Selenium Migration Mode in Coal Seams: Insights from Multivariate Analysis, Leaching Investigation, and Modelling

Processes controlling selenium concentrations ([Se]) in mine waters were studied at an operating coalmine district in Xuzhou city, China. The geochemistry and mobility of selenium was studied through leaching experiments, multivariate analysis, and numerical modeling. Results showed that selenium leaching was inﬂuenced by selenium occurrence in minerals, pH, electron activity (pe), and sulfur concentration in the water. Selenium occurrence in host rock was mainly sulﬁde minerals, and clay minerals in coal, respectively. Therefore, the oxidation and dissolution of sulﬁde minerals and transformation of clays may control the release of selenium. Experimental leaching experiments suggested selenium tends to leach more when the solution has more sulfur dissolved. A positive relationship is established between pH and the amount of Se released into solution with four times more Se released at pH 12 compared to pH 2 when leached with high-purity water. This release behavior is higher in O 2 -rich environments. The numerical modeling results showed that pH, pe, and sulfur presence in the solution play important roles in selenium adsorption. Selenium was desorbed from adsorbing surfaces under alkaline conditions, speciﬁcally when the solution pH was higher than 8. Higher pe values in the solution caused reduced selenium adsorption. In addition, dissolved sulfur competed with selenate for surfaces of adsorption, thus, selenium adsorption decreases as the sulfur concentration increased.


Introduction
Selenium is both essential and toxic for animals and humans [1,2]. e occurrences of diseases caused by deficiency or toxicity of selenium have been reported in New Zealand, USA, Australia, and China [2][3][4][5][6][7][8][9]. In the target area of the present study-Xuzhou-Datun coal mine district, high selenium concentrations in soil (0.21 μg/mg-4.08 μg/mg) and irrigation water have been reported in an earlier study [10]. Considering the low background of selenium level, the high selenium concentrations observed were believed to have come from coal mine water. By the process of weathering, selenium is released from sulfide minerals or organic matters in coal seams [11][12][13][14][15].
In coal mining processes, mineral-water interaction is extremely important as this is the medium of the release of toxic metals which endangers human health and environmental alike. is interaction process includes a series of equilibrium and kinetic reactions, namely oxidation-reduction, dissolution-precipitation, adsorption-desorption, and ion exchange. ese processes are governed by water or soil parameters such as pH, redox potential/electron activity (pe), organic matter, mineral composition, soil type, microbial activity, and competing anions [12,[16][17][18].
In different surrounding environments, trace elements show various behaviors. e typical metal elements have been found to have high concentrations in acid environments [19,20]. On the other hand, oxyanionic species, like those of certain elements including arsenic (As, AsO 2 − , . . .  6 ] − , . . . etc.), usually have low concentrations in acid coal mine water, but can be found in relatively high concentrations in alkaline water [12,[21][22][23].
Besides the release from solid matrix, adsorption also controls the migration of selenium. Selenium can be preferentially adsorbed by different surfaces according to the pH of the solution [24][25][26][27]. Furthermore, selenium adsorption by iron oxides and clay minerals was up to 100% for pH < 8 and declined rapidly when pH increased [24,[26][27][28]. In addition to the influence of the solution pH, pe also controls selenium adsorption because its valence state changes in different pe environments. us, it has been reported that Se (IV) has higher adsorption capacity onto iron-coated sand [29] and stronger adsorption ability than Se (VI) [30][31][32][33].
Previous studies have made a lot of successful discussions on selenium migration and adsorption behavior in water. However, there are gaps in the researches and understandings: (1) the experiments of coal and coal ash leaching are usually carried out in an ideal condition, and it is needed to investigate the leaching behaviors in real coal mining environment; (2) although coal and coal ash have been researched thoroughly, the leaching behavior of host rock is seldom studied; (3) in the coal mine water, the selenium concentration is controlled by both dissolution and adsorption; most researches have only focused on one of them; (4) most of the researches have only reported a shorttime leaching behavior, a long-term observation is needed to be investigated to demonstrate its full-view leaching behavior.
is study was carried out to explain the high concentrations of selenium in the coal mine water at Xuzhou-Datun coal mine district, although the concentration of this element in the coal seam is low. At the same time, this study is devoted to fill the gap of research and insufficient understanding. In this study, coal and host rock are collected from the active coal mines for the investigation of selenium occurrence and leaching behavior and mechanism. Leaching experiments and modeling were carried out together to investigate migration and adsorption behavior and mechanism on the selenium migration process. As an innovate method, multivariate analysis was used to cluster the elements in solid and liquid samples, then the selenium occurrence and its migration mode could be inferred.

