1 Introduction

Tungsten(W) is a transition element listed in the Group VI B of the periodic table, sharing the same subgroup as chromium (Cr) and molybdenum (Mo). As a strategic metal, W has been widely used in light sources, ammunitions, electronics, metallurgies, and chemical industries (Koutsospyros et al., 2006). Metallic W and its alloys can be quickly oxidized and dissolved into tungstate oxyanions at appropriate pH and redox potential in the environment (Hobson et al., 2020). To some extent, these soluble tungstate oxyanions are usually manifested as high mobility and can migrate with surface flows or be leached into groundwaters where they can negatively impact the biota and aquatic ecosystems (Lin et al., 2014; Guo et al., 2019; Cidu et al., 2021). In Falon, Nevada, USA, exposure to high levels of W in aquifers used as drinking water may be the leading cause of local childhood leukemia (Seiler et al., 2005). Natural surface water bodies normally contain < 2 μg/L W (Gaillardet et al., 2003; Steenstra et al., 2020), nevertheless, in mining/smelting-impacted regions, W concentration of the surface and pore waters can be as high as 7.1 mg/L (Petrunic and Al, 2005; Du et al., 2021).

Once present in solution, the concentration and mobility of W are greatly affected by adsorption on and desorption from the surface. In general, W might exhibit similar interfacial behaviors to other oxyanion-forming metals such as arsenic (As) that it is strongly sorbed to Fe/Mn (oxyhydr)oxides in oxic-to-anoxic waters (Johannesson et al., 2013; Cao et al., 2019; Wasylenki et al., 2020; Cui and Gomes, 2021; Cui et al., 2021). In addition, aluminosilicates such as montmorillonite, kaolinite, and illite can also sequestrate tungstates at acid-to-neutral pH (Li et al., 2014; Iwai and Hashimoto, 2017; Du et al., 2022). In addition to minerals, the roles that organic surfaces, e.g., natural organic matter (NOM), play in determining W complexation and mobility in aquatic systems are still unknown.

Natural organic matter (NOM), particularly dissolved organic matter (DOM), is highly reactive and plays an essential role in the geochemical cycling of trace elements (Mostofa et al., 2013). Dissolved organic matter has a large number of functional groups on its surface, including phenolic hydroxyl (–OH), amino (–NH2), and carboxyl (–COOH). In one aspect, these groups can complex metals via the formation of strong inner-sphere type complexes (Xiong et al., 2013; Shi et al., 2016), and the formed soluble DOM–metal complexes may increase their mobility and potential risks (Zhang et al., 2021). In another aspect, DOM may enhance metal cations or compete with oxyanions for binding onto surfaces such as ferrihydrite or alumina (Du et al., 2018a; Fariña et al., 2018; Xue et al., 2019), altering their distribution coefficients at the solid–liquid interface.

For the same subgroup elements Mo and Cr, the importance of carboxylic groups of NOM in metal–complexation reaction has been documented (Gustafsson et al., 2014; Gustafsson and Tiberg, 2015), and a Cr/Mo–O–C structure was observed in the second coordination shell. To the best of our knowledge, existing data regarding DOM–W interactions are scarce. In particular, the functional groups responsible for the complexation of W(VI) are still unclear. Moreover, NOM has unique redox properties and is therefore capable of reducing redox-sensitive metals such as Cr(V), Hg(II), and Cu(II) (Wittbrodt and Palmer, 1995; Gu et al., 2011; Pham et al., 2012). Whether and to what extent DOM can reduce hexavalent W(VI) to lower oxidation state species need to be illuminated. The above fundamental information is a prerequisite that enables us to evaluate the controls on W mobility in surface and groundwater flows in a more comprehensive way.

To fill this knowledge gap, this study aimed to examine W(VI) complexation behaviors by representative DOM, i.e., humic acid (HA) and fulvic acid (FA). We first determined the conditional distributions coefficients (Dom) for W(VI) on HA and FA by a batch equilibrium method at varied pHs and W concentrations. In addition, the influences of cations, Ca2+ as a representative, were explored. Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS), combined with isothermal titration calorimetry (ITC) were employed to figure out the complexation mechanisms. The information gleaned from this study has substantial implications for understanding the fate of tungsten in aquifers and evaluating its bioavailability and environmental risks.

