Sulfur, lead, and mercury characteristics in South Africa coals and emissions from the coal-fired power plants

Coal-fired power plants (CFPPs) are the dominant source of electricity in South Africa due to coal abundance in the country. However, emissions of SO2, Pb, and Hg have raised serious environmental and public health concerns. Hence, to reduce emissions and utilize coal efficiently, it is essential to estimate emissions trends, understanding existential forms of the elements in coals, and their affinities to minerals, organic matter, and pyrite. Therefore, this paper aimed to assess the forms of elemental occurrence of sulfur (S), lead (Pb), mercury Hg affinities in the coals using statistical correlations and their isotopic compositions. This study also estimated SO2, Pb, and Hg emissions from 1971 to 2018 from the CFPPs based on activity data and emission factors. Based on the results, South African coals mainly comprise equivalent fractions of organic and pyritic S. The Pb were correlated with ash content (R = 0.61), Si, Al, and Ti, which indicates clay mineral-bound Pb. However, the highest Pb206/Pb207 and the lowest Pb208/Pb206 in South Africa coals which contain high inertinite (organic matter) and low S, also reveal organically associated Pb. Similarly, clay minerals linked Hg appeared as of Hg relationship with ash (R = 0.641) and major elements, and the remaining could be an organic matter associated. As an organic matter-associated element least cleanability and readily oxidizing nature, burning South African coals containing a substantial quantity of organic S and organically bound Pb and Hg without washing results in higher emissions. The estimated SO2, Pb, and Hg emissions were 355.84 Gg, 168.91 tons, and 4.17 tons in 1971, and increased to 1468.13 Gg, 696.89 tons, and 17.20 tons in 2018, respectively. The values approximately increased by a factor of 4.


Introduction
In South Africa, coal-fired power plants (CFPPs) generate more than 90% of the total electricity due to the natural coal abundance and the consequent comparative cost advantage. However, despite the CFPPs providing the much-needed low-cost energy for economic development, their existence has become a concern for ecosystem sustenance (Dunmade et al. 2019). The emission of harmful gases and trace metals has attracted environmental and public health attention (Rowe et al. 2002;Hossain et al. 2015). Several factors can influence CFPPs' emissions: concentration of the elements, affinities to other elements and minerals, and forms of elemental occurrence (an existence of element dominantly occur with) in the burned coals are among the considerable drivers (Vejahati et al. 2010). South African coals are high in minerals and ash, but low in S (sulfur) and trace metals than the global coals average (Wagner and Hlatshwayo 2005;Bergh et al. 2011). Despite the low S and trace metals, emission of SO 2 , Pb, and Hg from coal combustion have posed a challenge in coal utilization to generate electricity (Maya et al. 2015;Munawer 2018). In view of this, understanding the occurrence forms of S, Pb, and Hg in the coals can provide better information about their cleanability from coals during coal washing (removal potential of an element via coal washing) and combustion behavior for minimizing emissions (Finkelman 1994;Vejahati et al. 2010). Tang et al. (2020) explained that ash and S contents in coals have a substantial role in trace metals' cleanability during coal washing. Hence, the affinity of Pb and Hg toward clay minerals, organic components, and sulfides that can be evaluated through their correlations with ash and S indirectly can illustrate the dominant forms of their elemental occurrence. The elements associated with ash demonstrate clay mineralbound, while those with S indicate sulfide-bound (Yudovich and Ketris 2005;Fang et al. 2014). Besides, elements associated with sulfides can be cleaned more readily (Tang et al. 2020), and organically or sulfide associated can easily vaporize during combustion (Vejahati et al. 2010).
Furthermore, besides its application for tracing emission sources and distribution, isotope chemistry of elements in coals can reveal forms of their elemental occurrence (Sun et al. 2014a;Wiederhold 2015). Most elements have more than one stable isotope. It has been established that the stable isotope ratio in a sample to a ratio in standard material gives isotopic composition expressed delta value (δ). Isotopic composition of the elements in coals does not only vary with geological deposits and geochemical processes, origin, and environmental factors during coalification (Smith and Batts 1974;Díaz-Somoano et al. 2007;Sun et al. 2016), but also vary with the form of elemental occurrence in the coals. S in coals exists in multiple forms, and isotopic composition (δ 34 S) of organic S form in low-S coals varies narrowly but widely in high-S coals (Smith and Batts 1974;Xiao and Liu 2011). Hg has seven isotopes with mass-dependent isotope fractionation (δ 202 Hg) and mass-independent fractionation (Δ 199 Hg and Δ 201 Hg) formed due to its behavior and reactions in the environment (Bergquist and Blum 2009). An inverse relationship between low δ 202 Hg and high Δ 199 Hg can demonstrate organically bound Hg, and the reverse could be an indication of inorganically bound Hg (Sun et al. 2014b). Similarly, Pb comprises four stable isotopes, and Pb isotopic ratio in coals differ with respect to the Pb form of occurrence in the coals (Díaz-Somoano et al. 2007).
