Distribution and Chemical Speciation of Molybdenum in River and Pond Sediments Affected by Mining Activity in Erdenet City , Mongolia

Rivers and ponds near the Erdenet mine, one of the world’s largest copper-molybdenum mines, exhibit high concentrations of molybdenum (Mo). This study evaluates the distribution and chemical speciation of Mo in surface sediments from ponds and rivers in Erdenet city to elucidate the mobility and solubility of Mo in the surface aquatic environments in the area. The waters and sediments were collected in two shallow ponds connected to the tailing pond and from three rivers flowing through Erdenet city. The distribution and chemical speciation of Mo in the sediments were examined using five-step sequential extraction and X-ray absorption fine structure (XAFS) analyses. The XAFS spectra of the sediments showed that large amounts of Mo in the sediments are molybdate or polymeric molybdate, weakly adsorbed onto ferrihydrite. Sequential extraction consistently showed a large amount of Mo distributed in the labile fractions. Results suggest that the surface sediments from ponds and rivers play a role as secondary contamination sources of Mo rather than as sinks of Mo in the area.


Introduction
Mining activities are constantly contaminating surrounding aquatic environments such as rivers, ponds, and lakes by deposition of heavy metals and metalloids [1][2][3].In Mongolia, the most important part of the economy is the mining sector.The International Monetary Fund (IMF) has estimated that about 71% of national income is earned from this sector [4].One can readily expect that Mongolia has many active mine blocks.At Erdenet mine (Erdenet city, Mongolia), one of the largest Cu-Mo mines in the world [5][6][7], mining wastes are disposed in a huge tailing pond that lies close to the Khangal River, flowing into Erdenet city [6].Recent reports have described that the rivers affected by the seepage from this tailing pond contain high concentrations of molybdenum [5,6].
Molybdenum (Mo) is an essential element for all living organisms with trace concentrations in natural surface waters ranging from 0.5 µg/L in streams to 10 µg/L in seawater [8].However, the excesses in environmental Mo can cause functional enzymatic disequilibria [9].The WHO drinking water guideline for Mo is less than 70 µg/L [10].Recently, Smeldley and Kinniburgh [11] and Frascoli and Hudson-Edwards [12] review the Mo concentrations in mining-affected water, soils and sediments.The highest aqueous Mo concentrations are between 10,400 and 13,900 µg/L from acid to alkaline mine drainage water at the San Telmo mine in the Spanish Iberian Pyrite Belt and at the Antamina mine in Peru [13,14].The reported highest Mo concentrations in tailings impoundment channel waters are 2670-3900 µg/L from Cu porphyry mines in the Machali Cachapoal Provcince, Chile [15].The Mo concentrations in the mining-affected river water are <0.3 µg/L from the Laver Cu-Ag-Au mine in Sweden and 0.6-9.7 µg/L from the Gyama Cu-polymetallic mine in central Tibet [16,17].The Mo contents in stream sediments and non-contaminated soils are typically around 10 mg/kg or less [11,12].The highest Mo concentrations in the mine tailings are 4000 mg/kg from the Gyama Cu-polymetallic mine in central Tibet [17].The highest Mo concentrations of the sediments is 1950 mg/kg from Surcheshmeh mine, Iran [18].
The dissolved heavy metals and metalloids in natural water commonly interact with minerals in soils and sediments [2,19,20].The concentrations of the elements in natural water are usually controlled by interaction with minerals [8,21].Therefore, understanding how Mo is sequestered by sediments and soils is crucially important for evaluating its mobility and solubility of Mo in natural water.The respective mobilities and solubilities of trace elements in soils and sediments depend mainly on the chemical speciation of the elements [22,23].Strong chemical bonding of elements to mineral surfaces usually does not result in the release of the elements to natural water, while weak electrostatic bonding engenders the release of elements into the natural water [23,24].Elements hosted in primary insoluble mineral phases or secondary insoluble phases are only slightly released to natural water, although those into soluble minerals are easily released during dissolution of the host phases [22,25].
Two important methodologies have been used to assess the chemical speciation of elements in natural solid samples.The first is the traditional sequential chemical extraction method [26,27].Sequential extraction can show the quantitative partitioning of trace elements in constituents of soils and sediments.However, extraction reagents can extract not only the target materials but also the unexpected mineral phases.Favas et al. [28] applied the sequential extraction technique to the tailings and the soils from the Ervedosa mine in northeastern Portugal.They observed that most of Mo is associated with the easily-to moderately-reducible fractions and interpreted that metals are bound to Fe, Al and Mn (oxy)hydroxides and ferric hydroxysulfates.The second method is sophisticated spectroscopy such as X-ray absorption fine structure spectroscopy (XAFS) [22,29].This methodology enables the direct observation of the chemical speciation of target elements.However, applications of spectroscopic methods to the elements present in very trace amounts in the soils and sediments are difficult because of the method's high detection limits.Blanchard et al. analyzed the mine tailings of Mo from uranium mine in norther Saskatchewan, Canada, by means of XAFS [30].They found that the tailings consisted primarily of Mo(VI) species: powellite (CaMoO 4 ), ferrimolybdeite (Fe 2 (MoO 4 ) 3 •8H 2 O), and molybdate adsorbed on ferrihydrite as well as a minor concentrations of a Mo(IV) species in the form of molybdenite (MoS 2 ).They also showed that Mo concentrations in the porewater in the tailing sediments are controlled by the mineralogical compositions containing Mo.
The combination of the two methodologies must engender a better understanding of the chemical speciation of trace elements in natural samples.By application of the sequential extraction procedure and the X-ray absorption fine structure spectroscopy, this study aims to identify the chemical speciation of Mo present in surface sediment from rivers and ponds in Erdenet city, which is affected by mining activity.