Study Area.
is study was conducted at the Xuzhou-Datun coal mine district located at Xuzhou city, Jiangsu province, eastern China ( Figure 1). e hydrogeology unit is encircled by a series of faults (Figure 1), including coal mines named Longdong Sanhejian, and Yaoqiao. Sediment strata in this area are Simian, Cambrian, middle-lower Ordovician, middle-upper Carboniferous, Permian, Jurassic, Cretaceous, Tertiary, and Quaternary system ( Figure 2). e mining seams (No. 7 and No. 9) are embedded in the Shanxi formation, carboniferous system. e Shanxi formation have an average thickness of 116 m, ranging from 62 to 131 m. Gray mudstone, sand-mudstone, and sandstone are the main rocks in Shanxi formation, while some siliconmudstone and siderite are also present.
Ground water could be found in six aquifers systems, which is shown in Figure 2. Limestone aquifers could be found in the Ordovician system and Taiyuan formation of Carboniferous system, which could be the main sources of the coal seams. Sandstone aquifers could be found in the lower-Shihezi formation and Shanxi formation of Ordovician system. A conglomerate rock aquifer in the Jurassic, and a grit aquifer in the Quaternary are also distributed in the coal mine area.

Sample Collection and Chemical Analysis.
e samples collected for test and leaching experiments include 30 rock/ coal samples and 48 water samples. Coal and rock samples were collected from the working area and sealed using plastic bags immediately. e water samples were collected in 1000 ml Nalgene bottles, which were cleaned by 3.2 mol/L HNO 3 and then washed by sample water. A portable JENCO 6010 pH/ORP meter was used to measure pe and pH values.
Major ions and physical parameters of water samples were determined according to Chinese standard in the Jiangsu Provincial Coal Geology Research Institute. Total dissolved solids and hardness were analyzed using the standard GB/T 8538 method. pH was also analyzed in the laboratory using the Glass electrode method (GB/T 6920-86). Mg and Ca were measured using atomic absorption spectrophotometric (GB 11905-89). A quantity of 1000 mg/L of Ca standard solution and 100 mg/L of Mg standard solution was diluted to 0.1, 0.2, 0.5, 1, 2, and 5 mg/L, the standard curve of Ca and Mg were obtained, while fitting the concentrations points of standard solutions minus concentration in blank solution. K and Na were analyzed by flame atomic absorption spectrophotometry (GB 11904-89). A quantity of 1000 mg/L of K and Na standard solution and 100 mg/L of Mg standard solution were diluted to 0.1, 0.2, 0.5, 1, 2, and 5 mg/L. e absorption line of K was set to 766.5 nm and 404.4 nm, and the absorption line of Na was set to 589.0 nm and 330.2 nm. Fe was measured by phenanthroline spectrophotometry (HJ/T 345─2007). Standard Fe solution was obtained by put weighted 0.7020 g of ammonium iron sulfate (NH 4 ) 2 Fe(SO 4 ) 2 ·6H 2 O into 50 ml of 1 : 1 sulfuric acid and then transferred to 1000 mL of volumetric flask. en the standard solution was diluted to 0.1, 0.2, 0.5, 1, 2, and 5 mg/L to generate standard curve. e wavelength of absorption line of Fe was set to 510 nm. Sulfate and chloride were determined using flame atomic absorption spectrophotometry (GB 13196-91) and silver nitrate titration (GB 11896-89), respectively. e concentrations of the major ions were represented as molality, while the weight concentration was divided by molar mass. e digestion process was designed in accordance with the earlier studies [34][35][36][37][38]. In the first step of the digestion procedure, 0.05 mg of solid samples were weighted and put into tetrafluoroethylene crucibles, then 3 ml of HNO 3 , 1 ml of HF and 1 ml of HClO 4 were added and heated until all liquid consumed up. In the second step, the crucible was added 3 ml of HCl and heated to evaporation. In the third step, high-purity water (18 MΩ·cm, Q-Pod Element, Millipore) was used to rinse the crucible three times and then transferred to a 50-mL volumetric flask, dilute with highpurity water to volume. All the acids are analytically pure. All the reagents, including HNO3, HCl, HF, and HClO4, were analytically pure grade, concentrations of which were 68, 38, 40, and 72%, respectively. e mineralogy of the coal and rock samples was determined using XRD, which was tested using a Philips PW 1830 diffractometer system with Cu