Utilization of the lowest grade and high ash content coal to generate electricity in South Africa (Jeffrey et al. 2014), in addition to other factors such as the form of elemental occurrence and concentration level of the substances (S, Pb, Hg) in the coals may lead to high SO 2 , Pb, and Hg emissions. These pollutants pose significant harm to the environment and equally constitute a public health risk. For instance, atmospheric SO 2 has been implicated in the formation of acid rain via atmospheric hydration. Acid rain induces the deterioration of physical infrastructures and negative impact on plants and microorganisms. In another vein, the disease burden of coal utilization emissions is of much concern to the global health community. Atmospheric SO 2 has been implicated in series of respiratory diseases and cancer complications. According to studies in China, a 1% increase in atmospheric SO 2 had raised the death rate resulting from SO 2 -induced respiratory problems by 0.067%, and SO 2 -related lung cancer by 0.004%, which translates to over 100,000 individuals (Chen et al. 2012(Chen et al. , 2018. Exposure to Pb emission has negative impact on kidney, hematopoietic, cardiovascular, renal, and reproductive systems (Demayo et al. 1982), while exposure to Hg emission causes neurological, renal, cardiovascular, and reproductive diseases (Kim et al. 2016). So far, a very handful of studies have focused on health risks in South Africa. A study done by Siyongwana and Shabalala (2018) reported that 61% of 100 residents near mining areas responded to lung-related problems in a small village in Mpumalanga, a province where most coal mining areas are located.
To date, few emission-related studies have been conducted in South Africa, and most concentrated on Hg emissions. Leaner et al. (2009) (Muntean et al. 2018). Later, Brunke et al. (2012) and Garnham and Langerman (2016) assessed 8 years emissions (2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015), and the values ranged from 14.8 to 23 tons. All previous Hg emission studies were limited to a relatively short period and were conducted years ago, and most of the reported values were also overestimated. Few studies are available on Pb and SO 2 emissions in South Africa. Lee et al. (2014) evaluated Pb emissions from 1971 to 2008, and the estimated were 309-1363 tons. Also, Girmay and Chikobvu (2017) assessed SO 2 emissions efficiencies of the CFPPs and established that the emissions ranged from 6 to 9 g SO 2 /kWh. As emissions of SO 2 , Pb, and Hg have severe environmental and health implications, there is an obvious need for continuous monitoring of their emissions. Estimating the annual emissions of these pollutants is essential to understand the trends and evolution of better emission control. Therefore, this study aimed to assess the existential forms of S and evaluate the affinities of Pb and Hg to clay minerals, organic matter, and sulfide via their association with ash and S as well as their isotopic compositions in the coals. The study further estimated SO 2 , Pb, and Hg emission trends from the electricity generated from CFPPs during 1971-2018 based on the activity data and emission factors.

Distribution of coal deposits and coal-fired power plants
As shown in Fig. 1, coal deposits in South Africa were mostly formed during the Permian period and the rest during the Carboniferous and Triassic periods. The deposits are abundantly distributed in the provinces of Mpumalanga (Witbank and Highveld), Free State, and KwaZulu-Natal. The strata of the coal-bearing Karoo sequence consist of Ecca, Dwyka, Beaufort, and Stormber Groups (Cairncross and Cadle 1988;Cadle et al. 1993), and coals at the Vryheld formation under the Ecca Subgroup are economically significant (Snyman and Botha 1993). The abundance of coal reserves motivates the country's dependence on CFPPs for electricity generation, and most of them are located in the coal-bearing region of Mpumalanga province (Fig. 1). In 1971, the generating capacity of the CFPPs was only 5.505 GW but raised over time to 40.591 GW as of 2018 (Fig. 6). Most CFPPs were established between 1981 and 1990. The CFPPs were designed to run on subcritical technology, with lower efficiency (average efficiency of 38%), higher coal consumption (≥ 380 g/kWh), and a higher emission rate (Muntean et al. 2018). At present, only Particulate Matter (PM) emission controls have been installed in the CFPPs such as Fabric Filter Plants (FFP), Electrostatic Precipitators (ESP) + Flue Gas Conditioning (FGC) with approximately 50% trace metals removal efficiency, and Electrostatic Precipitators (ESP) with around 10% removal efficiency (Table S1) (Dabrowski et al. 2008;Garnham and Langerman 2016). These PM emission controls can reduce trace metal emissions (Rallo et al. 2012). However, almost all CFPPs lack SO 2 emission control (Kolker et al. 2014).

Data sources
Annually generated electricity from coal was obtained from the International Energy Agency (IEA 2019). Total SO 2 emitted from public electricity and heat from 1971 to 2015 in South Africa were gathered from the global EDGAR inventory (version 4.3.2) (http:// edgar. jrc. ec. europa. eu), and SO 2 detected (Ozone Monitoring Instrument, OMI) from 2005 to 2018 from National Aeronautics and Space Administration (NASA) Goddard Space Flight Center (https:// so2. gsfc. nasa. gov). A total of 54 samples of elemental concentrations (S, Pb, and Hg, ash content, and major elements in the coals analyzed on a dry whole-coal basis) of the South Africa coals were obtained from the United States Geological Survey world of coal quality inventory (USGS WoQI) (Tewalt et al. 2010) and a study in South African coals by Wagner and Hlatshwayo (2005) (Table S2). The Pb isotope ratios (Pb 206 /Pb 207 and Pb 208 /Pb 206 in the coals, soil, aerosols, Pb ores, coal fly ash, and leaded gasoline) (Tables S3  and S4), and Hg isotopic compositions (δ 202 Hg and Δ 199 Hg) (Table S5) in the coals were compiled from various studies (Bi et al. 2017;Bollhöfer and Rosman 2000;Díaz-Somoano et al. 2009;Erel et al. 1997;Farmer et al. 1999;Monna et al. 2006;Soderberg and Compton 2007;Sun et al. 2014a;Witt et al. 2006).