Studied Area
The study area is the third-largest city in Mongolia: Erdenet city in the northern part of Mongolia (Figure 1a) lies in a valley between the Selenge and Orkhon rivers.The Erdenet mine, fourth largest copper-molybdenum (Cu-Mo) mine in the world, has been active since 1978 [5][6][7].The Erdenet mine possesses porphyry Cu-Mo deposit, developed in a continent-continent arc collision zone, within the central Asian orogenic belt in the early to middle Triassic.The Erdenet ore-bearing porphyry complex was preceded by multistage gabbro-granitoid magmatism [7].Molybdenite occurs in monomineralic veins and in finely laminated with massive quartz-molybdenite veins [31].
The vast tailing pond is located in north of Erdenet city (Figure 1b).The tailing dam repositories are a 5 km long dyke of standard design, of which 3 km are covered with water and 2 km are exposed tailing beaches.The dam contains 400 million tons of mine tailings, as well as 15 million m 3 of supernatant water [4].The mining slags have been transported with wastewater to the tailing pond through the pipeline from the refinery factory.The slags have been settled down in the tailing pond.Some of the wastewater is evaporated from the surface of the tailing pond, and some of it is pumped to the refinery factory for re-use.Moreover, some of the wastewater is penetrated the bottom of the pond, and makes runoff through the sub-surface, and released as seepage in a southward of the pond, where is close to the Khangal River.The mining company had built secondary ponds to interrupt the flow of seepage water to the Khangal River.However, the seepage water, which is temporary stored in the secondary ponds, flows to the Khangal River, continuously.There are three rivers in Erdenet city: Erdenet River, Gavil River and Khangal River (Figure 1b,c).The Erdenet River flows from the west of the city.Gavil River flows from the north of the city.These two tributaries join to become the Khangal River, which flows to the east of the city.Their junction point is surrounded by a densely populated ger district, an agricultural area, and the tailing pond.

Studied Area
The study area is the third-largest city in Mongolia: Erdenet city in the northern part of Mongolia (Figure 1a) lies in a valley between the Selenge and Orkhon rivers.The Erdenet mine, fourth largest copper-molybdenum (Cu-Mo) mine in the world, has been active since 1978 [5][6][7].The Erdenet mine possesses porphyry Cu-Mo deposit, developed in a continent-continent arc collision zone, within the central Asian orogenic belt in the early to middle Triassic.The Erdenet ore-bearing porphyry complex was preceded by multistage gabbro-granitoid magmatism [7].Molybdenite occurs in monomineralic veins and in finely laminated with massive quartz-molybdenite veins [31].
The vast tailing pond is located in north of Erdenet city (Figure 1b).The tailing dam repositories are a 5 km long dyke of standard design, of which 3 km are covered with water and 2 km are exposed tailing beaches.The dam contains 400 million tons of mine tailings, as well as 15 million m 3 of supernatant water [4].The mining slags have been transported with wastewater to the tailing pond through the pipeline from the refinery factory.The slags have been settled down in the tailing pond.Some of the wastewater is evaporated from the surface of the tailing pond, and some of it is pumped to the refinery factory for re-use.Moreover, some of the wastewater is penetrated the bottom of the pond, and makes runoff through the sub-surface, and released as seepage in a southward of the pond, where is close to the Khangal River.The mining company had built secondary ponds to interrupt the flow of seepage water to the Khangal River.However, the seepage water, which is temporary stored in the secondary ponds, flows to the Khangal River, continuously.There are three rivers in Erdenet city: Erdenet River, Gavil River and Khangal River (Figure 1b,c).The Erdenet River flows from the west of the city.Gavil River flows from the north of the city.These two tributaries join to become the Khangal River, which flows to the east of the city.Their junction point is surrounded by a densely populated ger district, an agricultural area, and the tailing pond.

Sampling Sites
The surface water (W) and sediment (S) samples were collected from the ponds and rivers in March 2017.Two pond samples, WS10.1 (49 • 04 41.21" N, 104 • 09 48.03" E) and WS10.2 (49 • 04 48.86" N, 104 • 09 30.45"E), were collected from the two secondary ponds connected to the big tailing ponds.As shown in Figure 1b, the seepage water flows between the tailing pond and the secondary ponds.The three river samples were collected at the junction point (Figure 1c).They are downstream of the Gavil River (WS7: 49 01" E).The water and sediments were separately collected at same locations by using a plastic ladle with adjustable rods from the bank of the rivers or the border of ponds.We collected the surface sediments that are expected to be contacted with the surface water in the shallow ponds and rivers.The volume of the plastic ladle is 500 mL.The depth of collected sediments were no deeper than 10 cm from the surface of the sediments.The obtained sediment samples were stored in the plastic bags with zips.The remained waters in the plastic bags were removed as much as possible.

