7

Lake Bed Processes

This chapter examines the (bio)geochemical reactions that control the partitioning of lead (Pb), cadmium (Cd), zinc (Zn), and arsenic (As) between lake sediments and porewater, ultimately leading to the specific conditions of the Coeur d’Alene (CDA) Lake. After an overview of metal(loid) retention processes and the diagenetic (redox processes) reactions, metal deposition and migration within the Lake are examined. Using modeled and measured conditions, the operative processes controlling dissolved porewater metal(loid) contaminant concentrations and possible entry into Lake waters are further assessed. The chapter ends with an assessment of the potential for eutrophication and pH changes to release metals from the Lake sediments into the water column.

INTRODUCTION TO BIOGEOCHEMICAL PROCESSES IN LAKE SEDIMENTS

The heavy metals cadmium, lead, and zinc and the metalloid arsenic all have the commonality of being present within the CDA River and CDA Lake due to their presence in sulfidic minerals of the mined orebodies. They are also all subject to changes in partitioning between the solid and aqueous phases, with biogeochemical transformations occurring within the water column and the sediments. Under sulfidic conditions, they all form mineral sulfides of limited solubility, while under oxic conditions their dissolved concentrations are largely controlled by binding to metal (hydr)oxides, principally those of ferric iron [Fe(III)]. Owing to the dynamic cycling of Fe(III) phases and sulfides with changes in oxygenation and nitrate concentrations, the dissolved concentrations of all four of the metal(loid)s will be subject to changes in the oxygen status of the water column and underlying sediments. Further complicating the behavior of arsenic is its changes in oxidation state with the redox conditions of the water and sediments.

This section provides the principal biogeochemical factors that drive the cycling and dissolved concentrations of arsenic, cadmium, lead, and zinc, and the critical control that Fe(III) phases exert as a function of oxygen level and pH on the dissolved concentrations of all four contaminants in CDA Lake waters and sediments.

Cadmium, Lead, and Zinc

Cadmium, lead, and zinc are all hydrolyzable divalent cations within conditions of the surface environment. They all have a strong affinity for sulfide (they are chalcophiles), forming metal sulfides of limited solubility and making them highly sensitive to dissolved sulfide concentrations. The primary control on dissolved sulfide

formation is microbial dissimilatory sulfate reduction (where sulfate is used in place of oxygen for respiration), driven largely by obligate anaerobic bacteria (Bradley et al., 2011). Dissimilatory sulfate-reducing bacteria (SRB) are nearly ubiquitous within lake sediments, and their activity and associated sulfide production is primarily dependent on oxygen levels, with upregulation of sulfate reductase at trace oxygen levels. Secondarily, their activity will then depend largely on temperature and microbially available organic carbon used in respiration (Rabus et al., 2013). SRB production of sulfide may then be scavenged by the differing metals within the sediments, making the total supply, as noted by dissolved concentrations, the important determinant of mineral sulfide formation. If, for example, the porewater sulfide concentration is 10 μM and the pH is 7, total dissolved Pb (largely as the neutral Pb(HS)₂⁰ aqueous complex) and Zn (largely as Zn²⁺ and ZnS⁰) will be approximately 1 pM (0.21 ng Pb/L, 0.065 ng Zn/L), while total dissolved Cd will be slightly higher at 10 pM [> 0.11 ng/L, largely as Cd(HS)₂⁰] and be controlled through the formation of PbS (galena), ZnS (sphalerite), and CdS (greenockite) solids, respectively.

Under oxygenated conditions, sulfide production is limited and mineral sulfides undergo oxidative dissolution. Ferric oxides, oxyhydroxides, and oxides (hereafter collectively referred to as iron oxides) are instead formed and often become the dominant regulator of dissolved metal concentrations (Stumm et al., 1992). Adsorption reactions on iron oxides, however, are pH-dependent and vary for the three metal contaminants here. Typical of cationic metals, the extent of adsorption increases with increasing pH and undergoes a rapid shift in partitioning across a small pH change (Figure 7-1). Within 1 to 2 units of pH change, the metal contaminants will go from nearly entirely aqueous (low pH) to entirely adsorbed (high pH); the pH at which the transition occurs is termed the pH edge. The pH edge for lead adsorbing on iron oxides occurs between pH values of 4.5 and 5.5. Thus, at most pH values encountered in the environment, and clearly those within the Coeur d’Alene system, lead would be partitioned on iron oxide surfaces provided that (1) the sediments are oxidized, (2) sufficient iron is present to generate iron oxides with the capacity to adsorb the quantities of lead, and (3) competing ions for the iron oxide surface do not limit lead adsorption. Given the high affinity of lead for iron oxides (Dewey et al., 2021), competing ions typically do not limit uptake.

The adsorption edge for zinc on iron oxides is substantially higher than for lead, occurring in the range of 6.0–7.0. Additionally, the affinity of zinc for the iron oxides surfaces is less than for lead, making it subject to competitive displacement. Nevertheless, iron oxides often are regulators of dissolved zinc concentrations.

Cadmium has the highest adsorption edge of the three heavy metal contaminants, occurring in the pH range of 6.5–7.5; it also has the lowest affinity for iron oxides and consequentially is not controlled by surface adsorption at the low concentrations noted for zinc and, most appreciably, lead. Critically, for systems such as CDA Lake that have pH values near neutral, slight changes in pH may have large impacts on the dissolved concentrations of zinc and cadmium under conditions where adsorption to iron oxide minerals is important.