Pb and Hg affinities with ash and S, and occurrence forms in coals
Based on previous studies, the degree of affinity of the elements in coal to other components of the coal matrix or elements can be assessed through statistical relationships (Eskanazy et al. 2010;Finkelman 1994). Linear regression and Pearson correlation between elements and ash content and among the elements themselves in coals are commonly used indirect approaches for interpreting the relationships and possible existential forms (Eskanazy et al. 2010). Zhao et al. (2019) explored nature of reporting basis of elemental and mineralogical contents (whole-coal basis or an ash basis) can influence the relationship between elements in coals, and contents analyzed on a whole-coal basis can provide a reliable relationship between two elements if either or both partially associated with organic matter. In this case, the strength of Pb and Hg association with ash and S in South African coals (a whole-coal basis) was evaluated using linear regression in Origin Pro-2016 to depict their affinity to clay minerals, inorganic, and organic components of the coal. In addition, Pearson correlation and stepwise multivariable regression were used to explore inorganic affinities of S, Pb, and Hg using the strength of their correlation coefficients and coefficient of determinations with the major elements in coals reported on a whole-coal basis (Chelgani 2018). The correlation coefficient defines the strength of a linear relationship, and the coefficient of determination explains the variability of one element caused by its relationship to another related element ). Moreover, their isotopic compositions (Tables S4  and S5) were compared with isotopic compositions of other countries' coals using R software (version 3.6.1) to describe the nature of the S, Hg, and Pb in South Africa coals. The concentration data of some elemental forms of S in South African coals collected from previous studies were used to evaluate their proportion.

Estimations of SO 2 , Pb, and Hg emissions from coal-fired power plants
SO 2 emissions from the electricity generated from CFPPs was estimated using Eqs. 1 and 2 (Bond et al. 2004;Zhao et al. 2010;Xiong et al. 2016), and similarly, Pb and Hg emissions via Eq. 3 (Li et al. 2012;Lee et al. 2014). These methods have been used by several studies in estimating historical emission trends of hazardous elements and gases from CFPPs using activity data and emission factors (Lee et al. 2014;Xiong et al. 2016).
where, E (SO2) refers to estimated SO 2 emission (Gg), AC coal is electricity generated from coal (kWh), C is the quantity of coal to generate 1 kWh electricity (kg), X is a fraction of fuel used by particular technology (Bond et al. 2004), and i represent total CFPPs.EF is emission factor of the emitted pollutant, and for SO 2 ( EF (SO2) ) can be calculated using Eq. 2, S coal is S content in coals, S r is S retention in ash, and is the removal efficiency of control technologies in CFPPs. E (Pb,Hg) are emissions (tons) of Pb and Hg, E coal is content of the emitted element in burned coal (mg/kg), and PM represents particulate matter control capacity to remove Pb and Hg emissions.
Due to the lack of coal consumption data from the CFPPs, coal-generated electricity was used to calculate the annually consumed coal data for generating electricity in South Africa. The value of C (Eq. 1) used to convert the generated electricity to consumed coal data was assumed from the average quantity of electricity to generate 1 kWh in Indian CFPPs (0.72 kg/kWh) and quantity of electricity to generate 1 kWh in the USA CFPPs (0.47 kg/kWh) (Guttikunda and Jawahar 2014; Lee et al. 2014). Power generations in South Africa have been and still use pulverized coal combustion boilers, a type of lower efficiency and higher emission rate (Muntean et al. 2018). During coal combustion, most of the S in coal is emitted to the atmosphere and some retain in coal ash. S retention in coal ash is characterized by the percentage of total S retained in the ash after coal combustion (Lu et al. 2010). The coal supplying coal mining located around the CFPPs have analogous geological formations and similar coal compositions. Hence, 0.92 ± 0.31% average S in the coals supplied to CFPPs (Kalenga et al. 2011) and S retention ratio of 0.15 in pulverized coal boilers (Lu et al. 2010) were used to evaluate the SO 2 emission factor. PM emission control has been deployed in most CFPPs (Table S1), but none of the CFPPs has been designed with SO 2 emission control (Kolker et al. 2014). In this incidence, the SO 2 is assumed zero in the calculation of EF (SO2) .