Kα radiation and fitted with a PW 1820 goniometer and a graphite monochromator. e concentrations of trace elements in water/coal/rock samples were determined by ICP-AES and ICP-MS. e ICP-AES analysis was carried out using a JY38S ICP-AES model. e limit of detection and deviation for the analysis carried out by such equipment are 0.01 μg/mL and less than 2%, respectively. e ICP-MS analysis was carried out using the X-Series ICP-MS-ermo Electron Co., at the following operating conditions: ICP-MS power: 1200 W; auxiliary flow rate: 0.75 L/min, nebulizer flow rate: 0.80 L/min, coolant argon flow rate: 13.50 L/min. An internal standard of Rh was used to determine limit of detection (0.5 pg/mL) and analytical deviation (less than 2%). e ICP-AES and ICP-MS instruments operated routinely. Before each set of testing, standard line for each trace elements was obtained by measuring the standard samples. All the samples were measured twice for its reproducibility.   Note. e study area is located in the Xuzhou city, Jiangsu province, eastern China. Samples were collected from Sanhejian, Longdong, and Yaoqiao coal mines. Surface water samples were collected from the Zhaoyang Lake which is located to the northeast of the coal mine district.

Leaching Experiments.
e batch mode was used in the leaching experiments to simulate conditions in a coal seam where water movement is slow, following a previous method (with low pO 2 ; see Stumm and Morgan for details [39]). Closed bottles were used to simulate close environment, and a thermostat which was set to 40°C was used to simulate temperature underground in some experimental sets. e pH and pe values were monitored using a JENCO 6010 pH/ ORP meter.
All glassware used in the experiments was previously soaked in 3.2 mol/L HNO 3 for 2 days. Rock/coal samples (Table 1) were ground and sieved to 120 μm. A quantity of 30 mg of solid sample and 1000 mL aliquot of high-purity water with initial pH values of 2, 5.6, 7, and 12 were used in each experimental set. Leachate solutions were collected using syringes at 2, 6, 24, and 48 h, some experiments lasted for up to 10 days. A quantity of 0.5 mol/L HNO3 was added into all collected samples. Selenium concentrations in collected sample are given in Table 2.
In addition to leaching experiments, water samples including those collected from the Zhaoyang and Yunlong Lakes ( Figure 1) were shaken every 2 hours for up to 10 days at 40°C. Results are given in Table 3.

Multivariate Analysis.
e solid samples are composed of minerals and elements, and the liquid samples are composed of ions and trace elements. Concentrations and coexisting behaviors of trace elements suggest the occurrence and leaching mechanisms during the water-rock interaction. As the sample-concentration pattern is a matrixlike system, the technology of the multivariate analysis can be used to identify the different and similar components, which usually suggest different occurrence in solids, and immigration mechanism in waters.
To understand the selenium occurrence in coal and host rock, we applied the method of principal component analysis (PCA) with R language. e data are first standardized using the original data matrix, then divided every value with each column by the column standard deviation. e orthogonal principals are then rotated to achieve the maximum deviation. After PCA, the matrix is reduced to two or three smaller ones that consist of PC loadings and PC scores. In our study, the PC loading means the values that every element have on principal components, and the PC scores means the values every sample have on principal components. e PC loadings suggest similar behavior of elements occurrence in rock and coal matrix, and the PC scores suggest the samples with similar composition.

Geochemical Modeling.
Besides the experimental work to characterize the behavior of leaching selenium in the water-rock/water-coal interaction at Xuzhou-Datun coal mine district, this was also modeled using the following aqueous geochemistry software: PHREEQC (version 2.15.05), Geochemist' Workbench (GWB version 7.0) and MINTEQ (version 2.53). Database Wateq4f.dat were used while modeling using PHREEQC and MINTEQ.
Two dominant processes controlling selenium distribution were modeled: after dissolution-precipitation and adsorption onto iron oxides and hydroxylated silicates.