Water Analysis
Temperature, pH, electrical conductivity (EC), dissolved oxygen (DO), and oxidation reduction potential (ORP) were measured on site using EC, pH, and ORP meters with electrodes (portable electrical conductivity pH meters, WM-32EP; TOA-DKK Corp., Tokyo, Japan) and a DO meter (DO-3IP; TOA-DKK Corp., Tokyo, Japan).Before the measurements, the pH electrode was calibrated with three pH buffer solutions (pH 4.01, 6.86 and 9.18 at 25 • C).Alkalinity of filtered solution with 0.45 µm membrane was measured by titration with sulfuric acid (H 2 SO 4 ).After the water samples were filtered using a 0.45 µm membrane cellulose filter to remove suspended particles, they were stored in two 50 mL plastic bottles that had been washed with diluted nitric acid and rinsed with deionized water before use.A small amount of ultra-pure grade nitric acid (HNO 3 ) was added to one bottle to make 0.6% HNO 3 for measurements of metals and silicon.Another bottle was left for major anion measurements.
The Mo and Cu concentrations in sample water were analyzed using an inductively coupled plasma-mass spectrometer (ICP-MS: X7; Thermo Inc., Waltham, MA, USA).The detection limit of Cu and Mo in this study was 100 ng/L.Metal concentrations for Na, K, Ca, Mg, Fe, Mn and Si were analyzed using inductively coupled plasma-optical emission spectroscopy (ICP-OES: ES-710; Varian Inc., Palo Alto, CA, USA).Anions for Cl − , NO 2 − , NO 3 − , Br − and SO 4 2− were analyzed using high-performance liquid chromatography (8020 Series; Tosoh Corp., Tokyo, Japan).

Sediment Analysis
Equipment and supplies to collect sediments were cleaned by the deionized water and rinsed by site water at least three times prior to use.The sediment samples in the air-tight plastic bags were stored in refrigerator (<10 • C) in laboratory at National University of Mongolia for two weeks before transport to Japan.After the transportation to Japan, the sediment samples were again stored in refrigerator (2-4 • C) in laboratory at Kanazawa University.The sediment samples were freeze dried (FDU-1200; EYELA Inc., Tokyo, Japan) in the laboratory after removal of stone and plant fragments.The dried sediments were sieved through a 64 µm sieve.The obtained samples were stored in desiccator and used for chemical and mineralogical analyses.

Sequential Extraction
All reagents used were analytical grade.HNO 3 used as a matrix of the ICP-MS and ICP-OES measurements was ultra-pure grade.Five-step sequential extraction was applied to elucidate the partitioning of Mo in the sediments (Table 1 was based on the modified BCR (Community Bureau of Reference) extraction procedure for the analyses of heavy metals in soil proposed by Žemberyová et al. [27].On the other hand, there is no extraction step specifically for the low crystalline oxides in the modified BCR extraction procedure.It is well documented that the low crystalline Fe and Mn oxides are good adsorbents for Mo [32].Therefore, present study modified the method to include the extraction step for low crystalline oxides by using pH 3 oxalate solution [33].At the first step in the modified BCR procedure, 0.11 mol/L acetic acid is used for the dissolution of carbonates and exchangeable metals.However, the pH of 0.11 mol/L acetic acid is less than 3, which results in the dissolution of low crystalline iron oxides [33].The sodium acetate solution adjusted to pH 5 can only dissolve carbonate but low crystalline iron oxides [26].Therefore, we used pH 5 sodium acetate solution instead of pH 3 acetic acid.Then we add the pH 3 oxalate solution for extraction of low crystalline oxide prior to the hydroxylamine hydrochloride in the modified BCR procedure.
The target of the first step (F1) was Mo partitioning in exchangeable (weakly adsorbed) and carbonate (soluble at pH 5) from the bulk sample.50 mg sediment powder samples were treated at room temperature for 15 h with 20 mL of 1 mol/L sodium acetate adjusted at pH 5 using acetic acid on mixed-rotor [26].The target of the second step (F2) was Mo partitioning in poorly crystalline iron and manganese oxides (soluble at pH 3 under reducing condition).The residue from F1 was extracted with 20 mL of mixed solution of 0.2 M ammonium oxalate solution adjusted at pH 3 by oxalate for 4 h at room temperature on mix-rotor [2,26,33].The target of the third step (F3) was Mo partitioning in crystalline Fe and Mn oxide (soluble at pH 2 under strongly reducing condition).The residue from F2 was treated with 20 mL of 0.5 mol/L hydroxylamine hydrochloride adjusted to pH 2 by 2 mol/L HNO 3 at room temperature for 16 h on mixed-rotor [27,34].The target of the fourth step (F4) was Mo partitioning in organic matter and sulfide minerals (soluble at highly oxidizing condition).The residue from F3 was treated further for 1 h at room temperature with 10 mL of 8.8 mol/L hydrogen peroxide on mixed-rotor [27,34].The vessels were heated at 85 • C until the volume decreased to less than 3 mL.After the second 10 mL of hydrogen peroxide was added to the vessels and heated again to dry at 85 • C, 20 mL of 1 mol/L ammonium acetate adjusted to pH 2 with HNO 3 was added to the residue for 16 h on mixed rotor.The target of the fifth step (F5) was Mo partitioning in the residues.The residue from F4 was digested completely in 3 mL of 48% HF with 3 mL of 60% of HNO 3 and was heated at 120 • C for 24 h.Then 3 mL of 30% hydrochloric acid (HCl) was added to the vessels and heated at 120 • C for 24 h to dry.The residues in vessels were extracted using 0.6% HNO 3 .The obtained extracted solutions were diluted adequately using 0.6% HNO 3 solutions and were analyzed using ICP-MS for Mo and ICP-OES for Mg, Ca, Fe, and Mn.We conducted the sequential extraction for the duplicated samples to check the reproducibility.