Cadmium, lead, and zinc may also form metal carbonates, and although more soluble than their sulfide mineral cousins, they may regulate the dissolved concentrations of these metals. Although metal coprecipitates or hydroxo carbonates, such as hydrocerussite, may lead to lower concentrations, considering the pure metal carbonate phases provides a useful reference point for the comparison in solubility of differing phases. At pH 7 and total dissolved carbonate concentrations of 2 mM, total dissolved lead and zinc would be near 1 μM (202 ppb and 65 ppb, respectively) while cadmium would be at 0.1 μM (66 ppb) if in equilibrium with PbCO₃ (cerussite), ZnCO₃ (smithsonite), and CdCO₃ (otavite), respectively. The metal carbonates should serve to regulate the upper limits of dissolved lead, zinc, and cadmium concentrations. Metal carbonates have, in fact, been noted to control the dissolved concentrations of contaminants in mining-influenced sites. For example, cerussite (PbCO_{3, crystalline}) was noted as a major secondary weathering phase in sediments influenced by a historic mining district of Wanlockhead, Scotland (Hillier et al., 2001).

Arsenic

Two oxidation states of arsenic, arsenate [As(V)] and arsenite [As(III)], predominate in surface and near-surface environments. At circumneutral pH, the partially protonated forms of arsenate, H₂AsO₄⁻ and HAsO₄²⁻, and fully protonated form of arsenite, H₃AsO₃⁰, dominate. Microbial activity may methylate As(V) or As(III), forming, for example, dimethylarsenic acid and monomethylarsonous acid (Cullen and Reimer, 1989). Thio- and carbonatocomplexes of arsenic also exist within anaerobic systems; thiolated forms of arsenic may, in fact, represent an important reactive component within sulfidic environments (Wilkin et al., 2003; Wang et al., 2020).

Partitioning of arsenic onto soil solids is foremost dependent on its oxidation state. In general, As(V) binds extensively and strongly to most mineral constituents of soils and sediments, often partitioning most appreciably on iron oxides, while As(III) retention is more convoluted and dependent on specific soil chemical conditions. As a consequence, under oxidizing conditions where arsenate predominates, dissolved arsenic concentrations are limited except at very high pH (> 8.5). Phosphate, having analogous binding properties to arsenate, also is strongly retained by mineral surfaces except at very high pH. The surface complexes of arsenite, although extensive, are far more labile than for its oxidized counterpart, leading to higher dissolved arsenite concentrations and a greater propensity for migration (Kocar et al., 2006; Tufano et al., 2008).

In surface and subsurface environments, changes in water chemistry often result in release of arsenic from solid phases through various desorption pathways. Processes leading to arsenic desorption can broadly be grouped into four categories: (1) ion displacement, (2) alkalinity (pH values > 8.5), (3) reductive dissolution of As(V)-bearing Fe(III) oxides, and (4) oxidative dissolution of arsenic sulfide phase. Although the latter process will increase dissolved concentrations of arsenic, the typical concomitant generation of Fe(III) oxides re-partitions arsenic to the solid phase as As(V) adsorbed on Fe oxides. Furthermore, the pH values of the Coeur d’Alene system negate the likelihood of alkalinity-promoted desorption. Competitive ion displacement can represent an important means by which arsenic is released to the aqueous phase, but it is most appreciable in regions where extensive fertilizer or pesticide runoff occur (Jain and Loeppert, 2000; Peryea and Kammerack, 1997), and hence it is likely to be limited in comparison to that imposed by the onset of anaerobic conditions. Thus, the process of greatest concern is the reductive dissolution of As(V)-bearing Fe(III) oxides.

Within most soils and sediments, total arsenic levels correlate with iron content rather than aluminum or clay content (Smedley and Kinniburgh, 2002), and thus reductive dissolution/transformation of Fe(III) phases should have a major impact on arsenic. As a consequence, the greatest likelihood for arsenic release in soils and sediments typically occurs upon a transition from oxidizing (oxygenated/aerobic) to reducing (deoxygenated/anaerobic) conditions. Arsenic may be displaced either through reduction of arsenate to arsenite or through reductive dissolution of Fe(III) oxides. Microbially driven oxidation of organic carbon coupled to the dissimilatory reductive dissolution

of As-bearing Fe oxides causes the transfer of arsenic from sediment solids to groundwater (Islam et al., 2004; Fendorf et al., 2010), as shown below:

CH₂O + 4FeOOH-(H₂AsO₄)_x + (7+3x)H+ 4Fe²⁺ + HCO₃^- + (6+x)H₂O + xH₃AsO₃

where CH₂O generically represents organic carbon and may include other fermentation products such as H_2(aq), As (as arsenate) is bound to sedimentary iron oxide (goethite as written in reaction 1), and x is the stoichiometric coefficient of arsenic content associated with the iron oxides. Dissimilatory As(V)/Fe(III) reduction requires (1) anaerobic conditions with low sulfate supply, (2) reactive As-Fe complexes, and (3) microbially available organic carbon.

Microbial Transformation of Iron and Arsenic

Under oxidizing conditions created largely by oxygen or nitrate, iron generally exists as Fe(III), forming Fe(III) oxides sparingly (Kappler et al., 2021). Cycling of iron oxides can exert a dominant control on nutrient and contaminant mobility and bioavailability due to their large surface areas and high reactivity (Cornell and Schwertmann, 2003). For example, within the large watersheds of the Ganges-Brahmaputra, Mekong, and Red Rivers in southeast Asia, iron oxides are the dominant hosts of arsenic in oxygenated surface environments (Fendorf et al., 2010). Accordingly, dissolution and transformation of iron oxides can have major implications for the fate and transport of metal(loid) contaminants (see, e.g., DeLemos et al., 2006); concomitant with the dissolution (inclusive of transformation) of iron oxides, adsorbed or coprecipitated contaminants may be mobilized.

Under oxygen-limiting conditions, microorganisms have the capacity to use alternate electron acceptors (to oxygen) in respiration, such as Fe(III). The general sequence of electron acceptors is O₂ followed by nitrate, manganese (Mn) dioxides, iron oxides, sulfate, and methanogenesis (Figure 7-2)—although the potential for iron oxides and sulfate often overlap, with As(V) reduction occurring concomitantly with Fe(III) reduction (Postma and Jakobsen, 1996; Kocar et al., 2008).

