1. Introduction
Human activities have injected significant amounts of heavy metals (HMs) into coastal waters since the last century. The potential of those HMs entering the food chain is of great environmental and public health concern. Assessing the bioavailable fractions of HMs in sediments, therefore, is of great importance. Earlier studies have indicated that sulfur chemistry in sediments plays a major role in regulating the bioavailability of HMs [
1,
2,
3]. This is because HM sulfides are extremely insoluble; as such, they can be excluded from the bioavailability assessments. Unfortunately, HM sulfides are difficult to quantify in sediments because they are considered acid-extractable (AE), at least to some extent. An indirect indicator of HM bioavailability in sediments based on sulfur chemistry, therefore, has been proposed. Casas and Crecelius [
4] examined the role of acid-volatile sulfide (AVS) in predicting the toxicity of zinc, lead, and copper in marine sediments. They found that, in most cases, HMs were not detected in the pore water until the atomic ratio between the simultaneously extractable metals (SEM) and AVS exceeded unity. Furthermore, mortalities of
Capitella capitata occurred when SEM/AVS > 1. Further evidence for the utility of the SEM/AVS ratio to predict Cd, Cu, Pb, and Ni toxicities in sediments was provided by Berry et al. [
5] and Hansen et al. [
6]. However, the SEM/AVS ratio of sediments is an indirect indicator, which is non-specific, semi-quantitative, and not working in quite a few cases [
6].
Traditionally, HM sulfides in sediments have been considered to be mono-sulfides (S
=), such as Hg(II)S, Cu(II)S, Zn(II)S, Cd(II)S, Ni(II)S, Co(II)S, Cr(II)S, and Pb(II)S. Metal mono-sulfides are AE under nitrogen. Metal bi-sulfides (S
2= or SS
=), such as pyrite (FeS
2), on the other hand, are non-acid-extractable (NAE) under nitrogen. There are NAE (or non-reactive) fractions of HMs in sediments other than those occluded in clay minerals [
7,
8]. Huerta-Diaz and Morse [
8] and Morse and Luther [
9] assumed that the NAE fractions of HMs were co-precipitated (or occluded) with pyrite. They used the degree of trace metal pyritization (DTMP) to explain the NAE fractions of HMs in sediments. Their extensive studies on various sediments in the Gulf of Mexico have shown that (1) the DTMP of Co and Ni correlated well with the degree of pyritization (DOP) in the sediments and agreed with the DTMP assumption; (2) the DTMP of Pb, Zn, Cd, and Cr were relatively low regardless of the DOP, which implied that they were not easily occluded by pyrite and tend to form discrete sulfides [
9]; and (3) the DTMP of Hg was unusually high even with low DOP. They speculated that Hg probably had a very high affinity to be incorporated into pyrite. Raiswell and Plant [
10] also observed a similar Cu sulfide anomaly during the diagenesis of black shales in England. They also speculate that the Cu was incorporated into pyrite. The Hg or Cu anomalies in sediments remain unexplained except for the speculation that Hg and Cu were incorporated into pyrite [
8,
10].
This study was initiated to (1) investigate the chemical nature of Hg and Cu sulfides precipitated from solutions and the cause of the Hg and Cu anomalies in sediments and (2) determine the NAE (or non-reactive) and AE (or reactive) fractions of Hg and Cu in the sediments of Apalachicola Bay, North Florida. Zn was included in this study for comparison because Zn sulfide is known to be mono-sulfide and AE.
2. Materials and Methods
Hg and Cu sulfide samples: Three mercury sulfides—(1) a commercially available black mercury sulfide (labeled as Hg(II) sulfide, Alfa Aesar, Ward Hill, MA, USA); (2) a precipitated mercury sulfide by reacting H2S with a pH 7.5 Hg(II)(NO3)2 solution under nitrogen (the pH 7.5 mercury sulfide); and (3) a precipitated mercury sulfide by reacting H2S with a pH 4.5 Hg(II)(NO3)2 solution under nitrogen (the pH 4.5 mercury sulfide)—and two copper sulfides—(1) a precipitated copper sulfide by reacting H2S with a pH 7.5 CuSO4 solution under nitrogen (the pH 7.5 copper sulfide) and (2) a precipitated copper sulfide by reacting H2S with a pH 4.5 CuSO4 solution under nitrogen (the pH 4.5 copper sulfide)—were investigated in this study.