Release rates of trace metals during coal combustion depend on type of boiler, operation conditions, and volatility of trace metals. Pb is a moderate volatile element. To estimate Pb emission, an average of 11.2 ± 0.08 mg/kg Pb (Cairncross et al. 1990;Wagner and Hlatshwayo 2005;Monna et al. 2006;Maya et al. 2015) with the assumption that 80% of the Pb in the coal tend to emit (Deng et al. 2014;Lee et al. 2014) were applied. PM emission controls reduction of Pb and Hg emissions vary in efficiencies (Fang et al. 2014), and an average of 0.45 was applied in this calculation as PM emission controls of FFP and ESP + FGC with approximately 0.5 reduction factor are installed in most CFPPs and ESP with 0.1 reduction factor in two of the CFPPs (Table S1). Hg is a high volatile element, and high temperatures in CFPPs' boilers emit a higher percentage of the Hg in coals (Dabrowski et al. 2008). The mass of Hg emitted to the air mainly depend on Hg content in the coals, and an average of 0.27 ± 0.07 mg/kg Hg in the coals burned in the CFPPs (Garnham and Langerman 2016;Sun et al. 2014a), scenario one emission reduction (50%) in CFPPs proposed by Muntean et al. (2018), and assumption of 95% of the Hg in coals emission (Pacyna and Münch 1991) were used to estimate Hg emission in this study.

Results and discussion
Sulfur in the coals S in coals varies within deposits where it formed (Liu et al. 2001 (Matjie et al. 2018). Furthermore, an inverse relationship between total S and ash content in coals can demonstrate organic S abundance (Shao et al. 2003). However, neither inverse nor direct correlation was observed in South African coals based on the data from USGS WoCQI (Fig. S1), signifying equal proportion.
Pearson correlation between total S and major elements in coals (Table 1) showed a relatively strong S correlation with Na but poor correlation with Fe, suggesting that sulfide S is not dominant. Besides, stepwise multivariable regression between total S and major elements (supplementary material part-II) demonstrated that the S affinity could be estimated by the concentrations of Na and Ca, possibly Nabearing minerals associated when Ca content is low (Chelgani 2018). According to Finkelman et al. (2018), 65% of the Na in low-rank coals occur organically associated and 35% as clay associated. Hence, the strong correlation of S with Na and moderate correlation (R = 0.78) of the S predictive equation by major elements (supplementary material part-II) may infer organically bound S. South Africa coals are generally low in S (< 1%) and high in inertinite maceral (organic constituent in coals), and organic S is dominant in inertinite rich and low-S coals (Roberts 1988). Inertinite is an oxidized organic material in coals. Onifade and Genc (2018) analyzed total S and pyrite in the coals from the Witbank area, and the correlation plotted between them showed more pyrite in the coals with S content > 1% (Fig. S2). In contrast, organic S is more dominant in the low-S coals (< 1%).
The δ 34 S in coals can reveal S forms. δ 34 S in South Africa coals ranged from + 1.1 to + 7.2‰ (Smith et al. 1981). The δ 34 S range is close to δ 34 S of organic S in low-S Australian coals (+ 4.8‰ to + 7.3‰) that has similar geological formation and S concentration to South African coals (Smith and Batts 1974) and low-S Danville coals in Indiana in the USA (Yaofa et al. 2008), + 6.2‰ to + 9.9‰. As δ 34 S in pyritic and organic S of high-S coals show a wide variation range (Westgate and Anderson 1984;Yaofa et al. 2008), the small δ 34 S variation range of the South Africa coals may indicate organic S domination originating from coal-forming plants decomposed at cold climatic conditions (Cairncross and Cadle 1988). The cold climatic condition can be an additional factor for a smaller δ 34 S variation range by impeding bacteria-initiated isotope fractionation, as bacteria are primarily active at warmer temperatures and can fractionate to broader and more negative δ 34 S of the organic S in coals (Hackley and Anderson 1986;Xiao and Liu 2011).

Sulfur dioxide (SO 2 ) emissions
Organic S removal from coals through coal washing is weak but oxidizes readily during combustion, whereas pyritic S can easily be cleaned (Tang et al. 2020). Coals supplied to the CFPPs in South Africa are not washed (Kolker et al. 2014). Hence, coal combustion with significant organic S and considerable pyritic S proportions without washing can significantly enhance SO 2 emissions. The historical SO 2 emission trends  from the CFPPs in South Africa using Eq. 1 ranged from 355.84 Gg in 1971 to 1468.13 Gg in 2018 (Fig. 2a). The SO 2 emission increased by a factor of 4 from 1971 to 2018, and South Africa ranked fourth the largest globally on SO 2 emission from CFPPs and related health impacts (Oberschelp et al. 2019). The risk could be more in Mpumalanga and Gauteng Provinces, where most CFPPs are located. A cumulative 52,110.53 Gg SO 2 was emitted since 1971, and comparatively, the higher emission was during 2000-2011. The SO 2 emission decrease after 2011 was due to a reduction in coal-generated electricity and some emission controls. An average of 6.9% reduction in electricity generation in 2011-2018 brought a reduction of 7.6% in SO 2 . However, coal-generated electricity reduction since 2011, despite a slight increase in generating capacity of the CFPPs (Fig. 6), could also be attributed to CFPPs' efficiency deterioration. In this case, higher emissions would be expected, and the suggestion can be supported by the increment in SO 2 emissions based on the annual OMI detected data (2005-2018) from NASA. However, the emissions around the Kriel area showed decreasing (Fig. 2b).