A host of bacteria and archaea can respire by coupling organic carbon or a fermentation product with Fe(III) oxides (processes known as dissimilatory iron reduction), reducing Fe(III) to Fe(II) (Gorby and Lovley, 1991; Nealson and Saffarini, 1994; Kappler et al., 2021) and releasing Fe²⁺ and adsorbed phases to porewater. Respiratory reduction of iron in sediments generally occurs in zones where O₂, NO₃⁻, and Mn(IV) are diminished (Lovley, 1997), as depicted in Figure 7-2.

Dissimilatory iron-reducing bacteria upregulate iron reductases at low concentrations of O₂ and depletion of nitrate (Kappler et al., 2021). Given atmospheric supply of O₂, and nitrate from the water column, sediments or bottom waters will have a depth-stratified succession in iron oxidation and reduction. Surface layers supplied with O₂ and nitrate will lead to the persistence or generation of Fe(III) oxides. Microbial utilization of O₂, followed by nitrate and then Mn(III/IV) oxides, will lead to anaerobic conditions and concomitant Fe(III) and As(V) reduction (Figure 7-2). The transition depth at which iron goes from being oxidized (by O₂ or nitrate) to being reduced (by iron-reducing bacteria) is largely dependent on the supply and demand of O₂ (Figure 7-2). The supply side is controlled by the diffusion through the water column while demand is dependent on microbially available organic carbon, temperature, and limiting nutrients (if any are limiting). Increased biological demand in the water column or sediments will lead to the drawdown of oxygen higher in the profile, moving the iron oxide reduction boundary higher in the sediment profile. If oxygen demand is limited by nutrient or organic carbon supply, then the reduction boundary will move lower in the profile and a greater zone of iron oxides will be generated. Provided sufficient iron oxide content and pH values that promote metal(loid) retention, contaminants such as arsenic, lead, and zinc, and potentially cadmium, will be retained by the iron oxide via adsorption. By contrast, if bottom waters become anoxic, the iron-reduction zone can move close to the water–sediment interface. In such cases, Fe²⁺ along with contaminants on the surface of the Fe oxides may then freely diffuse into the water column.

Possibly limiting metal(loid) release under anaerobic conditions is the generation of sufficient sulfide (from dissimilatory sulfate reduction) to result in metal(loid) sulfide formation, as noted above. However, many freshwater systems are sulfur-limited (compared to metal demand), which can limit secondary sequestration of contaminants in sulfidic phases (e.g., O’Day et al., 2004); this appears to be the case for CDA Lake (Toevs et al., 2006).

METAL(LOID) DYNAMICS WITHIN CDA LAKE SEDIMENTS

This section examines metal deposition and migration within the CDA Lake sediments using modeled and measured conditions, in order to further assess the operative processes controlling dissolved porewater metal(loid) contaminant concentrations and possible entry into Lake waters.

Information from Sediment Core Studies

Metal Deposition

Effective impoundment of milling wastes was largely completed by 1968 (Morra et al., 2015), but the historic deposition along the river corridor remains a source of metal deposition to CDA Lake. The stockpile of contaminated sediments upstream from the Lake remains 100 times more enriched than background conditions 52 years after the first effective constraints on primary waste inputs were imposed (Sprenke et al., 2000; Langman et al., 2020). Bookstrom et al. (2013) estimated that in 1998 about one-third of the lead load entering CDA Lake in any year was derived from the upper basin, with a small fraction derived from erosion of mainstem river banks (lower basin floodplains); the large majority entering the Lake was probably from the mainstem river bed. Bookstrom et al. (2013) note that as the metal-enriched sediments in the riverbed progressively undergo remobilization and dilution during transport, they will re-deposit to form and re-form persistent “hot spots.” The hot spots will continue to function as secondary sources of lead and other metals for the Lake.

In depositional environments of the Lake, contaminated sediments are slowly buried by successively younger layers of progressively diluted and less metal-enriched sediments. Indeed, sediments deposited on lower basin floodplains averaged 5,400 ± ~ 2,000 μg/g lead between 1903 and 1980, 4,000 ± 1,000 μg/g lead between 1968 and 1980, and 3,100 ± 1,950 μg/g lead between 1980 and 1998 (Figures 21, 23 in Bookstrom et al., 2013).

Furthermore, on the basis of two cores collected in 2010, Morra et al. (2015) illustrate that there was a distinct drop in sedimentation rates in the Lake after impoundment was completed in 1968. Those rates remain low, at 0.1 g cm⁻² yr⁻¹ compared to the maximum of 0.8 g cm⁻² yr⁻¹. Using distinct markers within the sediment profile, such as the eruption of Mount St. Helens in 1980, provides easy demarcation of sedimentation rates and metal loading over different time intervals (see Bookstrom et al., 2013; Morra et al., 2015). Zinc and cadmium have declined in concentration in lake sediment cores over the past few decades. Cadmium concentrations were 80 μg/g prior to 1968 and dropped to an average of 25.6 μg/g in the 1980–2010 interval; zinc concentrations similarly declined, to an average of 3,350 μg/g in the 1980–2010 interval. Concentrations of arsenic were at a maximum in the surface sediments, probably as a result of diagenesis. Lead concentrations, however, were not different in the 1980–2010 segment compared to the 1980–1998 segment reported by Bookstrom et al. (2013; ~ 3,000 μg/g dw).

Estimates using different methods show that CDA Lake continues to be a net sink for metals coming in from the watershed. Using cores from 2010, Morra et al. (2015) estimated that annually 91 percent of incoming lead was retained, and 56–58 percent of the zinc and 30 percent of dissolved cadmium that enter the Lake are retained. Based upon fluxes as calculated in Chapter 3, trapping of zinc ranged from ~ 45 to 65 percent between 2010 and 2015; furthermore, although lead retention in the Lake is highly variable from year to year, the ratio of lead export to import is decreasing. Trapping has increased since the 1990s from as low as 80 percent in 1991 to 93–98 percent between 2015 and 2020 (consistent with Morra et al., 2015). This is a positive development for lead not being exported downstream, but its meaning for future trends in dissolved lead concentrations in the Lake is unclear, depending upon the degree of internal flux from sediments to the water column. Box 7-1 further explores the trend of increasing CDA Lake lead trapping efficiency.