Modelling consisted of three parts at varying experimental conditions. First, species of Se were calculated; then the adsorption of Se species onto selected surfaces was simulated; and finally, the interaction of the water-rock/ coal-assuming equilibrium-was modeled under closed and open conditions. e speciation of selenium was modeled using the GWB software; assuming a solution volume of 1000 ml with ionic strength of 0.1 mol/L (this last assumption was made based on the analytical results obtained from the Sanhejian coalmine water samples). Se in the coal mine water was set to 6 nmol/L. Other major ions were limited to concentrations found in the water from the Sanhejian coal mine, shown in Table 3. Surface water in the table stands for the water samples that were collected from surface water in the coal mine district, the carbonateseam water means the water samples that were collected from carbonate seams from the coal mine district, and the coal mine water means the water samples that were collected from coal mines underground.
e system temperature was kept at 40°C, same as the coal mine environment.
Both, the adsorption of selenium species, including selenite and selenate and adsorption competition by sulfate were modeled using the PHREEQC and MINTEQ programs. e adsorption model used was the Diffuse Layer/ Surface Complexation model (DL/SC) as suggested in [40,41] and coded in PHREEQC and MINTEQ. e adsorbent was set to be amorphous iron oxides, also called ferrihydrite (HFO). is assumption was based on the fact that although selenium can be adsorbed by iron oxide, clay, carbonate, pyrite, etc., it has both a greater adsorption capacity than clay minerals and a better ability to adsorb trace elements [42]. us, the HFO was used to represent adsorbents. e specific surface of the HFO was assumed to be 600 m 2 /g as recommended [40], of this 2.24e − 3 mol/g and 5.6e − 5 mol/g were considered to be weak and strong sites, respectively.
pe, pH, and some major ions concentrations were varied during modeling. Changes in pe combined with air phase were used to simulate "closed" and "open" system behavior.
us, equilibrium with the air phase meant an open environment, while no air phase signified a closed system. Modeling results showed that open system had a high pe-in the range of 10-15; this meant that selenium adsorption was likely to be impacted by water parameters. In closed system, pe varied from 0 to 4. e system pH was controlled by hypothetical additions of NaOH in the modeling; at the beginning of the model, running pH was set to zero, then as modeling time evolved, NaOH was "added" to increase the pH of the system. During every step, the system was allowed to achieve equilibrium after that PHREEQC was used to calculate the adsorption ratio of selenium.
To find out the impact of ions at different concentrations on the adsorption of selenium species, two sets of 4 International Journal of Chemical Engineering concentrations of dissolved major ions were used in the modeling: high-purity water and coal mine water composition.
e initial settings simulated 0.6 nmol of selenium added to 1000 ml of high-purity water containing 0.1 g HFO of adsorbent. e following assumptions were taken:   International Journal of Chemical Engineering (a) e coal seam system is closed before mining, thus it is isolated from oxygen and any organic rich waters (b) e coal seam system is open during mining, then it exposed to oxygen (c) e redox potential of water surrounding the coal seam changes at open system conditions (d) Selenium will change among its four oxidation states depending on the pH and pe conditions (e) e dominant selenium species are H 2 Se and HSe − and Se precipitates under closed system conditions (f ) e concentration of dissolved selenium in the coal mine water is controlled through adsorption [28] rather than precipitation [21] processes (g) Iron minerals are the most important media for Se adsorption (h) e chemistry of iron mineral is highly influenced by pH and pe conditions (i) e dominant iron mineral in the coal bearing seam at closed system is pyrite, this however will oxidize, dissolve, and reprecipitate as iron oxide and iron hydroxide minerals such as hematite and goethite when the coal bearing seam is at open conditions  4 2− are similar for both carbonate seam and coal mine waters, but lower for surface water. e concentrations of CO 3 2− and NO 3 − are usually low in carbonate seam water but vary for the other two types of waters. e concentration of selenium in coalmine water is the highest of the three types of water under study; this value exceeds the Chinese standard limits for ground and surface waters if destined for drinking water purposes; 0.127 nmol/L and 0.253 nmol/L, respectively. It is also higher than the standard limit-independently of the origin-for irrigation; 0.253 nmol/L. [Se] in the coal mine water is about 0.2-0.9 nmol/L, this implies that this water needs to be treated before being used for either drinking or irrigation purposes. As the coal mine water was discharged into surface water, selenium concentration may be lowered due to adsorption. However, we compared the selenium concentration in surface water both from the coal mine district and non coal-mine district of Xuzhou-Datun area. According to our observation, the selenium concentration in coal mine district surface water was higher (0.11 nmol/L) than that of noncoal mine district surface water (0.0536 nmol/L). Huang et al. [10] reported that the selenium concentration in irrigation water in the Xuzhou area was from 0.03 to 3.67 nmol/L with a medial value of 0.63 nmol/L which is comparable with our research. At the same time, soil in the Xuzhou area was found to be contaminated by selenium. e results suggested that the selenium in coal mine water is an important source of surface water.