The ratio of the currently estimated to the total SO 2 emissions of various sources during 1971-2015 collected from EDGAR global emissions inventory (Fig. 2a) ranged from 32.52 to 50.41%, showing CFPPs are the primary source of the SO 2 in the region. The comparison of the present estimations with data  of total SO 2 emissions from public electricity and heat production provided by EDGAR global emissions inventory (Fig. 2a) demonstrated a similar pattern but somehow lower in quantity due to the emission from electricity and heat production includes other smallscale heat sources. CFPPs generated approximated 90% of the electricity and the remaining from other sources (Jeffrey et al. 2014). The average of the presently estimated was 1533.15 ± 47.36 Gg (mean ± SD), matches to the average during 2005-2018 (1522.52 ± 95.82 Gg SO 2 ) detected using OMI by NASA and to the emissions in 2016 (1553.81 Gg SO 2 ) by Gray et al. (2019). Furthermore, the average emission rate was 6.52 g SO 2 /kWh (10.9 g/kg coal), and this was close to the rate (6-9 g SO 2 /kWh) calculated by Girmay and Chikobvu (2017). The rate is also close to the value of 6.32 g SO 2 /kWh obtained by dividing OMI detected SO 2 with coal-generated electricity during 2005-2018 in South Africa (IEA 2019). Thus, the similarities in emission factors could indicate the credibility of the currently estimated historical emission trend. However, Marais et al. (2019) have used CFPPs' generating capacity to evaluate SO 2 emission in 2018 in South Africa, and the result (2176.6 Gg) was higher than presently estimated in that year by a factor of 1.5. Some uncertainties could relate to the assumptions of S retention and the absence of SO 2 emission control. The consumption of low standard and high ash coals to generate electricity in South Africa (Jeffrey et al. 2014) can reduce the efficiency Fig. 2 a Annually consumed coal in the CFPPs calculated from electricity generated from coal, total SO 2 and from public electricity and heat during 1971-2015 collected from EDGAR global emissions inventory (https:// edgar. jrc. ec. europa. eu/ emiss ions_ data_ and_ maps), and the historical (1971-2018) SO 2 emissions estimated in this study. b Ozone Monitoring Instrument (OMI) detected SO 2 emissions in South Africa from NASA Goddard Space Flight Center of the CFPPs and subsequently raise S retention and SO 2 emissions/kWh (Xiong et al. 2016).

Lead in the coals
The average Pb concentration range in South Africa coals (5.92-17 mg/kg) is almost half the global coals average (35 mg/kg) (Cairncross et al. 1990;Wagner and Hlatshwayo 2005;Monna et al. 2006;Maya et al. 2015;Díaz-Somoano et al. 2009). Pb enrichment in coals is ascribed to the geological and geochemical processes, coal-forming plant species, and depositional environment. Gangamopteris and Glossopteris gymnosperms were the common coal-forming plants during the Permian period in South Africa (Falcon 1986), in which their Pb content range is 0.9-13 mg/kg (Fang et al. 2014) could be the origin of the Pb in the coals. To reduce Pb concentration from coal before combustion is economically and environmentally advantageous over installations of emission control devices. So, assessing Pb affinity to ash and S in coals is essential to predict the ease of cleanability.
Linear regression between Pb and S and between Pb and coal ash content plotted using the data in Table S2 are shown in Fig. 3a, b, respectively. The correlation of Pb with ash content (R = 0.610) and with the total S (R = − 0.268) could depict more clay minerals bound Pb than sulfides bound. Pearson correlation between Pb and major elements showed Pb strong correlation with Si, Al, and Ti at P < 0.01 (Table 1), suggesting clay minerals affinity Pb as clay minerals (primary kaolinite) are common Al-bearing minerals (Finkelman et al. 2018). In addition, stepwise multivariable regression between Pb and the major elements (supplementary material part-II) suggests that Pb affinity could be influenced by Ca and Ti contents in the coals. The Pb may exist bound with Ti-bearing minerals when Ca content is low (Cheng et al. 2014). Overall, Pb moderately related with the major elements and coal ash, whereas a weakly related with S, could indicate a certain organic affinity Pb. In Fig. 3b, most Pb contents are observed at relatively low-S contents (< 1%) containing high inertinite coals (Roberts 1988), which implies organically associated Pb. Generally, South Africa coals are low-rank, high ash, and have low calorific value (Onifade and Genc 2018). Interestingly, the coals of the lowest quality are supplied to the CFPPs. Pb in low-quality coals mostly appeared as organically associated (Goodarzi and Swaine 1993), an existential form that can be cleaned less efficiently but vaporize easily during combustion than inorganic bound Pb (Kershaw and Taylor 1992). Hence, CFPPs have become the dominant Pb emission source in South Africa.