The Bookstrom et al. (2013) and Morra et al. (2015) studies suggest some progress in declining metal inputs to the sediments of CDA Lake. However, as noted in Morra et al. (2015), the surface sediment concentrations still exceed EPA’s 50th percentile probability of adverse biological effect by approximately tenfold or more and exceed background concentrations by 10–100 times (Table 7-1). Bookstrom et al. (2013) noted that natural processes that would be expected to reduce inputs to CDA Lake will take centuries, “depending on the size and metal contents of secondary sources, and the size and discharge of the drainage basin they occupy.” Coulthard and Macklin (2003) modeled long-term contamination in river systems from historical metal mining showing that more than 70 percent of the deposited contaminants remain within river systems for more than 200 years after mine closure. Natural recovery of the Rio Tinto in Spain and its floodplain, a river originally mined by the Romans (during the Copper Age), took 450 years.

Another example is provided by Diamond (1995), who used a mass-loading model to illustrate one scenario for the fate of water quality in Lake Moira, Ontario, a lake subject to high mass inputs of metals from mining wastes (Box 7-2). In this case, metal trapping, mixing and redistribution within the lake greatly delayed responses of the lake to changes in arsenic inputs once the legacy of contamination was established in lake sediments. CDA Lake is a deeper system, with longer residence times, than Lake Moira. The fate of arsenic is to some degree specific to that metalloid as well. Nevertheless, the example presented by Diamond (1995)

TABLE 7-1 Concentrations of Selected Metal(loid)s in Cores from Peaceful Point Showing Concentrations within Surface Sediments

^a SOURCE: Horowitz et al. (1993).

^b SOURCE: Morra et al. (2015).

clearly demonstrates that a lake’s response to changes in inputs can lag by a considerable period (decades). Moreover, CDA Lake remains in a phase of mass metal influx exceeding mass metal export (see Box 7-1). Thus, the Lake remains a net metal sink and is accumulating lead, zinc, cadmium, and arsenic. Morra et al. (2015) concluded that “without a concerted effort on remediation to prevent silt-sized (Toevs et al., 2008), lead-laden particles from entering the lake, lead concentrations in the sediment will remain at toxic levels with substantial potential to cause negative environmental impacts.”

It is worth noting that the core dating from Morra et al. (2015) and Horowitz et al. (1993) was performed from localized, highly contaminated areas near the delta region of CDA River (see Figure 7-3 for a map). Areas that see more frequent anoxia relative to Peaceful Point could have a more depleted Fe(III) oxide layer or greater concentration and migration of redox active elements within the upper 10 cm of sediments. Likewise, areas with differing sedimentation influences (stream inputs or bays) may show decreasing, rather than constant, deposition of lead since 1968.

Metal Associations within the Lake Sediments

The metal(loid)s deposited in the lake sediments may associate with different solids or partition into the porewater, with the possibility of dynamic cycling between different solids and the aqueous phase. A number of studies have examined the partitioning of the metals and their solid phase associations in the sediments of CDA Lake (see Balistrieri and Blank, 2008; Bostick et al., 2001; Harrington et al., 1998a,b; Haus et al., 2008; Horowitz et al., 1993; Kuwabara et al., 2003; Moberly et al., 2009, 2016). Horowitz et al. (1995) found lead, cadmium, zinc, and arsenic in extractions that target iron oxides, with only minor amounts of these metals in extractions that seek to target sulfide minerals (Horowitz et al., 1993, 1995). Additionally, studies of mixing experiments of acidic and metal-contaminated water with uncontaminated surface waters illustrate metal binding to iron oxides (Balistrieri et al., 1999, 2003; Paulson and Balistrieri, 1999; Tonkin et al., 2002). By contrast, Harrington et al. (1998a) found that heavy metals in the delta region of CDA Lake were extracted dominantly using a procedure targeting sulfide minerals (reaction with KClO₃ as an oxidant followed by reaction with concentrated HCl and then 4 M HNO₃ while boiling). Examining the molar masses of the potential sulfur-binding metals (Pb, Cd, Zn, and Fe) and arsenic relative to sulfur helps illustrate the likely dominant phases controlling the contaminants. Table 7-2 provides the elemental abundance of the heavy metals along with iron, arsenic, and sulfur in sediment taken from two locations within the Lake in 2002. These are Peaceful Point and Harlow Point, less than 1 km from the mouth of the CDA River, as shown in Figure 7-3.

TABLE 7-2 Elemental Concentrations in Near-Surface Sediments of CDA Lake

SOURCE: Toevs et al. (2006). Reprinted (adapted) with permission from Toevs, G. R., M. J. Morra, M. L. Polizzotto, D. G. Strawn, B. C. Bostick, and S. Fendorf. 2006. Metal(loid) diagenesis in mine-impacted sediments of Lake Coeur d’Alene, Idaho. Environ. Sci. Technol. 40(8):2537–2543. https://doi.org/10.1021/es051781c. Copyright 2006 American Chemical Society.

Summing the molar abundances of arsenic, cadmium, lead, and zinc (and neglecting iron) yields values ranging from 60 to 80 mmol/kg across depths for the two sites, which are slightly less than the range of total sulfur that varies from ~ 100 to 120 mmol/kg. Thus, there is sufficient total sulfur such that if it is all in sulfidic form, monosulfide phases such as CdS, PbS, and ZnS could form and control dissolved metal concentrations. However, when considering the molar abundance of iron in the sediments, the projections change substantially. If one assumes all of the sulfur were present in sulfidic form and FeS₂ is considered as a dominant sulfur product, there is only sufficient sulfur to react with ~ 2–4.6 percent of the total iron. Furthermore, Toevs et al. (2006) found that the maximum amount of total sulfur in sulfidic form (within the upper 36 cm) was just less than 50 percent, indicating that there is sufficient sulfide to react with only 1–2.3 percent of the total iron. Consistent with this estimate based on mass ratios, Toevs et al. (2006) found that FeS₂ was the main sulfide mineral in contaminated areas, increasing with sediment depth to a maximum of ~ 50 percent of total sulfur at ~ 36 cm, and that only 2–2.5 percent of the total iron is associated with pyritic minerals. Winowiecki (2002) reports similar trends in pyrite with depth, and in iron association with pyrite, using acid volatile sulfide and sulfur X-ray absorption near-edge structure spectroscopy for lake sediments from contaminated and uncontaminated sites sampled between 2000 and 2002. Thus, it appears that sulfidic phases are unlikely to be major hosts of heavy metal contaminants within CDA Lake, although they may vary and have locally important impacts.