2 Experimental section

2.1 Humic and fluvic acid preparation and characterization

HA and FA are both commercial products, provided by Sigma-Aldrich (product code, H16752) and Chengdu Euenreisi Chemical Reagent Co., LTD., China, respectively. HA was originally extracted from peat while FA was isolated from forest soil. Although the properties of humic substances vary differently with different sources, the major functional groups for metal(loid) complexation are always carboxylic and phenolic hydroxyls. Many studies have shown that FA has a higher acidity and more functional groups than HA in spite of environmental sources (Rashid and King, 1970; Weber and Wilson, 1975). Therefore, these two commercial and easily accessible humic substances were chosen as representatives. To prepare the dissolved humic substances, 500 mg of HA or FA was dissolved in 100 mL “Milli-Q” water, and 1 M NaOH or HCl was added if necessary to facilitate the dissolution. The solution was then filtered through a cellulose nitrate filter (specific diameter of 0.45 µm) and stocked in the refrigerator at 4ºC before use. The carbon (C) concentration was determined by a Shimadzu TOC analyzer (Shimadzu, Japan). The carboxylic and phenolic hydroxyl groups were determined by potentiometric acid–base titrations using a Metrohm titrator 836. The titration data were processed in the FITEQL 4.0 software using a 2-sites non-electrostatic model (Fig. S1). FA and HA elemental compositions were determined using an elemental analyzer (Vario unicube, Elementar, Germany). The basic properties of HA and FA are listed in Table 1.

Table 1 Chemical properties of humic acid (HA) and fulvic acid (FA)
Full size table

2.2 Batch equilibrium experiments

The W(VI) stock solution, ~ 1000 mg/L, was prepared by diluting of sodium tungstate (Aladdin) in “Milli-Q” water. Equilibrium complexation experiments were performed as a batch-type in 50 mL polypropylene centrifugate tubes. The reaction solutions (30 mL) contained 0.05 g C/L HA or FA and varied W(VI) concentrations (0 - 30 mg/L) in 0.01 M NaCl background electrolyte solutions. The suspension was adjusted to desired pH values (~ 5, 7, 9) using dilute NaOH and HCl, and shaken end-over-end for 48 h at 25 ºC to an equilibrium. The suspension was centrifuged at 4500 g for 20 min and then filtrated through a 3 k Dalton ultrafilter (Millipore, Amicon) (Du et al., 2018b). Through this approach, the free and DOM bound W could be almost completely separated because 1) the Dalton of HA and FA is commonly larger than 8 k Dalton (Kim et al., 1990); 2) HA or FA might contain minimum smaller molecular entities that were smaller than 3 k Dalton, however, our preliminary experiment showed that the DOC in the 3 k Dalton filter was below the LOD (limit of detection, ~ 50 μg/L). Even if there were some DOM-bound W in the filtrate, they were negligible compared to the total DOM-bound W (dozens of mg/L, see the Results section). The filtrate was immediately analyzed for dissolved W(VI) concentration using ICP-OES (PerkinElmer Optima 8300). Quantification by ICP-OES was based on a comparison to a 5-point standard curve using the 207.9 nm emission line which had a detection limit of 10 μg/L. The DOM-bound W was calculated by the difference between the total added and the dissolved concentrations. The effect of Ca2+ was investigated for an initial W concentration of 15 mg/L and varied Ca-to-W molar ratios (1:2, 1:1, and 5:1). Other treatments were the same as above. All batch experiments were performed in triplicates.

Conditional distribution coefficients (KD) for W(VI) binding to DOM (KD, L/kg) were calculated as follows (Buschmann et al., 2006):

$${K}_{\mathrm{D}}=\frac{\left[{C}_{\mathrm{b}}\right]}{\left[{C}_{f}\right] \left[\mathrm{DOC}\right]}$$

where Cf and Cb are the free and DOM-bound W(VI) concentrations (mg/L), respectively; and [DOC] is the concentration of organic carbon in solution (kg/L).

2.3 Spectroscopic measurements

The complexation mechanisms including binding functional groups and possible reduction reactions were investigated by FTIR and XPS. XPS measurement was conducted on a KRATOS Axis Ultra X-ray photoelectron spectrometer (Thermo Fisher Scientific, US). Spectra of W 4f (5p) were collected with a step size of 0.05 eV. The C 1 s peak at ~ 284.8 eV was used to calibrate the binding energy, and the spectra were curve-fitted using XPSPEAK41 software. FTIR measurements were performed on a PerkinElmer infrared spectrometer. Each scanning spectrum was recorded from 4000 cm–1 to 650 cm–1 at a 4 cm–1 resolution. Before XPS and FTIR measurements, liquid solutions were freeze-dried into a fine powder.