Rock Composition, Selenium Occurrence, and Its Influence on Selenium Leaching.
e coal host mineralogy, determined by XRD, is shown in Table 1. As seen, quartz and kaolinite are the main minerals in the coal seam host rock, with the former being the most abundant-52-85%-in three of four locations sampled; only in the roof of the Yaoqiao mine, kaolinite was the most abundant mineral-53%. Because of the low mineral proportion, mineralogy in the coal was not quantified. e relationship of elements in rock and coal samples was calculated by using PCA method. For the rock samples, two PCs were selected based on the screen test, which explained 63% (34% and 29%, respectively) of all deviation, the PC loading is shown in Table 4. e PCA result was clustered using GMM algorithm (shown in the Figure 3(a)), in which the group one includes Zn, Ba, Mn, Fe, Mg, As, Hg, Se, Cd, represented by hollow squares, and the group two includes Mo, Pb, Cr, V, Cu, Ti, Sr, Ca, Al, represented by solid circles, respectively. Considering the element combination behavior and previous studies [43][44][45][46], the RC1 stand for sulfide minerals, especially pyrite. erefore, selenium is mainly occurring in coal host rock as pyrite.
For the coal samples, also two PCs were selected based on the screen test, which explained 69% (45%, 24%, respectively) of all deviations; the PC loading is shown in Table 5. e PCA result was clustered using GMM algorithm (shown in the Figure 3(b)), in which the group one includes Mo, Pb, Cr, V, Cu, Ti, Al, As, Hg, Se, represented by solid circles, the group two includes Zn, Fe, Cd, represented by hollow squares, the group three includes Ba, Mn, Sr, Mg, Ca, represented by solid triangles, respectively. Considering the element combination behavior and previous studies [15,47,48], the RC1 stands for clay minerals. erefore, it is concluded that selenium is mainly occurring in coal as clay minerals.
e roof rock has more pyrite than the floor rock in the samples of both mines; this is relevant because sulfides, such as pyrites, may be carriers of selenium in coal and coal seam host rocks, where they are either replaced by some other elements (such as sulfur) or as selenium minerals. Before and after leaching experiments, illite tends to transform to kaoline, and albite and pyrite tend to consume. erefore, the transformation and dissolution are important mechanism of selenium release. In host rock, the oxidation and dissolution should be the key mechanism of selenium release, and the transformation of clay mineral should be the key mechanism in coal.
In case of sulfite occurrence, selenium would be stable under anaerobic conditions, and dissolve in an oxygen-rich environment. In case of clay occurrence, selenium may be released during the water-rock interaction, especially in an oxygen-rich environment. 6 International Journal of Chemical Engineering To illustrate the effect of anaerobic and oxygen-rich conditions on the dissolution of selenium minerals in the coal mines under study, the speciation modeling of selenium in coal mine water was carried out; results are given in Figure 4. e gray area delimits the typical conditions of pH and pe for the coal mine water. As seen, SeO 3 2− and SeO 4 2− are the main species in that region. If the pH of coal mine water is below 7, the main species would be HSeO 3 − . In an aerobic environment, SeO 4 2− will be more abundant than SeO 3 2− .

Selenium Leaching Behavior and Mechanism.
e geochemistry and mobility of selenium in coal mine water were studied using leaching of coal seam samples with both surface water and high-purity water, at three groups of different conditions; results are shown in Figure 5.