Rather than a dominance of metal sulfides, Toevs et al. (2006) show that Fe(III) oxides and Fe(II) as siderite (FeCO₃) are present down to at least 36 cm, with the proportion of Fe(III) (hydr)oxides, relative to siderite, decreasing with depth. Although the redox gradient produces dominant Fe(II) forms, Fe(III) phases persist beyond the oxic upper layer (Toevs et al., 2006). The reactive transport modeling of S‚ engör et al. (2007) suggests the dominant role of Fe(III) oxides in metal and arsenic partitioning with the solids of the upper layer (top 30 cm) of the sediments. Additionally, in experimental work conducted by Balistrieri et al. (2003), correlation between metal concentrations in sediments and those predicted by the formation of ferrihydrite (nominally ferric hydroxide) suggest that arsenic is predominantly associated with iron oxides, with lead and cadmium also associating largely with iron oxides at low to moderate loadings, while other phases (such as metal sulfides or carbonates) contribute at the highest metal loadings.

Modeling and Measurement of the Processes Controlling Metal Binding

The data from sediment cores suggest that Fe(III) oxides have a dominant role in controlling the dissolved concentrations of the heavy metal contaminants. The iron mineral ferrihydrite is likely the predominant phase within the surface and thus, based on ion affinities, is a prime determinant of dissolved metal concentrations.

TABLE 7-3 Surface Complexation Reactions and Optimized Binding Constants

SOURCE: Nomaan et al. (2021). Reprinted from Chemical Geology 573, Nomaan, S. M., S. N. Stokes, J. Han, and L. E. Katz. Application of spectroscopic evidence to diffuse layer model (DLM) parameter estimation for cation adsorption onto ferrihydrite in single-and bi-solute systems. https://doi.org/10.1016/j.chemgeo.2021.120199. Copyright 2021, with permission from Elsevier.

The binding and related dissolved concentrations of metal(loid) contaminants resulting from reaction with iron oxides was examined by the committee using the surface complexation model (SCM) described in Nomaan et al. (2021) with the physical-chemical characteristics of CDA Lake sediments described in S‚ engör et al. (2007). The surface reactions and binding constants for cadmium, lead, and zinc are provided in Table 7-3 and are taken from Nomaan et al. (2021); the reaction and binding constants for arsenic are from Dzombak and Morel (1990). The model includes the simultaneous reaction of all three metals and arsenic with the iron oxide surface. Although ignoring the full suite of reactions that control solute concentrations, the SCM illustrates the extent to which iron oxides may control heavy metal and arsenic concentrations in porewater (Figure 7-4). (Note that this exercise could also have been done with other models that capture transport processes, such as the USGS software PHREEQC.)

There are two striking outcomes from the SCM (Figure 7-4). First is the role pH plays in the binding of cadmium, lead, and zinc to ferrihydrite. The near-neutral pH of the Lake sediments is a critical threshold, such that even slight changes in pH will have pronounced impact on dissolved concentrations for cadmium and zinc. Second, arsenic, when residing as As(V), is completely bound to the ferrihydrite, with nondetectable dissolved concentrations. While highly sensitive to pH change, at pH 7.2, dissolved lead (< 1 μg/L) and cadmium (~ 1 μg/L) concentrations are also predicted to be near the limits of detection. By contrast, even at pH 7.2, a zinc concentration of 1,100 μg/L (16.8 μM) is projected, consistent with measured porewater concentrations (Harrington et al., 1998a; Balistrieri, 1998).

The projected regulation of dissolved metal concentrations by Fe(III) [and Mn(III/IV)] oxides is consistent with measured porewater concentration (Balistrieri, 1998; Winowiecki, 2002; Toevs et al., 2006). Moreover, Balistrieri (1998) allows estimates of benthic fluxes from the sediment into the Lake bottom water. Albeit with sparse temporal and spatial data, Balistrieri (1998) projected that sediments would provide a positive flux only for zinc. However, due to the dynamic biogeochemical cycling of the Lake sediments, seasonal alteration of environmental conditions and substantial presence of various metals and inorganic and organic sulfur and carbon create a dynamic source of potential metals for the overlying Lake water, in both dissolved and colloidal forms.

Despite being highly simplistic, the sediment diagenesis reactive transport model developed by S‚ engör et al. (2007) provides a useful means of further examining the processes likely controlling metal contaminant fate within CDA Lake. The predominant biogeochemical processes for anaerobic diagenesis encapsulated within the model are shown in Figure 7-5 and the biogeochemical outcomes are presented in Figure 7-6. The model is predicated largely on the field analysis conducted by Cummings et al. (2000) on microbial Fe(III) reduction in CDA Lake sediments, the sediment characterization by Toevs et al. (2006), and porewater data from Winowiecki (2002) and Balistrieri (1998).

The biogeochemical (diagenetic) processes (as modeled in S‚ engör et al., 2007; Figure 7-6), supported by depth profiles (particularly the detailed core analysis of Morra et al., 2015, shown in Figure 7-7) paint a picture of metal(loid) cycling within the near-surface sediments. As shown schematically in Figure 7-8, over the approximate upper 30 cm, oxygen introduced from the Lake water (which in turn is introduced from the atmosphere) is consumed, increasingly leading to anaerobic processes (with depth) in the sediments. Iron(III) oxides occur throughout the upper sediments and largely dominate the retention of lead, zinc, and cadmium. Despite dissimilatory Fe(III) reduction proceeding with depth, and the generation of diagenetic pyrite, lead, zinc, and cadmium depth abundances are unchanged (Figure 7-7), indicating that metal repartitioning within the profile is not a major factor. Porewater concentrations of these three metals are unlikely to vary greatly with depth (as noted by the SCM results; Figure 7-4), where only zinc is appreciable in the aqueous phase.