2.4 Calorimetric titration

Isothermal titration calorimetry (ITC) was applied to directly measure the heat exchange during W(VI) complexation by DOM. Calorimetric titration was performed using a thermal activity monitor (TAM III, TA instruments) equipped with a 1-mL reaction vessel. Before measurements, 0.7-mL of HA or FA solution (0.05 g/L) was added to the reactor stirring at 120 rev/min. The reactor was loaded inside the instrument and left to stabilize to achieve a stable heat flow, i.e., a signal excursion < 250 nw/h. A W(VI) stock solution was subsequently titrated into the reactor using a fully automated micro-syringe. Each titration was 10 μL and the corresponding reaction heat flow was automatically recorded. In total, 250 μL of W(VI) stock solution was titrated. A blank experiment was performed in the same way but using the 0.01 M NaCl instead of the DOM solutions, to exclude the dilution heat that was not caused by W(VI) complexation. All measurements were conducted at pH ~ 7 and 25ºC. Thermodynamic parameters including enthalpy (ΔH), entropy (ΔS), Gibbs free energy (ΔG), and binding affinity (Ka) were determined using the software package TAM assistant (ver. 1.4) and NanoAnalyze (ver. 3.6) followed the methods described elsewhere (Du et al., 2020):

$$Q=V[M]\frac{n\Delta HK}{1+K[L]}$$

where Q is the reaction heat, V is the reactor volume, [M] is the concentration of humic substances, ΔH represents the enthalpy (kJ/mol), K is the thermodynamic affinity, [L] is the total concentration of W(VI), and n represents the number of binding sites. The calculation of ΔG was: ΔG =  − RTlnK (R = 8.314 J/mol/K, T = 298 K). Entropy ΔS was determined by: ΔS = (ΔH − ΔG) / T.

3 Results

3.1 DOM-bound W(VI) concentrations and conditional distribution coefficients

A noticeable fraction of W(VI) was bound to HA and FA at all studied pHs, with higher proportions at low initial W(VI) concentrations (Fig. 1). For HA, the bound W was from 0.47 mg/L (initial W = 1 mg/L) to 3.1 mg/L (initial W = 30 mg/L), corresponding to 47%–10% of the total W(VI) in solution. The bound W on FA ranged from 0.55 to 4.9 mg/L, equivalent to 55%–16% of total W(VI). Fulvic acid seemed stronger than HA to complex W(VI), which further can be inferred from its larger KD values (Fig. 2). The values of KD decreased with the increase of initial W(VI) concentrations and remained almost constant at higher W(VI) concentrations. Both HA and FA showed pH dependence of KD values which increased with the decrease of pH, suggesting a pH-dependent W complexation behavior. Combined with the significance analyses (Fig. S2), lower pH favored W(VI) complexation by DOM, and FA showed a stronger complexation ability for W(VI) than HA.

Fig. 1
figure 1

Concentrations of free and bound W in humic acid (HA) and fulvic acid (FA) solutions (50 mg C/L) at pH ~ 5, 7, and 9. The initial concentrations of W spiked into the DOM solution are denoted on the x-axis. Error bars indicate standard deviations of three replicates

Full size image
Fig. 2
figure 2

Conditional distribution coefficients (KD, y-axis) of W(VI) between water and DOC as a function of free W(VI) concentration (Cf, x-axis) at pH ~ 5, 7, and 9. The DOC concentration in HA/FA solutions was 50 mg/L. Data points represent experimental results. Solid lines are fitted lines with nonlinear regression. Error bars indicate standard deviations of three replicates

Full size image

When a specific complexation mechanism is assumed (see the Discussion section), the occupation of functional groups of DOM can be estimated (Buschmann et al., 2006). At the highest W concentration (30 mg/L), only ~ 2.5% and 2.2% of the total functional groups were occupied for HA and FA, respectively.