Precipitation of Hg and Cu sulfides: A total of 10 mL of 0.1 M Hg(II)(NO3)2 or CuSO4 solutions, which were purged with nitrogen gas and adjusted to their respective pH using either NaOH or HCl solutions, was placed in a 20 mL scintillation vial. The scintillation vial was then inserted into the center of a 300 mL distillation bottle containing 100 mg of ferrous sulfide (FeS) powder at the bottom. The distillation bottle was then closed with a two-outlet stopper and purged with pure nitrogen. Then, the two outlets closed. A total of 20 mL of 1 M HCl solution was injected into the FeS powder in the bottom of the bottle through one of the outlets, and the outlet closed again. The released H2S from the FeS powder was allowed to react with the metal solution in the scintillation vial overnight under nitrogen. The scintillation vials were taken out the next day, and the contents were immediately transferred to centrifuge tubes, and the precipitate was separated from the solutions via centrifugation. The precipitate was washed twice with distilled water via centrifugation and then freeze-dried. The freeze-dried samples were kept in the refrigerator until they were analyzed.
Sediment samples: Sediment samples were collected from Apalachicola Bay, North Florida. Apalachicola Bay is a bar-built estuary located on the northwestern coast of Florida. It is approximately 63 km long, 12 km wide, 2–3 m deep, and covers an area of around 539–542 km
2 [
11,
12]. Seventy-six grab samples were collected from 16 stations. The grab sediments were immediately subsampled into 20 mL cut-off plastic syringes and sealed with rubber stoppers. Sample syringes were stored in ice during transportation to the laboratory, and the subsamples were processed, respectively, for moisture content (60 °C for three days), acid-volatile sulfide (AVS), chromium-reducible sulfide (CRS), HCl-extractable Hg, Cu, and Zn under nitrogen gas. The sulfide and metal analyses were carried out following the procedures described below.
AVS and CRS determinations: The AVS and CRS were determined using the diffusion methods [
13,
14]. A total of 20–30 mg of mercury or copper sulfide samples or 4–19 g of wet sediments (2–12 g dry weight) were used for the analysis. The AVS (mono-sulfides) and CRS (bisulfides and/or poly-sulfides) were analyzed sequentially; as such, the CRS did not include AVS. (That is, total sulfides = AVS + CRS.)
Acid-extractable (AE), non-acid-extractable (NAE), and total Hg, Cu, and Zn: The AE-Hg, AE-Cu, and AE-Zn extractions were conducted under nitrogen. Briefly, 15–20 mg dried sulfide specimens or 2–7 g wet sediment (1–4.42 g dry weight) samples were placed in a 300 mL distillation bottle and closed with a two-outlet stopper. The bottle was purged with nitrogen through the outlets, and then the outlets closed. A total of 10 mL of a 1 M HCl solution (prepared from double-distilled HCl and water and purged with nitrogen) was added to the bottle through one outlet of the stopper, and then the outlet closed again. The bottle was shaken gently overnight. Then, the contents of the bottle were separated via centrifugation. The supernatant was collected and filtered through a 0.45 µm filter for a graphite furnace atomic absorption spectrophotometry (GFAAS) analysis of Cu and Zn and a cold vapor atomic fluorescence spectrophotometry (CVFAS) analysis of Hg.
For total Cu, Zn, and Fe determinations, the conventional hot plate digestion method [
15,
16] was used. Pre-weighed samples and 10 mL of double-distilled, concentrated HNO
3 were added to Teflon beakers, covered with Teflon watch glasses, and heated up to 110 °C for two to three hours. Upon cooling, the digested samples were centrifuged to separate the supernatant and the solid. The supernatant was diluted to a volume of 50 mL with 0.1 N double-distilled HNO
3 and analyzed for Cu, Zn, and Fe using the National Institute of Standard and Technology (NIST) Standard Reference Material SRM 2704 Buffalo River Sediment (Cu 98.6 ng/g, Zn 408 µg/g, and Fe 39.7 mg/g) was also treated along with the samples to check the recoveries of the digestion methods, which were satisfactory. The NAE-Cu and NAE-Zn were obtained using the difference between the total and the AE concentrations. All results were reported on a dry-weight sediment basis.