X-ray Diffraction
X-ray diffraction (XRD) patterns for the sieved sediment samples were obtained using an X-ray diffractometer (Ultima IV; Rigaku Corp., Tokyo Japan) with monochromatic Cu Kα radiation operated at 30 kV and 20 mA from 2 • to 65 • .The step size was 0.02 • .The scan speed was 1

X-ray Absorption Fine Structure
Molybdenum K-edge XAFS spectra were obtained in the beamline BL01B1 of SPring-8 (Hyogo, Japan).The white beam from a bending magnet was monochromatized using a Si(111) double-crystal monochromator.The K-edge XAFS spectra for the natural samples were measured in fluorescence mode using a 19-element Ge solid-state detector, although those for reference materials were measured in transmission mode.The measurements were conducted under ambient conditions.The X-ray energy was calibrated with the first peak of Mo metal foil at 20.004 keV.The MoO Extended X-ray absorption fine structure (EXAFS) analysis was performed using software (REX2000; Rigaku Corp., Tokyo, Japan) with theoretical backscattering amplitudes and phase-shift functions obtained using FEFF 8.5 [35].The Fourier transformation of the k 3 χ(k) oscillation from k space to R space was performed in the range of 3.0-13.25Å −1 to obtain radial structural functions (RSFs) for Mo.The quality of fit was given by the goodness of fit parameter.Details of the experiments and fitting procedure were presented by Kashiwabara et al. [32].

Water Chemistry
Table 2 shows the water chemistries from the rivers (W4, Erdenet River; W7, Gavil River and W8, Khangal River) and pond (W10.1 and W10.2) samples.The water chemistry of the Erdenet River (W4) resembled that of the Khangal River (W8) but differed from that of the Gavil River (W7).Although the Erdenet River and Gavil River join to form the Khangal River (Figure 1c), the contribution of the Gavil River to the Khangal River might be minor compared to that of the Erdenet River.The Erdenet River (W4) and the Khangal River were characterized as having neutral pH and low EC values.The Ca 2+ and SO

XRD
The crystalline mineral assemblages are quartz, calcite, chlorite, mica, plagioclase, and amphibole (Figure A1).Minerals with high intensities are quartz and plagioclase.All samples contain quartz, plagioclase, mica, and chlorite.S4, S7 and S10.1 also contain amphibole, although S8 and S10.2 does not contain it.Calcite was detected from S4, S7, and S10.1 but not from S8 or S10.2.Peak intensities of quartz of S10.2 were found to be the highest among the samples.

Sequential Extraction
Analytical chemical extraction has been applied to understand the partitioning of Mo into five extractions: exchangeable (weakly adsorbed) and Ca and Mg carbonates (F1), amorphous Fe and Mn oxides (F2), reducible Fe and Mn oxides (F3), oxidizable (organic matter and sulfides) (F4) and residual (F5).Ca, Mg, Fe, and Mn are the important elements which constitute the target minerals in F1, F2, and F3.These elements were measured in each extraction.We conducted the duplicate sequential extraction analyses.Both results are similar (Table A1).Therefore, the elements concentrations are given as the average of two measurements.
Highest total Mo content (1100 mg/kg) was observed from the pond sediment sample S10.2 (Figure 2a and Table A1).The Mo in S10.2 dominated in F1 (600 mg/kg), followed by F2 (200 mg/kg), F4 (160 mg/kg) and F5 (50 mg/kg).The contributions from F3 were negligible.The Mo contents (70 mg/kg) of S10.1 are significantly lower than S10.2, although S10.1 also came from the pond sediment connected to the vast tailing pond.The Mo in S10.1 dominated in F2 (50 mg/kg), following by F5 (20 mg/kg) and F1 (7 mg/kg).Contributions from F3 and F4 were negligible.The Mo contents and the distributed fraction of S7 (Gavil River sediment) was comparable to S8 (Khangal River).The total Mo contents were around 30 mg/kg.The Mo mainly distributed into F1, F2 and F5.The lowest Mo content in the Erdenet River sediment was less than detection limits from each step (<4 mg/kg).Among all samples except for S4, most Mo is distributed in F1, F2, and F5.Mo in F3 and F4 are minor, except for F4 from S10.2.
Among the sediments, the distributions of Fe and Mg dominated in F5 (Figure 2b,c, and Table A1).Fe in F5 fraction was well correlated with Mg.Mica, chlorite and amphibole detected from the XRD patterns usually contain Fe and Mg in their structures [36,37].Fe and Mg in F5 are attributable to the clay minerals and amphibole.However, the second important Fe was in F2, and Mg was in F1.The Fe in F2 is most likely attributable to low crystalline Fe oxides (ferrihydrite), which cannot be detected by XRD because of their low crystallinity [38].Fe in F2 ranges from 0.2 to 1.3 wt % among samples.S10.1 sediment exhibited the highest Fe concentration (1.3 wt %) in F2.Calcite commonly contains Mg in the Ca sites in the structure [39].Therefore, Mg in F1 must be attributable to calcite.The extracted behaviors of Ca differed from Fe and Mg (Figure 2d and Table A1).The contributions of F1 in Ca were comparable to those in F5, although those of other fractions (F2, F3, and F4) were minor.S7 and S10.1 exhibit the higher Ca concentrations in F1 at around 3 wt %, although S10.2 exhibited the lowest concentrations at around 0.2 wt %.The calcite peaks were not detected from XRD patterns of S8 and S10.2 (Figure A1).The variation in Ca amount from F1 among the samples is consistent with the abundances of calcite in XRD analyses.The total amounts of Mn were lower than other major elements (Fe, Ca and Mg), with ranges from 0.08 wt % to 0.3 wt % (Figure 2e and Table A1).Mn dominates in F2 and F5, but the contribution of F1 was also important.F2 was more important for the river sediments (S4, S7 and S8).The total measured major element (Mg, Ca and Fe) concentrations of S10.2 were markedly lower than those of other samples, which suggests that the chemical composition of S10.2 dominates SiO 2 as was supported from XRD result.and F4) were minor.S7 and S10.1 exhibit the higher Ca concentrations in F1 at around 3 wt %, although S10.2 exhibited the lowest concentrations at around 0.2 wt %.The calcite peaks were not detected from XRD patterns of S8 and S10.2 (Figure A1).The variation in Ca amount from F1 among the samples is consistent with the abundances of calcite in XRD analyses.The total amounts of Mn were lower than other major elements (Fe, Ca and Mg), with ranges from 0.08 wt % to 0.3 wt % (Figure 2e and Table A1).Mn dominates in F2 and F5, but the contribution of F1 was also important.F2 was more important for the river sediments (S4, S7 and S8).The total measured major element (Mg, Ca and Fe) concentrations of S10.2 were markedly lower than those of other samples, which suggests that the chemical composition of S10.2 dominates SiO2 as was supported from XRD result.).Although the specific mineral phases containing Mo cannot be ascertained solely from this information, molybdate and heptamolybdate can be formed by the dissolution of Mo into water, which can subsequently be adsorbed or incorporated into secondary phases during solid-water interactions.In particular, molybdate ion is adsorbed on ferrihydrite by forming outer-sphere complex, which leads to the similar spectra of hydrated molybdate ion and that adsorbed on ferrihydrite [32].MoO2 species can be present in the solid phase, which is not related to the solid-water interactions, in particular for the samples formed under oxic conditions for the samples in this study.).Although the specific mineral phases containing Mo cannot be ascertained solely from this information, molybdate and heptamolybdate can be formed by the dissolution of Mo into water, which can subsequently be adsorbed or incorporated into secondary phases during solid-water interactions.In particular, molybdate ion is adsorbed on ferrihydrite by forming outer-sphere complex, which leads to the similar spectra of hydrated molybdate ion and that Minerals 2018, 8, 288 9 of 16 adsorbed on ferrihydrite [32].MoO 2 species can be present in the solid phase, which is not related to the solid-water interactions, in particular for the samples formed under oxic conditions for the samples in this study.