In the first group- Figure 5

pH and Pe Influences on Selenium Leaching and Adsorption.
e pH and pe may influence the selenium leaching in two ways. First, the selenium-contained minerals . Note. Features of the element contents in rock and coal were reduced by PCA method and then clustered using GMM algorithm, shown in (a) for rock and (b) for coal, respectively. In (a), the elements were clustered into two groups, the group one includes Zn, Ba, Mn, Fe, Mg, As, Hg, Se, Cd, represented by hollow squares, the group two includes Mo, Pb, Cr, V, Cu, Ti, Sr, Ca, Al, represented by solid circles, respectively. In (b), the elements were clustered into three groups, the group one includes Mo, Pb, Cr, V, Cu, Ti, Al, As, Hg, Se, represented by solid circles, the group two includes Zn, Fe, Cd, represented by hollow squares, the group three includes Ba, Mn, Sr, Mg, Ca, represented by solid triangles, respectively. As mentioned previously, the pH of the solution controls the leaching behavior of selenium; this effect is clearly depicted in Figure 5(a). As seen, when using high-purity water of pH 2 to leach host rock samples, the [Se]-after 48 hours-is 0.0386 nmol/L in the resulting solution. When using high-purity water of pH 5.6 and 12, [Se] resulted in 0.0796 and 0.154 nmol/L; respectively. us, it can be concluded that selenium is likely to be released as pH increases.
Similar results to the abovementioned were reported by Iwashita et al. [23]. ey researched the leaching of Coal Fly Ash (CFA) and found that the selenium leaching was pHdependent. When the pH solution was under 6, less than 20% selenium leached, while up to 50% leached when the pH solution was over 10.
Canty and Everett [50] also reported the influence of pH on selenium leaching in coal. ey studied the leaching of fluidized bed combustion ash injected into abandoned coal mines. After 24 hours of contact of water with the ash, both the leached solution alkalinity and pH gradually increased, the concentration of metals such as Al, Zn, and Ni decreased, but the concentration of Se was well above the criteria for continuous and maximum concentration levels [51][52][53][54][55]. Figure 6 shows the modeling results of adsorption of selenium at open and closed system with varying pH from 1 to 12. As seen for the open-system case-with pe above 10-selenium is almost 100% adsorbed in the pH region 2 to 6, outside such that the Se adsorption sharply reduced. Similar results were obtained for the closed system-with pe around zero-the only difference was that the pH region affecting Se adsorption was larger covering values from 2 to 8.
ese results show that in open as well as in closed systems, selenium adsorption was in inverse relation to pH. e modeling results are comparable to those obtained by leaching experiments in this research and to previous studies. For example, You et al. [32] reported that the SeO 3 2adsorption on hydroxides with double layer of Mg-Al and Zn-Al declined when pH was above 10 and below 4.5. El-Shafey [28] found that SeO 3 2− adsorption had a peak ratio when pH was equal to 6 and decreased when pH was above 8. Modelling and laboratory results in this study suggest that selenium adsorption from the coal mine water onto the present surfaces will be high at the typical range (7.5-9) found in the mine waters at the coal mines in Xuzhou area, China.
Experiments in this study also showed that selenium concentration in the leached solution increased with pe. After 48 hours of leaching host rock samples with highpurity water (initial pH was 5.6), the concentration of selenium in the leached solution was 0.0796 and 0.0634 nmol/ L, under open-and closed-system regimes, respectively. e pe values were 3.93 for the former and 2.68 for the latter. Schwartz et al. [31] studied the selenium leaching behavior from coal ash, and it is found that [Se] increased sharply and then decreased. Under the aerobic condition, the [Se] increased to 53 μg/L at 24 h and decreased to 29 μg/L at 2 weeks. Meanwhile, the [Se] increased to 94 μg/L at 4 h and decreased to 1.8 μg/L at 2 weeks under the anaerobic condition.
