reactions between H₂SO₄ and ammonia during normal days and pollution days [1]. Except for H₂SO₄, HNO₃ can also react with ammonia to produce secondary inorganic ions; for example, HNO₃ reacts with ammonia to form NH₄NO₃ [16]. Generally, ammonia is required to react with H₂SO₄ to generate NH₄HSO₄ or (NH₄)₂SO₄. If ammonia is sufficient to produce H₂SO₄, the excess ammonia reacts with HNO₃ to generate NH₄NO₃ [43]. In our results, the excess NH₄⁺ was >0 in most of the samples, and the NO₃⁻ was positively and linearly correlated with excess NH4+ in this case (Figure 6b). This suggested that NH₄NO₃ was also the main chemical form of NO₃⁻ and NH₄⁺. Compared to the dry season, more samples in the wet season were NH₄⁺-deficient (excess NH₄⁺ < 0), which may be related to the volatilization of NH₄NO₃ in the high-temperature condition in the wet season (Figure 6b) [16]. The formation of NO₃⁻ in NH₄⁺-deficient circumstances could be related to crustal species or hydrolysis of N₂O₅ in PM_2.5 [43]. Both the NH₄⁺ vs. SO₄²⁻ concentration and excess NH₄⁺ vs. NO₃⁻ concentrations were high on pollution days; this indicates that more NH₄HSO₄ and NH₄NO₃ were formed in these days.

The sum of SO₄²⁻ and NO₃⁻ was linearly correlated with NH₄⁺ with the slope <1 (Figure 6c), which may suggest there were redundant SO₄²⁻ or that NO₃⁻ existed in other forms. CaSO₄ and Ca(NO₃)₂ from soil dust may be chemical forms of SO₄²⁻, NO₃⁻, and Ca²⁺, as the weak correlations in [Ca²⁺] vs. [SO₄²⁻] and [Ca²⁺] vs. [NO₃⁻] (Table 3) indicate.

4.6. Factors Affecting SOR and NOR

Generally, NO_x and SO₂ are the precursors of NO₃⁻ and SO₄²⁻ formation, respectively, and they are the main factors regulating SOR and NOR [10,52]. Both on normal and pollution days, SOR and NOR were positively and linearly correlated with NH₄⁺ (Table 2 and Figure 7), indicating that ammonia was also an important factor impacting the transformation of SO₂ and NO₂ [35,53]. This showed that when the NH₄⁺ concentrations were low (NH₄⁺ <4 μg·m⁻³), the SOR increased faster, while the NOR increased more slowly; when the NH₄⁺ concentrations were high (NH₄⁺ >4 μg·m⁻³), the SOR increased slowly, while the NOR increased more quickly. The different growth trends of SOR and NOR with the NH₄⁺ concentrations confirmed the different formation mechanisms of SO₄²⁻ and NO₃⁻. The SO₄²⁻, NO₃⁻, and NH₄⁺ in PM_2.5 were mainly formed by the reaction of H₂SO₄ and HNO₃ with ammonia. When ammonia was low, ammonia preferentially reacted with H₂SO₄ to form NH₄HSO₄ or (NH₄)₂SO₄; more SO₄²⁻ was formed and the SOR increased rapidly. When ammonia was high, excess ammonia reacted with HNO₃ to produce NH₄NO₃; during this time, more NO₃⁻ was formed and NOR increased rapidly [11]. This also confirmed the deduction of the discussion in Section 4.5.

It has been suggested that both OH radicals and O₃ concentration and particulate matter liquid water content (represented by RH) also had a significant influence on the secondary transformation of NO₂ and SO₂. The moist particulate matter (high RH) and intensive photochemistry process (high OH radicals and O₃ concentration) were favorable for the formation of SO₄²⁻ and NO₃⁻ [44,53,54]. The RH was positively correlated with SOR and NOR, and O₃ was negatively correlated with NOR on pollution days (Table 2, Figure 7c,d), which indicated that high RH could promote the secondary formation of SO₄²⁻ and NO₃⁻, and secondary formation of NO₃⁻ could consume more O₃. However, on normal days, the effects of RH and O₃ on SOR and NOR were not significant. There were also some differences between the formative mechanisms of SO₄²⁻ and NO₃⁻. The formation of NO₃⁻ was not only affected by O₃ but also by temperature, as indicated by the negative correlation between NOR and temperature both on normal and pollution days (Table 2). It has been confirmed by previous studies that the reaction of the gas-solid equilibrium for NH₄NO₃ was largely influenced by temperature [55,56]; the lower temperature was conducive to NH₄NO₃ formation, while the higher temperature was favorable to volatilization of NH₄NO₃. This was also the reason why the contribution of automobile emissions decreased in the wet season (relatively high-temperature conditions), as discussed in Section 4.3.

5. Conclusions

WSIIs in PM_2.5 were measured in Nanning, showing that SNA were the main components of WSIIs (contributing about 80% to the total WSIIs), and the total WSIIs were the main pollutants in PM_2.5 (accounting for 51.65% of PM_2.5 mass). These WSIIs were mainly from secondary inorganic ions, soil dust, biomass burning, and coal combustion emissions (primary emissions). Furthermore, secondary inorganic ions that form fossil fuel combustion, agricultural activities, and automobile emissions contributed more than 60% to PM_2.5. This result was also consistent with the results of an air pollutant emission inventory in Nanning, which suggested that stationary combustion sources (including thermal power plants, industrial combustion, and civil combustion), vehicle exhaust, and urban road dust were the three major sources for air pollutants in Nanning.

Compared with the wet season, the contributions of different sources increased in the dry season (including pollution days); of these sources, automobile emissions and coal combustion emissions increased the most (about nine times and seven times, respectively). Seasonal weather and climate change affected the concentration levels of WSIIs. During the wet season, higher temperatures and abundant rainfalls contributed to the volatilization and removal of WSIIs in PM_2.5, while in the dry season, lower temperatures and little precipitations, and higher emissions contributed to the increase of WSIIs in PM_2.5. During the pollution days, besides low removal rate and high emissions, the poor dispersion weather and high transformation efficiency of NO_x and SO₂ were in favor of the accumulation of pollutants. Furthermore, the air masses brought more WSIIs pollution in the dry season and pollution days, while they brought less WSIIs pollution in the wet season.

NH₄HSO₄, (NH₄)₂SO₄, and NH₄NO₃ were the main chemical forms of secondary inorganic ions. Except for the NO_x and SO₂, the NH₃, O₃, RH, and T also had important impacts on the formation of secondary inorganic ions. Sufficient NH₃, intense solar radiation (high O₃ concentration), and moist particulate matter surfaces (high RH) promoted the formation of secondary inorganic ions. The high temperature contributed to the volatilization of secondary inorganic ions.