XAFS
(A) (B) Figure 4 portrays XANES spectra of S8 (a), S10.1 (b) and S10.2 (c) with linear combination fitting (LCF) to the reference materials.High-quality XANES spectra were not obtained from other samples.Good fitting to the spectra of sediment samples was obtained by LCFs of three reference spectra of molybdate adsorbed on ferrihydrite, MoO2 and heptamolybdate, which are inferred from EXAFS analyses.The similar XANES spectra between aqueous MoO4 2− and that adsorbed on ferrihydrite, suggested that the contribution of molybdate adsorbed on ferrihydrite estimated by the LCF included aqueous MoO4 2− species and other adsorbed onto mineral phase.Similarity of XANES spectra was also found for aqueous heptamolybdate ion and hexaammonium heptamolybdate tetrahydrate (solid material), suggesting that successful fitting indicates the presence of a similar compound (e.g., outer sphere complex) in the solid phase for heptamolybdate species.The LCF showed that the most important Mo species in S10.2 was molybdate adsorbed on ferrihydrite (40%) followed in order by heptamolybdate (35%) and MoO2 (25%) (Table 3).S8 and S10.1 can be fitted by considering the same three Mo species.The most important species of S8 is molybdate adsorbed on ferrihydrite (48%), followed in order by MoO2 (24%) and heptamolybdate (28%), while contributions of 50%, 37% and 13% of the three species were estimated to S10.1, respectively (Table 3).Good fitting to the spectra of sediment samples was obtained by LCFs of three reference spectra of molybdate adsorbed on ferrihydrite, MoO 2 and heptamolybdate, which are inferred from EXAFS analyses.The similar XANES spectra between aqueous MoO 4 2− and that adsorbed on ferrihydrite, suggested that the contribution of molybdate adsorbed on ferrihydrite estimated by the LCF included aqueous MoO 4 2− species and other adsorbed onto mineral phase.Similarity of XANES spectra was also found for aqueous heptamolybdate ion and hexaammonium heptamolybdate tetrahydrate (solid material), suggesting that successful fitting indicates the presence of a similar compound (e.g., outer sphere complex) in the solid phase for heptamolybdate species.The LCF showed that the most important Mo species in S10.2 was molybdate adsorbed on ferrihydrite (40%) followed in order by heptamolybdate (35%) and MoO 2 (25%) (Table 3).S8 and S10.1 can be fitted by considering the same three Mo species.The most important species of S8 is molybdate adsorbed on ferrihydrite (48%), followed in order by MoO 2 (24%) and heptamolybdate (28%), while contributions of 50%, 37% and 13% of the three species were estimated to S10.1, respectively (Table 3).

Characteristics of Water Chemistry in the Ponds and Rivers
All the water was characterized as Ca-SO4 type (Table 2).High concentrations of SO4 are most likely attributable to the oxidative dissolution of sulfides in the mining area.The very high concentration of SO4 in the secondary ponds was attributable to the dissolution of the waste rock dumped in the huge tailing pond.However, the Erdenet River, flowing from the upstream area, also exhibits high concentrations of SO4, which suggests that the SO4 production from sulfides occur not only from mining activity but also from natural processes in this area.The oxidative dissolution of sulfides usually engenders acid production [44].The mining slags have been transported with wastewater to the tailing pond through the pipeline from the refinery factory.Prior to transport the wastewater through the pipeline, lime has been added to neutralize the acidity.High concentrations of Ca in the secondary ponds most likely derive from the lime treatment.The river waters also exhibit high Ca concentrations, which might be attributable to the natural neutralization of sulfuric acid by reaction with the calcite present in this area (Figure A1).
Concentrations of Cu in the river and pond waters were lower than 10 μg/L, although the Erdenet mine produces Cu (Table 2).The lower concentrations of Cu are related to the artificially or naturally neutralized pH.The dissolved Cu strongly adsorbs to mineral surfaces, especially for low crystalline iron oxides at neutral to alkaline conditions [45].By contrast, the Mo concentrations in