Modeling results also showed that pe had a high impact on selenium adsorption, as seen in Figure 7. is shows the modeling at varying pe from −2.5 to 13.5 and constant pH of 7. Under such conditions, the selenium adsorption ratio on HFO was reduced from nearly 100% to 0. us, selenium  Figure 4: pH-pe diagram of selenium species. Note. Two assumptions were taken to build this diagram; the first is that [Se] in the coal mine water was 0.6 nmol/L, while the second is that temperature was 40°C which is same as the temperature underground. 8 International Journal of Chemical Engineering absorption is nearly 100% up to pe 7, from that value onwards selenium is more likely to be released into solution, from 7 to 9 the selenium adsorption plummeted to negligible levels. ese modeling results are like those obtained in the laboratory. ese pe modeling predicts what will happen to the selenium in the coal seam when mined. Before production, there is an anaerobic environment where pe values are low and selenium is likely to be absorbed into the coal seam. When the coal seam is open to air during mining, pe values will increase and the selenium adsorption will be reduced, thus selenium will be leached from the coal seam. e selenium leached into the solution will adopt a variety of chemical forms-or chemical species-depending on the pH and pe values of the solution, as shown in Figure 4.
is shows that at aerobic conditions, chemical species of Se +6 are more likely to form than Se +4 . e opposite will occur at anaerobic environments. is is relevant because the chemical form of a substance determines its absorption/ mobility behavior as well as toxicity.
In this study, PHREEQC modeling results showed that SeO 3 2− has stronger adsorption ability than SeO 4 2− . At high pe values, the SeO 4 2− /total selenium ratio increases, indicating that the selenium adsorption is reduced. Similar results have been reported in literature; for example, Fernández-Martínez and Charlet [56] found the Se (VI) oxyanion selenate has little tendency to adsorb to solids or to precipitate out of solution compared to selenite Se (IV), which has greater sorption affinity to metal oxyhydroxides, clays, and organic matter. Lopes et al. [57] found that Se sorption is mediated by its oxidation state, as selenite has more adsorption capacity than selenate, with the best affinity for metallic oxides and hydroxides mostly iron, manganese, and aluminum.

Competitive
Adsorption. e effect of ionic strength on selenium leaching in coal seam was studied using two types of surface water samples, one of ionic strength 24.6 mmol/L and sulfur content of 1.994 mmol/L; and the other had ionic strength of 15.6 mmol/L and sulfur concentration of 0.849 mmol/L [58][59][60][61][62].
Results depicted in Figure 5(c) show that the [Se] in the solution obtained from the more concentrated surface water decreased during the first two hours of the experiment, increased slowly for four hours after that, to be increased promptly for the remaining time, reaching a concentration of 0.133 nmol/L after 48 h. In contrast, the [Se] in the solution got from the less concentrated surface water, increased steadily from beginning to end, reaching 0.0903 nmol/L after 48 h.
As seen, the effect of ionic strength on selenium leaching in host rock is better illustrated in the more concentrated solution, where a clear example of competitive adsorption is displayed. Such competition is likely to arise from sulfate and phosphate ions taking the place of selenium on adsorbent surfaces [32], such as clay minerals and HFO.
To study the effect of the dissolved ions on the adsorption of dissolved selenium, two "hypothetical" solutions were designed with major ions compositions similar either to coal mine water or surface water. e hypothetical coal mine water was modeled at both anaerobic and aerobic environments; while the surface water was modeled only considering that it was in contact with air. Under such conditions, for all cases, major ions were separately added and withdrawn from the hypothetical water in question to examine the major ions impact on the selenium adsorption, during these modeling experiments, the ionic strength was kept by addition of nonreactive ions such as nitrate.
Modelling results showed that only dissolved sulfate reduced the selenium adsorption. Figure 8 shows the effect that sulfur has on selenium adsorption. In the hypothetical coal mine water under anaerobic conditions, sulfur starts to compete when sulfur concentrations were above 4 and up to 12 mmol/L, reaching its highest interfering effect at around 8 mmol/L when selenium adsorption is reduced in almost 80%.