The profiles for the redox active elements stand in stark contrast to lead, zinc and cadmium, which do not undergo redox transformation. The profiles of iron and manganese (Figure 7-7) indicate upward migration within the near-surface, oxygenated sediment, which is particularly apparent relative to depositional fluxes (Morra et al., 2015). The depth profile of arsenic is particularly striking in its surface accumulation, even in comparison to iron and manganese, which results from a combination of ferrihydrite accumulation and the formation of As(V), which has a greater affinity for Fe(III) minerals (Tufano et al., 2008). The combined reduction of As(V) [to As(III)] and Fe(III) results in a weaker partitioning of As(III) and increased dissolved concentrations. Upward diffusion of arsenic leads to As(III) oxidation to As(V) [coupled with Fe(II) oxidation] and the concomitant strong partitioning back onto the solids (see Figure 7-8). As an outcome, arsenic has a distinct peak abundance in the upper profile (Figure 7-7) that results from biogeochemical processes rather than deposition (Morra et al., 2015). The buildup of arsenic in the near-surface sediments also illuminates the concern of bottom-water anoxia. If the sediments become devoid of oxygen, arsenic will no longer repartition in the near-surface and would rather diffuse into the overlying lake water.

Manganese is more sensitive to low oxygen levels than arsenic or iron (Stumm and Morgan, 1981; Davison, 1993) owing to a higher redox potential (Figure 7-2). Reduction of manganese oxides, prevalent under oxygenated conditions, leads to their dissolution and resulting Mn(II). Accordingly, dissolved concentrations of Mn(II) within bottom waters may serve as a sentry for the potential release of arsenic, indicating when the oxygen-created Fe/Mn oxide barrier at the surface sediments is breached. Indeed, large spikes in filterable¹ manganese in bottom waters occur in association with anoxia at C6 in CDA Lake (Figure 7-9A). Low-concentration spikes of filterable manganese also occur coincident with hypoxia² in the Lake bottom at C5 (e.g., Figure 7-9B).

Arsenic is also seen in bottom waters coincident with seasonal anoxia. Chess (2021) showed that release of arsenic occurred from the lateral lakes during periods of anoxia. Figure 7-10 shows a peak in arsenic concentrations in bottom waters of C6 coincident with anoxia in two examples (2012 and 2019). Concentrations of arsenic in bottom sediments of the St. Joe River watershed are not enriched in arsenic compared to locations in CDA Lake affected by mine wastes (Toevs et al., 2006). Nevertheless, dissolved arsenic in C6 bottom waters reached 5–10 μg/L during anoxia (see Figure 7-10).

C5 is the mine-effected monitoring location in CDA Lake where the seasonal sag in oxygen concentration is the largest. The cycle of arsenic in the waters at C5 also reflects hypoxia to some degree (Figure 7-11). Elevated concentrations of dissolved arsenic in bottom waters at C5 also coincide with the fall oxygen sag, but peak concentrations are ~ 10 percent of those at C6, coincident with less hypoxia (Figure 7-11). Detectable arsenic concentrations were consistently available only from 2018 to 2020. Concentrations of arsenic in mine waste-impacted surface sediments in CDA Lake are typically 10–20 times higher at C5 than at the mouth of the St. Joe River. It would be important to understand how these higher arsenic concentrations in sediment would affect peak arsenic concentrations in bottom waters under anoxic conditions.

___________________

¹ Filterable can include dissolved Mn(II) or nano-, colloidal-size Mn particles.

² The term “hypoxia” is used to describe situations in which oxygen is depleted below saturation levels (saturation refers to the maximum amount of oxygen water can hold as a function of temperature and pressure), and that depletion is detrimental to organisms of interest.

The biogeochemical cycling drawn from modeling and measurement lead to four principal conclusions. First, Fe(III) oxides are dominant phases restricting the dissolved concentration of heavy metals in the upper 35 cm of the lake column. Second, reductive dissolution of the Fe(III) phases transpires in the sediments, becoming increasingly prominent with depth. Third, generation of sulfide (through sulfate reduction) would likely fail to sufficiently sequester heavily metals owing to the large Fe:S ratio within the CDA Lake system, consistent with the conclusions of Toevs et al. (2006). Fourth, alterations to lake chemistry that (1) further promote Fe(III) reduction (such as periods of anoxia) and/or (2) stimulate processes that acidify (even slightly) the sediment porewater, may lead to release of metal(oid)s from the sediment and thus enhance their concentrations in the water column.

KEY ISSUES ABOUT SEDIMENT PROCESSES UNDER CHANGING CONDITIONS

The previous sections suggest that CDA Lake sediments could affect Lake water quality in the future if either enhanced anoxia or lowering of pH of the Lake sediments were to occur. Both enhanced anoxia and lowering of pH have the potential to release phosphorus and/or metals from sediments, as discussed below.

Changes Promoting Anoxia

Loss of oxygen from the bottom waters and sediment porewaters of CDA Lake could lead to geochemical changes that release metals from the solid phases in which they reside in the sediment. The primary concern is the upward migration of arsenic into the water column if oxygen levels were to decline at the sediment–water interface. The extent of release is proportional to the concentration of the arsenic in the sediment and length of time the bottom waters and sediment–water interface are under anoxic conditions. Because of the abundance of iron oxide in the sediments, oxygen levels need to be depleted (creating anoxia) at the sediment–water interface before arsenic is released to the water column.