Characteristics of Water Chemistry in the Ponds and Rivers
All the water was characterized as Ca-SO 4 type (Table 2).High concentrations of SO 4 are most likely attributable to the oxidative dissolution of sulfides in the mining area.The very high concentration of SO 4 in the secondary ponds was attributable to the dissolution of the waste rock dumped in the huge tailing pond.However, the Erdenet River, flowing from the upstream area, also exhibits high concentrations of SO 4 , which suggests that the SO 4 production from sulfides occur not only from mining activity but also from natural processes in this area.The oxidative dissolution of sulfides usually engenders acid production [44].The mining slags have been transported with wastewater to the tailing pond through the pipeline from the refinery factory.Prior to transport the wastewater through the pipeline, lime has been added to neutralize the acidity.High concentrations of Ca in the secondary ponds most likely derive from the lime treatment.The river waters also exhibit high Ca concentrations, which might be attributable to the natural neutralization of sulfuric acid by reaction with the calcite present in this area (Figure A1).
Concentrations of Cu in the river and pond waters were lower than 10 µg/L, although the Erdenet mine produces Cu (Table 2).The lower concentrations of Cu are related to the artificially or naturally neutralized pH.The dissolved Cu strongly adsorbs to mineral surfaces, especially for low crystalline iron oxides at neutral to alkaline conditions [45].By contrast, the Mo concentrations in the river and ponds water were significantly higher than the average level in streams (0.5 µg/L, [8]).The Mo concentrations of the secondary ponds exceed the level (70 µg/L) stipulated in WHO guidelines [10].The high Mo concentration must be attributable to speciation of the Mo in the sediments, as discussed in the next section.

Chemical Speciation of Mo in Sediments
Results from sequential extraction revealed that most Mo is distributed in F1, F2, and F5.F5 is a fraction from total digestion after a four-step sequential procedure (Figure 2 and Table A1).It can be considered that Mo in F5 is immobile like the inside the matrix of the primary minerals or sulfides which remained after H 2 O 2 treatment (F4).XANES and EXAFS spectra of all samples suggest the presence of MoO 2 .Mo in MoO 2 takes tetrahedral coordination, which is the same as Si in SiO 2 and Al in feldspar [46] and mica [36].The MoO 2 might represent the form of Mo(IV) in the primary minerals.
A large amount of Mo in S10.2 is distributed in the F1 (Figure A2).The contributions of Mo in F1 of S7 and S8 samples were found to be around 30% of total Mo.The target of F1 is the Mo partitioning in exchangeable (weakly adsorbed) and calcium carbonates.Mo exists as molybdate anion in oxidizing surface conditions [47,48].In neutral pH and oxic conditions, the chemical form of Mo(VI) is MoO 42− [47,48].XANES analyses of all samples suggested the presence of molybdate adsorbed on ferrihydrite.Fe and Mn oxides have been documented as excellent scavengers of MoO 4 2− [32,48,49].
Molybdate adsorbs to Mn oxides as inner sphere complex in a form of distorted Mo(OH) 6 complex, which is direct chemical bond sharing the surface oxygen [32].On the other hand, it adsorbs to Fe oxides (ferrihydrite) as an outer sphere complex, which is a weak bonding form via electrostatic attraction [32].Sequential extraction revealed that Mn from oxide forms (F2 and F3) are 40-900 mg/kg (Figure 2e and Table A1), whereas Fe from F2 and F3 are 3000-14,000 mg/kg (Figure 2d and Table A1).
The amounts of Mn oxides in the sediments were markedly lower than Fe oxides in the sediments.XANES analyses suggest the presences of polymeric form molybdate (heptamolybdate).Spectroscopic studies have suggested that molybdate adsorbs to mineral surfaces not only as monomer, but also polymeric forms [49,50].The heptamolybdate observed from EXAFS and XANES analyses might correspond to the polymeric adsorbed species.F2 was an important Mo fraction from the sequential extraction.The target of F2 was Mo partitioning in poorly crystalline iron and manganese oxides.The reagent of F2 was oxalate solution at pH 3, which enables the dissolution of poorly crystalline iron and manganese oxides.A significant amount of Mo is expected to be present as the outer spherically adsorbed form to ferrihydrite.The reagent of F1 was 1 M acetate at pH 5. The high concentration of acetate can be expected to replace the weakly adsorbed species.However, MoO 4 2− is an anionic species.The adsorption of MoO 4 2− on ferrihydrite is favored at lower pH conditions at around pH 5 [32,48].Even in the presence of high concentration of acetate, some MoO 4 2 might remain in the ferrihydrite surfaces.Moreover, ferrihydrite occurred as nanoparticles that aggregate in the natural water [51].The adsorbed MoO 4 2− present inside the aggregates might only slightly access the external solutions in F1 treatment.We inferred that the Mo in F2 might also be attributable to the adsorbed MoO 4 2− in the ferrihydrite.