Once the dissolved major ions competing with selenium for adsorption was identified and its behavior was estimated; this study focused on identifying the selenium species most affected by sulfur at the existing conditions at the coal mines in Xuzhou district. As mentioned before and shown in Figure 7, pe is around 0 in the anaerobic environment and above 10 in the aerobic environment; as a result, Se +4 species will be more likely to form under the anaerobic environment, while Se +6 species will be more abundant under the aerobic environment. us, the adsorption of SeO 4 2− is quite likely to be reduced by the presence of sulfur under anaerobic conditions; in contrast to the adsorption of SeO 3 2− that is less impacted by the changing concentration of sulfur at aerobic conditions. is study's results on the impact of sulfur on selenium adsorption could be explained by the fact that selenium and sulfur have similar structure and chemical character. Furthermore, similar findings have been reported previously; for example, You et al. [32]  ] ratio was higher than 30, the selenium adsorption ratio reduced from 6 cmol/kg to about 1 cmol/kg. Goh and Lym [30] also reported similar findings to those shown here; these authors studied the competitive adsorption of SO 4 2− on SeO 3 2− and SeO 4 2− and observed that the latter showed a significant decrease with increasing concentrations of SO 4 2− . In contrast, SeO 3 2− did not show a   significant impact. Previously mentioned authors [63][64][65] have adduced these differences could be explained if one considers that SeO 3 2− could be adsorbed on HFO surface by strong specific binding mechanism (inner-sphere complex), while SeO 4 2− may only form weak bonding on the HFO surface (outer-sphere complex) so it could be easily displaced by SO 4 2− . Mo et al. [66] reported that in divalent sulfate anions, Se (VI) sorption is almost completely inhibited at sulfate concentration close to 8.3 mmol/L.

Dissolved Selenium in the Xuzhou-Datun Coal Mine: A
Hypothesis. Based on the analytical and modeling results, the high concentrations of selenium found in the Xuzhou-Datun coal mine water could be briefly explained as follows: e selenium occurs in coal as clay minerals, and sulfide minerals in host rock. ese minerals release selenium during the process of water-rock interaction. Accompanying the mining activity, pH and pe changes, then the release behavior may change either. In a closed environment, pyrite and other sulfide mineral is relatively stable. When the minerals open to air, pe value would rise, then the sulfide mineral reacted with water faster, so the selenium would be released to water. Clay minerals may transform from illite to kaolinite, then the selenium would be released.
In aerobic environments, higher pe and higher sulfur concentrations in water were found, which led to a high concentration of dissolved selenium in coal mine water. us, the oxidation and dissolution of selenium mineral and/or releasing from clays becomes the controlling impact on the selenium leaching. is explains why, despite that this coal mine district has coal seams with low concentrations of selenium, this element is in such high concentrations in the coal mine water. Moreover, dissolved selenium will be found in different chemical species that determine their mobility and adsorption.
In the Xuzhou-Datun coal mines, selenium-bearing minerals associated with coal seam are dominantly of a sulfidic type; however, selenium could also be in clay or organic form. All forms are unstable under aerobic environments.
In addition to pe, pH will have an important impact on selenium minerals dissolution and speciation of dissolved selenium species. Figure 9 shows the selenium pH-pe diagram in the coal mine water in Xuzhou-Datun coal mine district. Typical pH (7.5-9) and pe (3 to 12) values in coal mine water are marked by the gray area in the diagram. As seen, selenium can be dissolved as SeO 4 2− in aerobic environments for pe values above 7 or 8. With increasing pe in the coal mine water, the pyrites, and some selenium minerals such as magnesium selenate, tiemannite, and clausthalite will be oxidized and dissolved.

Conclusions
e concentration of selenium in the coal mine water from the Xuzhou-Datun Coal mines is high and poses a potential risk to pollute the surrounding surface water, though the selenium in coal and host rock is relatively low.
Leaching experiments carried out during this study showed that high values of pH, pe, and ionic strength enhanced selenium leaching in the coal mine water at the Xuzhou-Datun coal mine district. Adsorption and  dissolution processes in the solid phase controlled the behavior of such selenium leaching. e selenium releasing mainly from sulfide minerals from host rock, and clay minerals from coal, respectively.
Modeling carried out in this study estimated that the main selenium species are weakly adsorbed at the typical conditions found in the Xuzhou-Datun coal mine environment. Such adsorption is impacted by pH, pe, and presence of dissolved major ion, especially sulfate. At acid pHs, the selenium adsorption on HFO is strong, and hence its leaching is limited; in contrast, at alkaline pHs, such adsorption is weak enhancing selenium leaching. e effect of alkaline pH will be increased if there is dissolved sulfur in the environment because sulfate will compete with selenium to be absorbed.
Oxidation and dissolution of selenium minerals may be the source of dissolved selenium in the coal mine water at the Xuzhou-Datun coal mine district. Pyrites, and some other selenium bearing minerals such as sphalerite, clausthalite, and tiemannite, may be the main selenium carriers in the coal seam; all of these are unstable at aerobic environments.

Data Availability
e datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest
e authors declare no conflicts of interest.