To put this in perspective, the relatively uncontaminated site C6 has released arsenic into the water column, reaching concentrations of 5–10 μg/L, under transient anoxia with an estimated sediment arsenic concentration of 12–15 mg/kg (Toevs et al., 2006). What would a similar period of anoxia produce at other CDA Lake sites with much higher sediment concentrations of arsenic? For example, arsenic concentrations of 140–330 mg/kg were observed in surface sediments at two sites in 2002 (Toevs et al., 2006). Mercury methylation can also be concerning under anoxia, but is not considered further here given the low abundance of mercury within the sediments.

There are two primary ways to enhance anoxia of bottom waters and underlying sediments: lake eutrophication and lake stratification. Eutrophication (or increased biological productivity) within the lake water results in an enlarged supply of decaying algal material to the lake bottom that increases the microbial oxygen demand and can lead to anoxia. The committee’s analysis of anoxia trends in Chapter 5 found that anoxia is not happening more frequently or for longer periods than in the past, but if lake water quality conditions changed, this trend could be reversed. These changes include (1) alterations in nutrient supply to the water column (Kuwabara et al., 2007; Moberly et al., 2010), (2) increases in water temperature from climate change (Rigosi et al., 2014, Carey et al., 2012; O’Neil et al., 2012; Paerl and Scott, 2010; Dziallas and Grossart, 2011; Moss et al., 2011), and (3) future nutrient release from the lake sediments into the water column. The phosphorus input analysis in Chapter 3 indicates that increases in external nutrient supply to the water column are unlikely because phosphorus loading from the watershed is trending down. Furthermore, phosphorus is likely transporting as an adsorbed phase on Fe(III) oxides and thus has limited bioavailability. An increase in lake water temperature that might lead to oxygen depletion is considered in Chapter 10. Here, the focus is on phosphorus release from the lake sediments (#3), which is of particular concern because primary productivity within CDA Lake is mainly phosphorus-limited (Woods and Beckwith, 1997; but see Chapter 5).

It is possible that phosphorus release from sediments during low oxygen conditions could create a positive feedback loop that leads to greater productivity and subsequently lowers dissolved oxygen even more. However, there is currently no evidence of this occurring in CDA Lake. It is possible that the Lake is nitrogen-limited during the summer, which would prevent the creation of a positive feedback loop between eutrophication and phosphorus release from the sediments (see Chapter 5). Also, iron (and manganese) oxides within the surface sediment strongly bind and limit phosphorus exports into the water column. Given the conditions of the CDA sediments, the committee’s model projections [based on surface complexation modeling using the diffuse layer model developed by Dzombak and Morel (1990) and using phosphorus and ferrihydrite concentrations for the sediment from Toevs et al. (2006)] are that porewater phosphate concentrations would vary only from 0.8 to 0.19 μM, which is lower than the reported mean concentration (13.1 μM). As reported by Morra et al. (2015), total phosphorus in the top 5 cm of sediment ranges from 0.94 to 1.44 g kg^-1, thus residing within the range of 0.5–1.6 g kg^-1 measured in 1992 (Woods and Beckwith, 1997), which is low compared to those reported for sediments of other oligotrophic lakes (Carey and Rydin, 2011). Nonetheless, the potential for phosphorus release from sediments during anoxia cannot be ruled out. The CDA Tribe has collected data showing elevated phosphorus and nitrogen concentrations in the hypolimnion at site SJ1 during anoxic conditions (Chess, 2021).

Lowering pH in Lake Sediments

Another master variable controlling metal partitioning to the sediment solids is pH. However, as illustrated by the surface complexation modeling presented in Figure 7-4, for the pH range of the CDA Lake sediments (pH 6.1–8), while all the metals increase in concentration, the magnitude of change for zinc dwarfs that of the other metals (by more than ten times).

Although lake sediments are typically well buffered due to the presence of oxide and carbonate minerals (Stumm and Morgan, 1981), and their pH is somewhat decoupled from fluctuations within the water column, the sediment–water interface would be influenced by changes in lake water pH. The bottom-water pH data at station C1 from 2005 through 2020 illustrate the wide range of values, spanning from 5.8 to 8.3 (Figure 7-12). Examining the dissolved zinc concentration in relation to pH (at C1 from 2005 to 2020) shows a rather poor correlation (Figure 7-12A). However, if the graph is restricted to the periods when stratification was likely (May through September) and it excludes early summer data from 2011 and 2017 when inflows included low pH waters, then a stronger correlation emerges that is statistically significant (p < 0.01) (Figure 7-12B). When flows subside in the

spring and early summer, the impact of dissolved zinc inputs from the watershed on dissolved zinc concentrations in the Lake wanes; during these more quiescent, stratified periods, decreasing pH may lead to zinc desorption from sediments at the sediment–water interface, leading to increases in dissolved zinc concentrations in bottom waters.

At depths below the sediment–water interface (likely a few centimeters), the largest influence on pH results from changes in oxygenation of sediments. Under oxygen-limited conditions, anaerobic processes generally move the pH toward neutrality, with reductive dissolution of iron and manganese oxides pushing pH upward, while being offset by metal carbonate precipitation, increases in carbonic and organic acids from microbial decomposition of organic matter, and sulfate reduction (McBride, 1994). An increase in oxygenation, by contrast, can lead to decreases in pH; the principal reactions that acidify lake sediments are oxidation of Fe(II) and reduced sulfur. Seasonal changes in pH resulting from shifts in oxygenation would thus be expected to impact porewater zinc concentrations and efflux to overlying waters, but due to complex and competing processes, it is difficult to predict the net effect. Lake conditions that lead to greater fluctuation in oxygen levels will lead to larger changes in pH, including periods of lower pH and release of zinc.