Solubility and Mobility of Mo in Erdenet Area
The XAFS spectra of the sediment samples show that large amounts of Mo speciation in the rivers and ponds sediments are weakly adsorbed molybdate or polymeric molybdate on ferrihydrite.Sequential extraction shows that these Mo species are labile, which is consistent with XAFS results.It can be inferred that the pond and river sediments play a role as a secondary contamination source of Mo rather than as a sink of Mo in the area.
The weakly adsorbed molybdate can be desorbed from mineral surfaces in solution with high salt concentrations, as demonstrated in the F1 treatment from the sequential extraction (Figure 2a).The increase of salt concentrations in river and pond water by some artificial or natural processes might increase the release of Mo to solution.Change of pH in the pond and river waters can also occur by some artificial or natural processes.The lower pH engenders the increase of positive charge on ferrihydrite surfaces, especially in the presence of SO 4 [52].The negatively charged molybdate anion adsorbs favorably to a positively charged ferrihydrite surface at a low pH condition.By contrast, the increase of pH engenders the decrease of positive charge on ferrihydrite surfaces.Then, the adsorbed Mo must be released to the solution.
The surface sediments examined in present study will be in burial condition after the accumulation of sediments with time.Blanchard et al. [30] examined the post-depositional mobility of Mo in mine tailings.They showed that the adsorbed molybdate on ferrihydrite and ferrimolybdeite (Fe 2 (MoO 4 ) 3 •8H 2 O) dissolves in burial condition.However, the dissolved Mo(VI) in the porewater effectively precipitates to form powellite (CaMoO 4 ) over time.The ponds and rivers from Erdenet area are also characterized to be high Ca concentration (Table 1).Therefore, the formation of powellite in burial condition possibly occurs in Erdenet area.Moreover, at more reducing conditions, the reduced Mo(IV) is expected to transform to molybdenum sulfide [53] because of the relatively high SO 4 concentration in pond and river in the area.The solubilities of sulfides are low at reducing conditions [8].Therefore, the mobility of Mo at the burial condition is possibly limited, but additional studies must be conducted to elucidate the post-depositional mobility of Mo in Erdenet area.

Conclusions
Waters collected from rivers and the ponds near the Cu-Mo mining area in Erdenet city were characterized as neutral pH and Ca-SO 4 rich type.Although the concentrations of Cu in the rivers and ponds were low because of their neutral pH, those of Mo were significantly higher than the world average.The Mo concentrations in the secondary ponds connected to the huge tailing pond exceed the level stipulated in WHO guidelines (70 µg/L).
The distribution of speciation of the Mo in the sediments was examined from five-step sequential extraction and XAFS.The XAFS spectra of the sediment samples revealed that high degrees of Mo speciation in the river and pond sediments are weakly adsorbed molybdate or polymeric molybdate on ferrihydrite.The sequential extraction consistently showed large amounts of Mo distributed in the labile fraction.Speciation of Mo in the surface sediments suggests that the pond and river sediments play a role as a secondary contamination source of Mo rather than as a sink of Mo in the area.

Figure 1 .
Figure 1.Study area with sampling locations: (a) Erdenet mine in Mongolia, (b) Erdenet mining area with sampling locations, and (c) the junction of the three rivers with sampling locations.

Figure 1 .
Figure 1.Study area with sampling locations: (a) Erdenet mine in Mongolia, (b) Erdenet mining area with sampling locations, and (c) the junction of the three rivers with sampling locations.

Figure 3 portrays
Figure 3 portrays EXAFS spectra in k and R spaces for the S10.2 sample and the reference materials.The Mo contents of other samples (S4, S7, S8 and S10.1) are too low (less than 100 mg/kg) to elicit reliable spectra.The EXAFS spectra provide information related to the coordination environment of the Mo species in S10.2.The RSF of S10.2 exhibits a prominent peak because of Mo-O at R + R = 1.1 Å with a shoulder at 1.7 Å suggesting that multiple Mo-O shells are present for Mo in S10.2.No evidence exists of the presence of Mo-S bonding in the sample because of the lack of Mo-S shells, which will also be confirmed by X-ray absorption near edge structure (XANES) (Figure 4).The Mo-O peak in the R-range of 0.65-2.05Å was fitted well by three Mo-O shells at distances of 1.72 Å (coordination number (CN): 2.3), 1.91 Å (CN: 0.61), and 2.11 Å (CN: 1.3), where the respective errors of the distance and CN are 0.02 Å and 20% [40].They are possibly ascribed to a mixture of various Mo-O species in molybdate ion (Mo-O distance: 1.77 Å [41]), heptamolybdate (Mo-O distance: 1.70-1.97Å [42]), and MoO2 (Mo-O distance: 1.95-2.07Å [43]).Although the specific mineral phases containing Mo cannot be ascertained solely from this information, molybdate and heptamolybdate can be formed by the dissolution of Mo into water, which can subsequently be adsorbed or incorporated into secondary phases during solid-water interactions.In particular, molybdate ion is adsorbed on ferrihydrite by forming outer-sphere complex, which leads to the similar spectra of hydrated molybdate ion and that adsorbed on ferrihydrite[32].MoO2 species can be present in the solid phase, which is not related to the solid-water interactions, in particular for the samples formed under oxic conditions for the samples in this study.