3.2 Influences of Ca2+ on W(VI) complexation

The presence of Ca2+ significantly promoted W(VI)–DOM complexation at almost all studied pHs, with greater W(VI) adsorptions at higher Ca:W ratios (Fig. 3). The highest W adsorption on FA, ~ 95 mg/g, occurred at a Ca:W ratio of 5 at pH 5, about 2 times larger than that without Ca2+. Noticeably, the promotion effect of Ca2+ on W(VI)–DOM complexation was also dependent on pH. On HA, W(VI) adsorption increased by ~ 43–61%, 28–55%, and 0–18%, respectively at pH ~ 5, 7, and 9. On FA, W(VI) adsorption increased by ~ 41–101%, 14–86%, and 13–68%, respectively at pH ~ 5, 7, and 9. Overall, Ca2+ can promote W(VI) complexation by DOM, and lower pH favored the DOM–Ca–W(VI) ternary interactions.

Fig. 3
figure 3

Complexed W(VI) amounts per mass (mg/g) on humic acid (HA) and fulvic acid (HA) (50 mg C/L) at different pHs (~ 5, 7, and 9) and initial Ca:W molar ratios. Error bars denote ± standard deviation (n = 3). Lowercase letters represent significant differences using one-way ANOVA (p < 0.05)

Full size image

3.3 Functional groups for W(VI) complexation

Two FTIR absorption bands have significantly changed after HA binding of W(VI): One was at ~ 1613 cm–1, corresponding to the antisymmetric vibration of COO– (Niemeyer et al., 1992), and it shifted to higher frequencies; the other one was at ~ 1403 cm–1, which represented the symmetric vibration of COO– and/or C–O stretching of phenols (Lumsdon and Fraser, 2005), showing reduced intensity and generating a new peak variation at ~ 1457 cm–1 (Fig. 4a). These two spectral changes probably resulted from the ligand exchange of carboxyl and phenolic hydroxyl groups by W(VI) oxyanions. Similar changes in antisymmetric vibration of COO– at ~ 1607 cm–1 were observed when FA binding W(VI) (Fig. 4b). These observations suggested that carboxyls were the major functional groups for W(VI) complexation by DOM, and the phenolic group was likely to complex a portion of W(VI). In contrast to variations of carboxylic and phenolic groups, the absorption bands between 1000 and 1150 cm–1, representative of C–O of ester and polysaccharide (Zhang et al., 2017), were almost unchanged, implying that the aliphatic structures were more resistant to W(VI) complexation.

Fig. 4
figure 4

FTIR spectra of humic acid (HA) and fulvic acid (FA) before and after reacting with W(VI). Experimental condition: 50 mg C/L, initial W concentration of 30 mg/L, pH 7, ionic strength 0.01 M

Full size image

3.4 Reduction of W(VI) to W(V) during complexation

In a typical W 4f7/2 XPS spectrum, the peak binding energies for W6+ and W5+ locate at ~ 35.6 eV and 35.0 eV, respectively (Du et al., 2020) (Fig. 5). Approximately 46% of the total adsorbed W(VI) was reduced to W(V) on HA, compared to a mere 10% reduction of W(VI) on FA. Ca2+ inhibited W(VI) reduction on DOM, with almost no existence of pentavalent W(V) on HA. Collectively, HA exhibited a higher reducibility than FA, and Ca2+ played an inhibiting role in W redox reactions.

Fig. 5
figure 5

Tungsten (4f, 5p) XPS spectra of W(VI)–DOM complexes under different conditions. Experimental condition: 50 mg C/L, initial W concentration of 30 mg/L, pH 7, ionic strength 0.01 M. The W 4f7/2 spectrum can be decomposed into W5+ 4f7/2 at ~ 35.0 eV and W6+ 4f7/2 at ~ 35.6 eV as appropriate, and the corresponding proportions(%) are given beside the peak

Full size image
Fig. 6
figure 6

Heat exchange (μJ) versus per titration for W(VI) complexation on humic acid (HA) and humic acid (FA) at pH ~ 7.0. Lines are the two-sites model fits

Full size image

3.5 W(VI) Complexation thermodynamics

For W(VI) complexation by HA and FA, the first 2 injections generated exothermic heat, followed by 8 releases of endothermic heat (Figs. S3 and 6). These observations suggested that two different complexation processes might occur, one was exothermic and the other one was endothermic. The peak amplitudes and areas decreased to almost zero after 10 injections, implying a complete consumption of available binding sites. For W–HA interaction, thermodynamic fitting results showed one process with negative ΔH (–5.2 kJ/mol) and positive ΔS (43.8 J/mol/K), and another with positive ΔH (30.8 kJ/mol) and ΔS (158.7 J/mol/K) (Table 2). The W–FA interaction generated a larger heat exchange of –17.8 kJ/mol and 158.6 kJ/mol. This result corroborated the batch adsorption results that more W(VI) was complexed by FA than by HA. The entropy changes for W–FA complexation were 6.7 and 597 J/mol/K. In conclusion, W(VI) complexation by HA and FA was a spontaneous process (ΔG < 0) of entropy increase, and two different complexation mechanisms might occur with opposite thermal effects.