The total Hg was determined directly with a mercury analyzer, the DMA-80 Evo (Milestone Srl, BR, Italy). Together with the samples, NIST Standard Reference Materials SRM 1570a spinach leaves (Hg 30 ± 0.03 ng/g) and SRM1515 apple leaves (Hg 44 ± 0.04 ng/g) were also analyzed to check the accuracy of the analysis, which was satisfactory. The NAE-Hg was obtained using the difference between the total and the AE concentrations. All results were reported on a dry-weight sediment basis.
Statistical Analysis: the standard deviation and the results of regression analysis were obtained using the statistics functions of the Microsoft Excel computer program (Microsoft Office 2021).
3. Results
3.1. Mercury and Copper Sulfides
Table 1 lists the results of the total, NAE, and AE of Hg and Cu, respectively, as well as AVS and CRS of the five sulfide specimens. Since AVS and CRS analyses were conducted sequentially, the CRS in this study did not include AVS. That is, the total sulfides (TS) were the sum of AVS and CRS.
The sulfide analysis of the five specimens showed that they were all dominated by CRS (95.5% to 100% of the TS). The metal analysis showed that the three mercury sulfide specimens had almost all NAE-Hg (95.6–99.2% of the total Hg). The two copper sulfide specimens had the majority of NAE-Cu (57.88% to 62.80% of the total Cu) but also had significant amounts of AE-Cu (37.2% to 42.1% of the total Cu).
Analytically, AVS represents mono-sulfides (HS
− and S
=), and CRS (excluding AVS) represents bi-sulfides (SS
=) or poly-sulfides [
13,
14,
17,
18]. This is because mono-sulfides can be dissolved in an HCl solution to form volatile hydrogen sulfide, which is then precipitated as ZnS for quantitation [
13,
19]. Bi-sulfides or poly-sulfides, on the other hand, cannot be directly dissolved in HCl. They need to be first reduced to mono-sulfide using an acidic Cr(II) solution and then quantified [
13,
14,
17,
18].
The results of AVS and CRS analyses indicated that the black mercury sulfide and the pH 7.5 copper sulfide had no AVS (mono-sulfide) but all CRS (bi-sulfide or poly-sulfides). The atomic ratios of CRS/NAE-Hg and CRS/NAE-Cu were 1.01 ± 0.01 and 0.98 ± 0.03, respectively. These results imply that the electron-balanced stoichiometric formula for the black mercury sulfide and the pH 7.5 copper sulfide samples should be Hg(I) bi-sulfide [(Hg+)2(SS=)] and Cu(I) bi-sulfide [(Cu+)2(SS=)], respectively, rather than the traditionally recognized Hg(II) mono-sulfide (Hg++S=) and Cu(II) mono-sulfide (Cu++S=), although all the atomic S/metal ratios are the same.
The AE-Hg in the black mercury sulfide was 1.17% of the total Hg, while the AE-Cu in the pH 7.5 copper sulfide was 37.2% of the total Cu. Since there was no AVS found in either of these samples, the AE-Hg and AE-Cu were presumably associated with anions other than sulfide.
The sulfide in the pH 7.5 mercury sulfide was 97.9% CRS and 2.1% AVS. Its Hg was 99.2% NAE-Hg and 0.8% AE-Hg. The atomic CRS/NAE-Hg atomic ratio was 1.29 ± 0.01, suggesting that the stoichiometry of the bi-sulfide was (Hg+)2(S1.58S=). The remaining 0.8% of AE-Hg had an atomic AVS/AE-Hg ratio of 3.27. There is no reasonable electron-balanced stoichiometric formula that has this atomic S/Hg ratio. The trace of AE-Hg, therefore, was not likely associated with sulfide but with other anions. The 2.1% AVS (mono-sulfides) were most likely associated with the cations of the impurities in the Hg(NO3)2 solution that was used for the mercury sulfide synthesis.