Pb isotope ratio in coals has several applications due to its resistance to isotopic composition fractionation during environmental processes (Erel et al. 1997;Díaz-Somoano et al. 2009). One of the most important applications of the Pb isotopic ratio is its application in identifying Pb emission sources. Moreover, it is also helpful to identify Pb existential forms in coals (Díaz-Somoano et al. 2007). Based on the data in Table S3, the average Pb 206 / Pb 207 and Pb 208 /Pb 206 in South Africa coals are 1.215 and 2.036, respectively. Comparing the Pb isotope ratios of the South African coals with the coals of similar geological formations (Gondwana) and coal-rich countries can demonstrate Pb nature in the coals. Relatively, the heaviest Pb 206 /Pb 207 (Fig. 4a) and the lowest Pb 208 /Pb 206 (Fig. 4b) in South Africa coals indicate more radiogenic Pb, unlike  (Tewalt et al. 2010;Wagner and Hlatshwayo 2005) the old Pb ores in the country (Fig. 5). In this case, it can be said that Pb in South Africa coals could be from the younger Pb ores (Witt et al. 2006). The insignificant difference Pb 206 /Pb 207 in South Africa coals (one-tailed t-test, P < 0.05) to the coals in India and Australia could be linked to similarities in their geological settings and compositions (high inertinite and low sulfur) (Chou 2012;Moroeng et al. 2018). Díaz-Somoano et al. (2007) found a correlation between the heavier Pb 206 /Pb 207 and high inertinite and between the lowest Pb 208 /Pb 206 and low-S contents in Spanish coals. Thus, the heaviest Pb 206 /Pb 207 and the lowest Pb 208 /Pb 206 in inertinite high and low-S South Africa coals (Roberts 1988;Bergh et al. 2011) can show organically associated Pb. In addition, some of the Pb can be found associated with clay minerals as the isotope ratio of the coals approximated to coal fly ash (Fig. 5).

Lead emissions
Consumption of leaded gasoline and coal are the principal Pb emission sources. According to Pacyna and Graedel (1995), 70.23% of the total Pb emissions in Africa in 1995 were from leaded gasoline, and only 3.6% were from fossil fuels. Figure 5 shows Pb isotope ratios of different endmembers to overview the primary Pb emission sources of earlier years in South Africa and identify their influence on Pb enrichment in soils. The closeness of Pb isotope ratios of some aerosols, Pb ores, and leaded gasoline clustered in group 1 of Fig. 5 implied leaded gasoline was the prior concern of Pb emission in the past decade in the country. The figure also presents that the Pb in the leaded gasoline was originally from old Pb ore geologically analogous to the Pb ore in South Africa, possibly from Broken Hill ore in Australia (1.04 for Pb 206 /Pb 207 and 2.23 for Pb 208 /Pb 206 ) (Chiaradia et al. 1997). Pb emission from leaded gasoline has decreased in South Africa after the use of leaded gasoline has phased out (Lee et al. 2014), and unleaded gasoline consumption has become prevalent since 2007 (DoE 2017).  (Erel et al. 1997;Farmer et al. 1999;Bollhöfer and Rosman 2000;Monna et al. 2006;Witt et al. 2006;Soderberg and Compton 2007;Díaz-Somoano et al. 2009;Bi et al. 2017) Page 9 of 15 116 Nevertheless, coal utilization for around 75% of the country's energy need has re-ignited the Pb emission problem in the region. Pb emission estimated in this study (Eq. 3) ranged from 168.91 tons in 1971 to 696.89 tons in 2018 (Fig. 6). The average emission in 2011-2018 (38.89-40.59 GW generating capacity CFPPs) was 1.32 times higher, and the average emission in 2001-2010 (38.90-40.59 GW generating capacity) was 1.29 times greater than the averagely emitted during 1990-2000 (547.34 tons). An average of 515.33 tons of Pb was emitted annually and could be more around the Mpumalanga, where most CFPPs are located. Comparing the average emission from the CFPPs to the total Pb emission (675 tons) over the Mpumalanga region (Zhu et al. 2020) showed that CFPPs are the dominant sources of Pb emission. As shown in Fig. 6, low emission was observed before 1992 despite a sharp increase in generating capacity of the CFPPs. However, higher emissions along with a slight increase in generating capacity after 1992 illustrated an increase of emissions with age of the CFPPs.
Pb in coal is a semi-volatile trace metal and thus partly released within combustion residue (Fang et al. 2014). Figure 6 also presents annual Pb (tons) release in the bottom ash, and it depicted relatively higher since recent years (41.71 tons/year during 1971(41.71 tons/year during -1980(41.71 tons/year during and 131 tons Pb/year in 2010(41.71 tons/year during -2018. The higher Pb content in the bottom ash since the last few years might be related to the consumption of poor-quality coal containing high ash content (average 35-45%) for electricity generation (Kalenga et al. 2011;Jeffrey et al. 2014) and clay mineral associated Pb in coals of high ash. Cumulatively, about 24,730 tons of Pb were emitted from 1971 to 2018, which could be the factor for high Pb enrichment in soils around the CFPPs (Okedeyi et al. 2014). The close Pb isotope ratio values of some aerosols (group 2 clustered in Fig. 5) to Pb isotopes of coal fly ash and coal (group 3 in Fig. 5) may indicate that coal combustion contributed Pb emissions was also substantial during the leaded gasoline consumption period. Besides, the close Pb isotope ratios in the soil to the values in coal fly ash and coals compared to the leaded gasoline Pb isotope ratios (Fig. 5) could signify that Pb emission associated with fly ash has a higher potential impact on soil Pb pollution. When the presently estimated were compared with the emissions from 1971 (309 tons) to 2008 (1363 tons) evaluated by Lee et al. (2014), the lower emission was observed as the study ignored the role of PM emission control deployed in the CFPPs (Masekoameng et al. 2010). However, the average emission rate (5.2 g Pb/ton) was close to the rate (5.52 g Pb/ ton) in a broad region that includes South Africa (Zhu et al. 2020). Pb emissions from the CFPPs will increase as long as electricity from coal is dominant, and installing effective emission control such as Wet-Flue-gas desulfurization and PM emission control may significantly support reducing the ongoing challenge (Deng et al. 2014).