Table 2 Complexation calorimetry parameters for W(VI) on humic acid and fulvic acid
Full size table

4 Discussion

4.1 W(VI) complexation mechanisms

Tungstate species are negatively charged, while HA and FA are overall negatively charged as well. Only weak W(VI) binding would be expected. However, we found quite a few W(VI) complexation by both HA and FA, suggesting a ligand exchange reaction might have occurred. For coordination numbers < 6, e.g., WO42–, an associative ligand exchange mechanism at positively charged metal centers can take place (Huheey et al., 2014). As the tungstate center has a high formal charge + VI, an addition of hydroxyl entity (e.g., carboxylic or phenolic) at the electrophilic center followed by protonation and water molecule release might occur. Although the overall charge was negative for both reactants, this formed species can be stabilized through additional processes such as H–bridges (Buschmann et al., 2006). Such a process would consume H+. That was why W(VI) complexation by HA/FA was stronger at acidic pH. Using FTIR, we only confirmed the participation of carboxylic groups in such ligand exchange reaction. ITC as a supplementary clearly showed two distinctly different complexation processes. Thus, carboxylic and phenolic hydroxyl groups both participated in the ligand exchange reaction (Fig. 7), consistent with previous observations for other metal(loid)s (Shi et al., 2016; Xu et al., 2016; Lu et al., 2017; Besold et al., 2019).

Fig. 7
figure 7

Ligand exchange reaction of W(VI) by different functional groups of HA and FA

Full size image

The pKa of carboxylic and phenolic groups were around 5 and 8, respectively (Table 1), so the carboxylic group seemed easier to protonate than the phenolic groups. However, previous studies have shown a higher affinity of proton binding to phenolic-type sites than to carboxylic-type sites (Xu et al., 2018), and phenolates are better Π-donors than carboxylates (Buschmann et al., 2006). That means when metal(loid)s, such as W(VI), are exposed to DOM, phenolic groups may first participate in the complexation reaction, followed by the carboxylic groups. Taken together with the ITC results that W(VI) complexation by DOM first generated an exothermic heat followed by an endothermic heat, we concluded that W(VI) complexation by phenolic-type sites of DOM was an exothermic process whereas the W–carboxylic complexion was an endothermic process. For either W–carboxylic or W–phenolic interaction, they both produced positive entropies (158.7 J/mol/K and 597 J/mol/K), likely due to the displacement of solvating water molecules from the coordination sites when ligand exchange was taken place (Chen et al., 2020; Du et al., 2020). The negative enthalpy and positive entropy of W–phenolic complexation suggested this process was driven by both enthalpy and entropy, while the W–carboxylic association was driven only by entropy.

It should also be pointed out that, at pH 5, W(VI) formed polymetric species such as H2W12O4210–, HW7 O245– and W7O246–, based on our solution speciation analyses by Visual Minteq (Table S1). Therefore, multiply species, including poly-tungstate species, such as RC–O–W–O–W, might be present on DOM surfaces. Polymeric species are commonly metastable compared to the mono-tungstates (Sun and Bostick, 2015), further influencing W’s solubility and mobility.

4.2 Complexation differences between HA and FA

Both batch adsorption and thermodynamic results showed a stronger complexation capacity of FA than HA, particularly at higher W(VI) concentrations. Although FA and HA had very close phenolic hydroxyl concentrations (0.32 and 0.37 mol/kg, Table 1), FA indeed possessed more carboxyls, which was 0.86 mol/kg for FA compared to 0.33 mol/kg for HA. At low W(VI) loading, phenolic hydroxyls preferentially participated in the complexation, hence the differences between HA and FA were not significant. We also observed that HA and FA could reduce W(VI) to W(V), indicative of the presence of redox-active functional moieties. To verify this assumption, we presented an electrochemical approach to study the redox properties of HA and FA (See supporting information). Direct electrochemical reduction of HA showed a decreased electron current from > 4 μA to ~ 1.8 μA when reached stable, while for FA, the decrease was only ~ 1.1 μA. Thus HA has a higher reductive potential than FA, leading to a greater reduction of W(VI) to W(V) as shown in XPS results. This reductive phenomenon is probably because DOM contains a variety of reductive moieties including quinones, substituted phenols, α-hydroxyl carboxylic acids, and α-carbonyls carboxylic acid (Wilson and Weber, 1979; Jiang et al., 2014), and humic substances are more redox-active than fulvic acid (Yang et al., 2016).