The sulfide in the pH 4.5 mercury sulfide specimen was 95.5% CRS and 4.5% AVS. Its Hg was 95.6% NAE and 4.4% AE. The atomic CRS/NAE-Hg ratio was 0.98 ± 0.03. The most likely electron-balanced stoichiometric formula for the Hg sulfide would also be Hg(I) bi-sulfide [(Hg+)2(SS=)]. The atomic AVS/AE-Hg ratio of the specimen was 0.98 ± 0.13. This 4.4% AE-Hg, however, was not likely Hg(II) mono-sulfide (Hg++S=) because divalent mercury could not be stable in the presence of sufficient mono-sulfide. (This will be discussed further in the Discussion section.)
The sulfide in the pH 4.5 copper sulfide specimen was 96.0% CRS and 4.0% AVS. Its Cu was 57.9% NAE and 42.1% AE. The atomic CRS/NAE-Cu ratio was 0.98 ± 0.03. The electron-balanced stoichiometric formula for the copper sulfide would be Cu(I) bi-sulfide [(Cu+)2(SS=)]. The atomic AVS/AE-Cu ratio of the 4.0% AVS was 0.057, far below the expected atomic ratio of Cu(I) mono-sulfide (Cu+)2(S =), which should be 0.5. The 4.0% AVS, therefore, was not likely associated with copper but with other metal impurities in the reagent. Currently, I am not sure what forms of the 42.1% AE-Cu could be in this specimen. I only know that it could not have been copper sulfide.
3.2. The Sediment Samples
The TS in the sediment samples ranged from 0.7 mg/g to 17.2 mg/g (average = 8.1 ± 4.6 mg/g, n = 76), indicating strong sulfidic conditions in the sediments. The TS was dominated by CRS (average = 98.7 ± 3.0% of the TS). The AVS ranged from trace to 0.360 mg/g (average = 0.102 ± 0.086 mg/g). There was no significant correlation found between AVS and TS (r2 = 0.24).
The total Hg contents of the sediment samples ranged from 1.2 ng/g to 102.5 ng/g (average 47.1 ± 27.5 ng/g,
n = 76). The AE-Hg concentrations of the sediments ranged from trace to 2.21 ng/g (average = 0.99 ± 0.51 ng/g,
n = 76). The AE-Hg in the sediments ranged from 0% to 15.5% of the total Hg (average = 2.1 ± 2.7%,
n = 76). The average NAE-Hg was 46.11 ± 29.29 ng/g, or 97.9 ± 2.8% of the total Hg. There was no correlation found between the total Hg and TS (
r2 = 0.011), the AE-Hg and TS (
r2 = 0.192), the AE-Hg and total Hg (
r2 = 0.014), or the AE-Hg/total Hg ratio and TS (
r2 = 0.001). The plot of percent AE-Hg/total Hg vs. TS showed a quite scattered distribution, but the scattering converged to a much narrower range (average = 1.41 ± 0.63%) when the TS concentration exceeded a level of 12 mg/g (
Figure 1).
The total Cu concentration in the sediment samples ranged from 0.4 µg/g to 19.4 µg/g (average = 6.8 ± 4.5 µg/g,
n = 76). The AE-Cu in the sediment samples ranged from 0.04 µg/g to 3.66 µg/g (average = 1.05 ± 0.96 µg/g,
n = 76). The AE-Cu in the sediment ranged from 0.1% to 36.6% of the total Cu (average = 15.4 ± 9.8%,
n = 76). The percentage of AE-Cu among the total Cu, therefore, was significantly higher than that of AE-Hg among the total Hg. The average NAE-Cu was 5.75 ± 3.83 µg/g, or 84.6% of the total Cu. There was no correlation found between the AE-Cu and TS (
r2 = 0.08). The AE-Cu, however, had a weak correlation with the total Cu (
r2 = 0.403,
Figure 2) or with the AE-Hg in the sediments (
r2 = 0.370,
Figure 3). The total Cu also had a weak correlation with the TS (
r2 = 0.349,
Figure 4). The detailed spatial distribution of NAE/AE Hg and NAE/AE Cu and their relationship with the organic matter content and the texture of the sediments will be discussed in a separate paper.