Mercury in the coals
Mercury (Hg) content in South Africa coals ranged between 0.16 and 0.3 mg/kg (Wagner and Hlatshwayo 2005;Tewalt et al. 2010;Bergh et al. 2011), and this is relatively higher than the global coals average range (0.02-1 mg/kg) (Mukherjee et al. 2008). Hence, washing the coals before combustion is highly essential to minimize Hg emission. The reducibility of Hg in coals depend on how it is associated with ash and S contents of the coals (Tang et al. 2020), as sulfide-bound Hg can be cleaned more readily than the clay and organic-bound (Kolker et al. 2006;Tang et al. 2020). Figure 7a was plotted from the concentration data in Table S2 to explore the relationship between Hg and S in South Africa coals, and the weak correlation (R = 0.170) could reveal that sulfide bound Hg is not dominant. Low-S coals contain a higher organic S fraction (Smith et al. 1981), and most Hg associated within the lower S coals in Fig. 7a is an indication of organically associated Hg. Hg in coals can also exist in association with ash (Yudovich and Ketris 2005), and the correlation (R = 0.641) between the total Hg and ash (Fig. 7b) depicted clay minerals bound Hg. A similar strong relationship between the Hg and ash (R = 0.948) in the Highveld coals was also given by Kolker et al. (2017).
Pearson correlation between Hg and major elements exhibited Hg strong positive correlation at P < 0.01 with Si, Al, Ca, Fe, Ti, and Mg (Table 1). Hence, the Hg weak correlation with total S that agreed to Wang et al. (2018) while relatively strong with ash and major elements inferred clay minerals associated Hg. Hg strong correlation with Fe, but weak with S, could specify Hg affinity to Fe-bearing minerals such as illite and ankerite (Pinetown et al. 2007).
Stepwise multivariable regression between Hg and the major elements showed that Ca, Na, and Fe contents can influence Hg content and mode of occurrence in the coals (supplementary material part-II), illustrating Hg associated with Ca-and Fe-bearing minerals at low Na concentration. According to Finkelman et al. (2018), the organic form occurrence of Al, Ca, Fe, Mg, and Na, in low-rank comprise 20%, 35%, 5%, 60%, and 65%, respectively. Hence, the strong correlation of Hg with the above elements may inform the presence of organically associated Hg. Lusilao-Makiese et al. (2012) explored Hg mode of occurrence in South Africa coals approximated 37-40% organic-bound. Furthermore, organic and mineral-associated Hg in this study agreed with findings from previous studies of organic and inorganic associated Hg in Waterberg coals (Wagner and Tlotleng 2012), and mineral pyrite and organic S associated in no. 4 seam coals in Witbank coalfield (Bergh et al. 2011). However, Jongwana and Crouch (2012) reported that 80% of the Hg were in the acid extraction showing pyrite-associated Hg followed by organic and carbonate in coal samples from CFPPs.
Hg isotopic composition in coals can reveal coal washing efficiency and Hg affinities with other elements in the coal (Lefticariu et al. 2011), besides its frequent application to identify Hg sources (Yin et al. 2014). The coal samples collected by Sun et al. (2014a) from South Africa during his/her global Hg isotopic composition study were collected from coals prepared for combustion in the CFPPs. Comparing the Hg isotopic composition of the South African coals with Hg isotopic compositions in coals from other regions having the same coal-forming period with that of the South African coals (Permian) ( Table S5) can inform coal washing efficiency and Hg occurrence in the coals. Based on the comparison, South Africa coals ranked comparably the heaviest in δ 202 Hg and the lightest in Δ 199 Hg (Fig. 8a, b). According to Sherman et al. (2011), relatively higher δ 202 Hg appears in unwashed coals than washed, and thus the highest δ 202 Hg in South Africa coals can signify either weak or absence of coal washing of the coals consumed to generate electricity. Fig. 7 The correlation of Hg concentrations (mg/kg) with a total sulfur (%) and b ash yield (%) in South Africa coals (Tewalt et al. 2010;Wagner and Hlatshwayo 2005). THg refers Total Hg Page 11 of 15 116 The conclusion of partial or no coal washing for the coals used in the CFPPs in South Africa was agreed to previous studies (Dabrowski et al. 2008;Kolker et al. 2014Kolker et al. , 2017. The Hg in coals with decayed plants origin shows relatively lower (negative) Δ 199 Hg, whereas those from aquatic water and organisms show the heavier Δ 199 Hg (Sherman et al. 2011;Yin et al. 2014). 199 The lowest Δ 199 Hg in South Africa coals (Fig. 8b) may inform decayed plants related origin Hg composition, and such Hg exists associated with organic components. However, as pyrite associated Hg generally shows the heavier δ 202 Hg that can be cleanable at a higher probability than the δ 202 Hg in organic matter associated Hg (Kolker et al. 2006;Lefticariu et al. 2011), the heaviest δ 202 Hg and the lowest Δ 199 Hg in South Africa coals could have expressed a mixed of pyritic and inorganic associated Hg. A similar investigation was explained by Kolker et al. (2017), R = 0.89 between Hg and pyritic sulfur in coals from Highveld coalfield.