Figure 3
Figure 3 portrays EXAFS spectra in k and R spaces for the S10.2 sample and the reference materials.The Mo contents of other samples (S4, S7, S8 and S10.1) are too low (less than 100 mg/kg) to elicit reliable spectra.The EXAFS spectra provide information related to the coordination environment of the Mo species in S10.2.The RSF of S10.2 exhibits a prominent peak because of Mo-O at R + ∆R = 1.1 Å with a shoulder at 1.7 Å suggesting that multiple Mo-O shells are present for Mo in S10.2.No evidence exists of the presence of Mo-S bonding in the sample because of the lack of Mo-S shells, which will also be confirmed by X-ray absorption near edge structure (XANES) (Figure 4).The Mo-O peak in the R-range of 0.65-2.05Å was fitted well by three Mo-O shells at distances of 1.72 Å (coordination number (CN): 2.3), 1.91 Å (CN: 0.61), and 2.11 Å (CN: 1.3), where the respective errors of the distance and CN are 0.02 Å and 20% [40].They are possibly ascribed to a mixture of various Mo-O species in molybdate ion (Mo-O distance: 1.77 Å [41]), heptamolybdate (Mo-O distance: 1.70-1.97Å [42]), and MoO 2 (Mo-O distance: 1.95-2.07Å [43]).Although the specific mineral phases containing Mo cannot be ascertained solely from this information, molybdate and heptamolybdate can be formed by the dissolution of Mo into water, which can subsequently be adsorbed or incorporated into secondary phases during solid-water interactions.In particular, molybdate ion is adsorbed on ferrihydrite by forming outer-sphere complex, which leads to the similar spectra of hydrated molybdate ion and that

Figure 3 .
Figure 3. Mo K-edge EXAFS spectra in (A) k and (B) R spaces for S10.2 and related reference materials.Spectra (a) in k and R spaces are presented as small circles for S10.2.Fitting of the spectrum (a) in R space by assuming Mo-O shells in heptamolybdate ion in water (c), molybdate ion in water (d), and MoO2 (e) is shown as a bold solid curve.Inverse Fourier transform spectrum for S10.2 (R range: 0.65-2.05Å) is shown as small circles of spectrum (b) in (A) with a fitted curve (bold) by assuming three Mo-O shells in heptamolybdate ion in water (c), molybdate ion in water (d), and MoO2 (e).

Figure 3 .
Figure 3. Mo K-edge EXAFS spectra in (A) k and (B) R spaces for S10.2 and related reference materials.Spectra (a) in k and R spaces are presented as small circles for S10.2.Fitting of the spectrum (a) in R space by assuming Mo-O shells in heptamolybdate ion in water (c), molybdate ion in water (d), and MoO 2 (e) is shown as a bold solid curve.Inverse Fourier transform spectrum for S10.2 (R range: 0.65-2.05Å) is shown as small circles of spectrum (b) in (A) with a fitted curve (bold) by assuming three Mo-O shells in heptamolybdate ion in water (c), molybdate ion in water (d), and MoO 2 (e).

Figure 4 portrays
Figure 4  portrays XANES spectra of S8 (a), S10.1 (b) and S10.2 (c) with linear combination fitting (LCF) to the reference materials.High-quality XANES spectra were not obtained from other samples.Good fitting to the spectra of sediment samples was obtained by LCFs of three reference spectra of molybdate adsorbed on ferrihydrite, MoO 2 and heptamolybdate, which are inferred from EXAFS analyses.The similar XANES spectra between aqueous MoO 4 2− and that adsorbed on ferrihydrite,

Figure 4 .
Figure 4. Mo K-edge XANES spectra for samples (S8, S10.1, and S10.2) and reference materials (MoO2, heptamolybdate ion in water, molybdate ion in water, molybdate ion adsorbed on ferrihydrite and MoS2).Fitting results for the samples are shown as bold curves.Deconvolution results for S10.2 are given by showing the contributions of the three reference materials.

Figure 4 .
Figure 4. Mo K-edge XANES spectra for samples (S8, S10.1, and S10.2) and reference materials (MoO 2 , heptamolybdate ion in water, molybdate ion in water, molybdate ion adsorbed on ferrihydrite and MoS 2 ).Fitting results for the samples are shown as bold curves.Deconvolution results for S10.2 are given by showing the contributions of the three reference materials.

Author
Contributions: T.S., K.F., O.A. and N.H. designed the fieldwork and analyses for water chemistry, sequential extraction and XRD.T.S., K.F., O.A., A.A., D.D., Y.A., I.E.and B.D. took samples and field measurements.Y.T. performed XAFS analysis.T.S., A.A. and G.B. conducted analyses for water chemistry, sequential extraction, and XRD.T.S., K.F., O.A. and N.H. wrote the paper.Funding: Financial support was provided by the Higher Engineering Education Development Project, Functional material based on Mongolian Natural Minerals for Environmental Engineering, Cementitious and Float Process (No. J11A15), Biological Active Compounds and Useful Genes from Mongolian Plants, Microorganisms and their Application (No. J12A15), the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) (No.JP17H06458), the Environmental Research Project (No. 163288) from the Sumitomo Foundation, the Department of Research and Innovation, National University of Mongolia and a cooperative research program of the Institute of Nature and Environmental Technology, Kanazawa University (Nos.17039 and 18036).
).The sequential extraction applied in the present study

Table 1 .
Sequential extraction scheme in present study.
That of the Gavil River is lowest (8 µg/L) among the samples.The Erdenet River and Khangal River show mutually similar Mo concentration of around 25 µg/L.The Cu concentrations of all samples were lower than 10 µg/L.

Table 3 .
Proportion of molybdate on ferrihydrite, heptamolybdate and MoO2 by Mo K-edge XANES.

Table 3 .
Proportion of molybdate on ferrihydrite, heptamolybdate and MoO 2 by Mo K-edge XANES.

Table A1 .
Total concentrations were calculated by ignoring the concentrations with under the detection limits.Concentrations of respective elements in each fraction after sequential extraction.
1Total concentrations were calculated by ignoring the concentrations with under the detection limits.