For comparison, the Zn concentration in the sediments ranged from 9.0 µg/g to 198.5 µg/g (average 102.7 ± 67.3 µg/g,
n = 41). The AE-Zn was, on average, 77.7 ± 24.9 µg/g, or 75.7% of the total Zn, a much higher percentage than those of the AE-Hg or the AE-Cu (2.1% or 15.4%, respectively, of the total), but within the range of other studies [
8,
9].
4. Discussion
Evidence of monovalent Cu(I) in copper sulfide has been reported in previous studies [
20,
21]. This study confirmed that copper sulfide precipitated from solutions is monovalent Cu(I) bi-sulfide, (Cu
+)
2SS
=, rather than divalent Cu(II) mono-sulfide (Cu
++S
=), although both have the same atomic S/Cu ratio. Why did Cu (II) ions in solution react with H
2S to form monovalent cuprous bi-sulfide instead of divalent cupric mono-sulfide? The following may provide an explanation: In a sulfidic environment, cupric ion (Cu
++) was reduced with mono-sulfide (HS
−) to form cuprous ion (Cu
+) and elemental sulfur (S
o). The elemental sulfur reacted with sulfide to form bi-sulfide (SS
=). Finally, cuprous ions react with bi-sulfide ions to form cuprous bi-sulfide:
The stepwise reactions of (1), (2), and (3) must be tightly coupled before Cu+ had a chance to disproportionate into Cu++ and Cu0 (2 Cu+ = Cu++ + Cu0) as well as So to polymerize into hydrophobic S8o (8So = S8o).
The above reactions are similar to the stepwise formation of pyrite (ferrous bi-sulfide) from solution by the reaction of ferric ions and mono-sulfide, as indicated in other studies [
22,
23,
24,
25]:
The formation of the Hg(I) bi-sulfide could be explained with a similar mechanism:
Hg and Cu may substitute Fe in pyrite to form solid solutions in sediments. However, Hg and Cu ions have higher water exchange rates than ferrous ions, so they tend to form discrete sulfides in sediments [
9]. This study showed that both discrete Hg sulfide and Cu sulfide are NAE because they are bi-sulfides. These results explain the results of Huerta-Diaz and Morse [
8] that a high proportion of NAE (non-reactive) Hg was found in sediments even with very low DOP. Mercury has the highest priority among metals to form sulfide in sediments because the pk
sp of HgS is 52.7, a much higher value than those of other metal sulfides (22–36.1) [
26]. Copper has the second highest priority among metals to form sulfide (pk
sp of CuS = 36.1) in sediments.
I could not find the values of ksp for mercury or copper bi-sulfides in the literature. I suspect that the ksp of mercury or copper sulfide reported in the literature were actually their bi-sulfides. (The commercially available black mercury specimen was labeled as Hg(II) mono-sulfide but tested to be Hg(I) bi-sulfide in this study.) This study failed to produce any Hg or Cu mono-sulfides via precipitation from solutions. Based on all the evidence, I concluded that all precipitated Hg and Cu sulfides in sediments are bi-sulfides (or maybe some poly-sulfides), which are NAE. The separation of Hg and Cu sulfides from their other chemical forms in sediments, therefore, can be performed with a simple AE procedure under nitrogen, as outlined in this study. The NAE Hg or NAE Cu in sediments are sulfides and can be excluded from the bioavailability assessment. The AE Hg and AE Cu were relatively low (on average 0.99 ± 0.51 ng/g or 2.1 ± 2.7% of the total Hg and 1.05 ± 0.96 µg/g or 15.4 ± 9.8% of the total Cu) in the sediments of Apalachicola Bay. Although AE Hg and AE Cu are non-sulfides, they are not necessarily bioavailable in sediments. The AE-Hg and AE-Cu levels in sediments, however, can be considered the upper limits of their bioavailable fractions. The AE Zn in the sediments (on average 77.7 ± 24.9 µg/g, or 75.7 ± 24.3% of the total Zn) was much higher than those of the AE Hg or AE Cu. The AE Zn, however, cannot be considered the non-sulfide fraction of Zn because discrete Zn sulfide is also AE.