Mercury emissions
Hg is a volatile trace metal in coals (Mukherjee et al. 2008). Due to high temperatures in CFPPs' boilers, a higher percentage of the Hg in coal feedstock is volatilized and subsequently emitted, while very little percentage remains within the solid waste. As the Hg in the coals appears mostly bound with organic and clay minerals (Sect. 4.5), coal combustion by CFPPs promotes more Hg emissions. Hg emission from South African coals calculated based on the average Hg content in coals supplied to the CFPPs (0.27 ± 0.07 mg/ kg) are shown in Fig. 6, and the results were 4.17 tons in 1971 and 17.20 tons in 2018. The emission trend can be partitioned into four phases in relation to the generating capacity of the CFPPs (Fig. 6) (Streets et al. 2018).
The presently estimated Hg emissions from the CFPPs were compared with available previous emission studies in South Africa (Fig. S3) to show the credibility of the currently estimated. The average emission factor evaluated from the current calculation was 0.13 g Hg/ton coal. It was within the previously estimated range (0.02-0.16 g Hg/ton coals) in South Africa by Dabrowski et al. (2008) and globally estimated range (0.1-0.3 g Hg/ton coals) by Pacyna et al. (2006) and Mukherjee et al. (2008). The emission trend from 1971 to 2000 (4.17-14.78 tons Hg with an average of 9.82 tons Hg) shows similar patterns and close in values (4.41-12.75 tons Hg with an average of 8.92 tons Hg) to the estimated by EDGARv4.tox2 (Muntean et al. 2018) (Fig. S3). However, the emissions from 2001 to 2018 of the current estimation were higher than the report by Muntean et al. (2018).  (Sun et al. 2014a;Yin et al. 2014) Besides, the emission range during 2007-2009 (18.86-17.75 tons Hg) was close to the range of 11.1-18.5 tons Hg estimated by Brunke et al. (2012), and the estimation during 2011-2015 (18.61-17.46 tons Hg) was close to the range of 17.19-23 tons Hg from 2011 to 2015 estimated by Garnham and Langerman (2016). The current estimate for 2004 (17.3 tons Hg) was very close to the previously estimated (17.6 tons Hg) by Dabrowski et al. (2008). However, it was substantially lower than the estimated (30.96 tons) in 2004 by Leaner et al. (2009) and 50 tons by Pacyna et al. (2006) due to the fact that the studies used higher emission factor and Hg content in coals than the other studies in the region. Hence, the correspondence and similarity of emission rates from the present study with some of the previously estimated values is an indication of the validity of the current estimation approach and values.

Limitations
In South Africa, limited studies have focused on SO 2 , Pb, and Hg emissions from CFPPs, and this study can provide valuable information on how the emission trends are going on. However, some limitations need to be considered that could have been produced owing to the use of total electricity generated from coal when calculating annually consumed coal by the CFPPs. The lack of data on electricity generated from individual CFPP precludes the estimation of emissions from every CFPP. Furthermore, additional limitations could have also appeared in the estimated emissions due to the fact that the emission factor and S retention used in the current study were collated from other studies reported in the literature. This is owing to the fact that there was no research/ practical investigation on emission factors and S retention for CFPPs in South Africa.

Conclusions
The emissions of SO 2 , Pb, and Hg from CFPPs are among the ongoing environmental and human health challenges in South Africa. Coal washing reduces more efficiently for pyritic associated elements and some inorganic related, whereas pyrite and organic elements can readily vaporize during combustion. However, reports from South Africa indicate that coal washing for the coal-feedstocks used for electricity generation is not a common practice, and this could be a reason for the highest δ 202 Hg in the coals. The relationship of the S, Pb, and Hg in coals with coal ash and major elements, isotopic compositions, and Pb and Hg relationship with S reveal that the South Africa coals contain significant organic and considerable pyritic S fractions, clay and organically bound Pb, and clay and organically associated Hg. Hence coal combustion with such S, Pb, and Hg existential modes could be among the main factors for higher SO 2 , Pb, and Hg emissions from the CFPPs despite their low concentrations in the coals. It means that the existential forms of the S, Pb, and Hg in coals and coal consumption without washing can substantially enhance emissions. Results from the current study showed the average emission rates during 1971-2018 for SO 2 , Pb, and Hg were 6.52 g/ kWh, 3.11 mg/kWh, and 0.076 mg/kWh, respectively. The emissions in [1981][1982][1983][1984][1985][1986][1987][1988][1989][1990] were doubled compared to 1971-1980 due to the surge in generating capacity from 5. . The decrease of emissions since 2011 is likely due to the reduction in electricity generation from coal. However, the reduction in contrast to the slight increase in total generating capacity can show CFPPs efficiency deterioration. It is expected that the use of the lowest-grade coal to generate electricity and the decline of good-quality coal reserves in the country will exacerbate emissions in the future and need attention.