Despite the complexity of oxygen dynamics and pH within the Lake sediments, major patterns in these two drivers and their effects on dissolved metal and arsenic concentrations are observed. As illustrated (and described) in Figure 7-13, Lake stratification in late spring and summer limits oxygen supply from the atmosphere (and upper waters) into the bottom waters. Microbial respiration of dissolved and suspended organic matter leads to a progressive decrease in pH. Although the sediments are well buffered, acidification of the sediment–water interface leads to release of zinc (and in some cases cadmium) into the water column (Figure 7-13A), as shown in Chapter 6. If microbial respiration is sufficient to completely consume oxygen in the water column (Figure 7-13B), anoxia at the sediment–water interface results in microbial Fe(III) and As(III) reduction, which has two consequences: (1) As(III) is released into the water column and (2) the pH increases abruptly, limiting (acid-promoted) desorption of zinc. Figure 7-14 is a finer scale schematic of the sediment–water interface showing both drivers of metals release.

CONCLUSIONS AND RECOMMENDATIONS

Unlike the previous three chapters, which analyzed trends in long-term monitoring data in the watershed and the Lake, this chapter relied on data collected during a handful of special studies of Lake sediments coupled with reactive transport modeling (conducted by the committee) and diagenetic modeling (conducted by S‚ engör et al., 2007), along with known biogeochemistry of elements, to evaluate the current status of zinc, lead, cadmium, and arsenic cycling. Bottom-water data revealed that certain metals are periodically released from Lake sediments during periods of thermal stratification at some Lake sites. Until cleaner materials are deposited in the Lake system, the potential for release of metals from sediments into the water column will persist. The exceptions are redox active elements or elements affected by redox processes, which will continue to migrate toward the sediment–water interface under anoxic sediment conditions, regardless of decreases in the metals concentration of overlying material.

1. Iron(III) (hydr)oxides are an abundant and dominant control on metal and arsenic concentrations in the Lake sediments (at the sediment–water interface). Data from multiple studies illustrate the preponderance of iron within the sediments, with Toevs et al. (2006) showing that it exists in the near-surface sediments as the poorly crystalline Fe(III) hydroxide ferrihydrite. With the pH conditions of the sediments, the Fe(III) oxides serve as principal adsorbents of arsenic, cadmium, lead, and zinc that regulate dissolved concentrations as shown through measurement and modeling.
2. The mass of sulfur relative to the composite of zinc, lead, cadmium, and iron would not allow metal sulfides to be dominant phases within the CDA Lake sediments. Due to the abnormally high metal concentrations (particularly iron, which makes up more than 10 percent of the solids) within the sediments, sulfur occurs in quantities insufficient to control the full suite of heavy metal concentrations and may exert local and selective controls, but not universal control, on metal retention in the sediments. Iron sulfide in the form of pyrite (FeS₂) accounts for the majority of the sulfur, leaving the possibility of small quantities of ZnS, PbS, and CdS that may have local impacts on metal contaminant concentrations but not sweeping controls.

3. The greatest threat of enhanced anoxia, should it occur, is release of arsenic into the Lake water column. However, there is no evidence that anoxia is getting worse (see Chapter 5). Because As(V) adsorption onto Fe(III) oxides is the dominant control on arsenic concentrations within the porewaters of lake sediments, decreased oxygen concentrations leading to anoxia in the bottom waters will promote the reductive dissolution of arsenic within the upper sediments and arsenic release into overlying lake waters. Presently, albeit with a few exceptions temporally and spatially, oxygenation of the upper sediments appears to provide a protective cap, where As(V) retention on Fe(III) oxides limits the dissolved concentrations of arsenic within the porewater and efflux from the sediments into the overlying lake waters.
4. Because phosphorus chemistry is similar to As(V), a second threat of bottom water anoxia is release of phosphorus. Although not redox-active itself, phosphate is largely bound to Fe(III) oxides; anoxia leads to the reductive dissolution of the Fe(III) phases and thus the concomitant release of phosphate. If the Lake is phosphorus-limited, it is possible that such release of phosphorus from the Fe(III) oxides could create a positive feedback loop that further promotes biological productivity, anoxia, and arsenic release from the sediment.
5. Because the adsorption edge for zinc on iron oxides occurs in the pH range of 6.0–7.0 (typical of CDA Lake), local lowering of pH can cause release of dissolved zinc from the sediment. A decline in pH to less than 7 occurs in bottom waters of CDA Lake in some locations and some years, with the onset of stratification. Of the metal(loid) contaminants, zinc has the highest dissolved concentration within the sediment porewater, while cadmium is present at much lower concentrations than zinc. Upward diffusion of zinc from porewaters and desorption from surficial layers of the sediment is possible at pH less than 7. Less of a concern for zinc release would be an alteration in oxygen concentrations within the bottom waters and top few centimeters of

5. the sediment. Although this could lead to a transient increase in dissolved zinc concentrations, zinc largely shifts fluidly between solids, partitioning to Fe(III) oxides under periods of oxygenation or to sulfide phases (precipitates such as ZnS or adsorbed on FeSx) and organic matter under periods of anoxia. The lack of upward zinc migration noted within the sediment profiles illustrates zinc partitioning to the solids under both oxic and anoxic conditions.
6. Lead partitions strongly to the Lake sediment solids both under oxygenated and anoxic conditions. Similar to zinc, it partitions strongly to iron oxides under oxygenation and to sulfide phases (PbS precipitates and is adsorbed on FeSx solids), along with organic matter, under anoxia. Lead adsorption on iron oxides is stronger than zinc adsorption, maintaining a lower dissolved lead concentration in porewaters, and lead desorption is unlikely to occur within the pH range of the CDA Lake system.
7. There are limited datasets of sediment cores from within CDA Lake, making predictions about the future of the Lake water–sediment interactions difficult. The addition of periodic core sampling coupled to existing sampling efforts would provide needed information on in-lake processes, such as redox changes in the sediment leading to metals migration and trends in metal deposition over time. Core sampling should include metal(loid) and nutrient (phosphorus) profiling, metal(loid) and nutrient speciation or phase association, depositional dating, and grain-size fractionation with metals content. Carefully designed collection of porewater data could also improve understanding of sediment–water exchanges and other in-lake processes. An initial core sampling campaign could also provide data needed to develop and calibrate a coupled hydrodynamic–biogeochemical model with specific information tailored to spatial regions within CDA Lake.

REFERENCES