The chemical reactions in Equations (1) and (9) can proceed spontaneously to the right-hand side under sulfidic conditions, according to the thermodynamic potentials. This implies that in a sulfidic solution, mercuric mono-sulfide (Hg++S=) or cupric mono-sulfide (Cu++S=) are not stable because they would form thermodynamically more stable Hg and Cu bi-sulfides. In theory, Hg(I) and Cu(I) could form stable mono-sulfides (Hg2S and Cu2S), which are AE with a S/metal atomic ratio of 0.5. This study, however, did not find any evidence of their existence. The reason for that could be because Hg (I) and Cu(I) bi-sulfides are much more stable than Hg(I) and Cu(I) mono-sulfides, respectively. There is a possibility that Hg(I) or Cu(I) can form poly-sulfides (Hg2SxS or Cu2SxS, x > 1) in sediments, which are also NAE. The pH 7.5 mercury sulfide had an atomic S/metal ratio of 1.29 (i.e., Hg2S1.58S), suggesting this possibility.
No significant correlation was found between the AE/NAE ratio of Hg and that of Cu (
r2 = 0.039) or the total Hg and the total Cu (
r2 = 0.035) in the sediments. The AE-Hg and AE-Cu in the sediments had a weak correlation (
r2 = 0.370,
Figure 1). Mercury in the sediments was not correlated with the TS except that the pattern converged to a lower value (average = 1.41 ± 0.63%) when the TS concentration exceeded 12 mg/g (
Figure 2). The affinity of Hg to sulfides is much stronger (pk
sp = 52.7) than that of Cu (pk
sp = 36.1) or Fe (pk
sp = 18.1 for mono-sulfide and pk
sp = 27 for pyrite). Mercury, therefore, has the priority, among all metals, to form sulfides in sulfidic sediments. This could explain why mercury was not correlated with the total sulfides in the sediments because almost all available Hg had already been sequestered as bi-sulfide, especially when the TS level exceeded 12 mg/g.
Total copper in the sediments was weakly correlated with the TS (
r2 = 0.349,
Figure 3). Apparently, the degree of Cu sequestration via sulfide is not as complete as that of Hg. As such, weak correlations existed between total Cu and total sulfide. AE-Cu was also weakly correlated with total Cu (
r2 = 0.403,
Figure 4).
Ravichandran et al. [
27] reported that Hg sulfide formation could also be interfered with using dissolved organic matter (DOM) to some extent. Could the highly scattered pattern of the AE Hg vs. TS plot (
Figure 2) be the result of the interference of DOM on Hg sulfide formation in the sediments? This speculation will be discussed in a separate paper.
The CRS in the sediments was six or three orders of magnitude greater than the mercury or the copper, respectively, on an atomic-scale basis. Even the AVS in the sediment was four or one order of magnitude greater than the mercury or the copper, respectively, on an atomic-scale basis. The vast majority of sulfides in the sediment, therefore, were associated with ferrous ions, which were much more abundant (on average 20.7 ± 9.4 mg/g, n = 41) than all the trace metals combined in the sediments.
We used Zn to represent the HMs that form mono-sulfides (AE) in sediments. The Zn in the sediments, therefore, had a much higher fraction of AE (on average 75.7% of the total Zn) in comparison with those of Hg (2.1% of the total Hg) or Cu (15.4% of the total Cu). The NAE Zn (on average, 24.3% of the total Zn) in the sediments was likely occluded in pyrite, as suggested by Huerta-Diaz and Morse [
8]. The AE Zn, however, cannot be considered as non-sulfide because discrete Zn sulfide is also AE.
DOM has been found to inhibit the precipitation and aggregation of the black HgS [
27] or affect the speciation, solubility, and bioavailability of Hg [
27,
28,
29,
30]. Those findings, however, should not affect the conclusion of this study that NAE Hg is sulfide and can be excluded from the bioavailability assessment.