4.3 The role of Ca2+ on W(VI) complexation by HA/FA

We showed that Ca2+ promoted the W–HA/FA interactions at environmentally relevant concentrations (1.6 to 16.3 mg/L). Ca2+ is one of the most common cations in natural aquifers, and usually affects the interaction of DOM with other metal pollutants such as arsenic (As). Ca2+ has a preference for binding to larger and more negatively charged molecules such as carboxylic groups, phenolic sites, and even amine moieties (Kinniburgh et al., 1999). Ca2+ can not only compete with other cations (i.e., Fe3+ and Cu2+) for DOM’s available sites (Adusei-Gyamfi et al., 2019), but also serve as bridges in oxyanions (i.e., As) binding to DOM (Ren et al., 2017; Zhang et al., 2021). Herein, the promoting role of Ca2+ in W(VI) complexation by DOM may be attributed to an indirect or a direct reason: 1) Binding of Ca2+ on DOM molecules reduced the negative charges and electric repulsion between DOM and W(VI) anions, hence allowing the two reactants to approach each other more readily; 2) By cationic bridging between carboxylic or phenolic groups and W(VI), a ternary DOM–Ca–W complex might be formed (Fig. 8, carboxylic groups as an example), and attracted additional W(VI) to be complexed by DOM. This bridging increased with increasing Ca2+ concentrations, which resulted in more bounded W(VI). Even though Ca2+ exhibits a relatively weak bonding with DOM, compared to trivalent cations such as Fe3+ and Al3+, this bridging cannot be ignored considering the high concentration of Ca2+ in most aquifers; 3) At the experimental condition, scheelite (CaWO4) was supersaturated according to our solution speciation analysis (Figs. S6–S7), approximately < 10% of total W(VI) can exist as precipitates which could also limit dissolved W concentrations. We also found that Ca2+ could almost completely inhibit W(VI) reduction by DOM. Because Ca2+ is not redox-active compared to, for example, Fe(III), thus Ca2+ was probably blocking the redox-active functional moieties of DOM to hinder the redox reaction.

Fig. 8
figure 8

Possible ternary DOM–Ca–W complex; carboxylic groups are taken as an example

Full size image

4.4 Environmental considerations

When studying tungsten speciation and mobility in the aquifers, W complexation by dissolved organic substances including HA and FA should be taken into consideration. In aquatic systems that are impacted by W mining-smelting activities where W concentration can as high as several mg/L, ~ 20% of total W(V) can be bound to dissolved DOM. In natural aquifers with much lower W concentrations, this percentage is even higher. Coexisting cations in natural waters should also be taken into account because we observed that in the presence of Ca2+ more W(VI) was bound to HA and FA. Moreover, DOM is also capable of reducing W(VI) to a lower valence state of W(V) which may subsequently affect their biotoxicity and environmental risks.

5 Conclusions

Tungsten species were shown to react with DOM (HA and FA), and this association was highly dependent on pH, coexisting cations, and the concentration of W(VI). Tungsten–DOM interaction is proved to involve ligand exchange on the O-containing functional groups including carboxylic and phenolic sites. DOM could reduce hexavalent pentavalent W(VI) to W(V) via its redox-active moieties. The presence of Ca2+ promoted W–DOM complexation and inhibited W(VI) reduction on DOM. We conclude that DOM servers as an important pool of chelatable and redox-active for W in natural aquifers. Nevertheless, how this association would affect W’s mobility, including the intensity of W adsorption onto the solid phases (i.e., iron (oxyhydr)oxides), needs further investigation. Overall, the study provides a basis for better evaluating the migration, bioavailability, and fate of tungsten in DOM-laden surface and ground waters and more appropriate designs of complexation-based treatment methods for tungsten in natural water.