Aqueous Geochemistry and Limnology of the Sleeper Pit Lake, Nevada, USA: Evidence for Long-Term Subaqueous Solute Generation in Mine Pit Lakes

Predictive geochemical and limnologic modeling of pit lakes is an important aspect of modern mine-site permitting. One of the key assumptions of many predictive pit-lake models is that open-pit high walls and in-pit backfill become geochemically unreactive once submerged by the filling pit lake. Existing pit lakes provide useful data to test this assumption. The Sleeper pit lake (northwestern Nevada, USA) is approaching hydrologic equilibrium and contains good-quality water that generally meets regulatory requirements for pit lakes. Despite the overall stable geochemical composition, seasonal trends in the hypolimnion indicate the generation of dissolved metals associated with the ore deposit (e.g., Mn and Zn) and cyclical variations in pH. This study applies mass balance, analysis of subaqueous pyrite oxidation, and trends in solute concentrations to evaluate the potential causes of long-term solute generation in the hypolimnion of the Sleeper pit lake. Three separate conceptual models (subaqueous pyrite oxidation; redox reactions; and diffusion of solutes from a permanently stratified bottom layer) were tested against the high-quality dataset available for the pit lake. Evaluation of the monitoring dataset for the pit lake indicates that the Sleeper pit lake has variable limnologic behavior, wherein the lake is consistently stratified (meromictic) in some years, while it undergoes full mixing (holomixis) in other time periods. Comparison of the data with the three conceptual models illustrates that none of these models can be completely implicated as causing the seasonal geochemical variations in the bottom of the pit lake, but that a combination of the processes is likely partially responsible. Additional data collecHow to cite this paper: Newman, C. P., Cluff, T., Gray, T., & Beale, G. (2019). Aqueous Geochemistry and Limnology of the Sleeper Pit Lake, Nevada, USA: Evidence for Long-Term Subaqueous Solute Generation in Mine Pit Lakes. Journal of Geoscience and Environment Protection, 7, 64-81. https://doi.org/10.4236/gep.2019.73004 Received: December 17, 2018 Accepted: March 11, 2019 Published: March 14, 2019 Copyright © 2019 by author(s) and Scientific Research Publishing Inc. This work is licensed under the Creative Commons Attribution International License (CC BY 4.0). http://creativecommons.org/licenses/by/4.0/


Introduction
Modern mining operations are becoming more reliant on open-pit mining methods to economically produce ore deposits of lower grade. Depending on the relationship of these open pits to the local groundwater table, active dewatering operations are in some instances required to allow mining to commence. When dewatering is discontinued at the conclusion of mining, groundwater and surface water fill the open pit, creating a mine pit lake (Castendyk & Eary, 2009). Pit lakes are becoming a prominent issue around the world (Kumar et al., 2009;Ramteke et al., 2016;Sánchez-España et al., 2008), and have resulted in substantial environmental contamination and been the subject of advanced remediation measures (Tucci & Gammons, 2015).
As mine pit lakes become more widespread, regulatory agencies are increasingly requiring advanced predictive modeling of pit-lake geochemistry during mine permitting. The end goal of all those involved in these permitting actions (i.e., the mining company, the regulatory agencies, and the public) is to have pit lakes become geochemically and hydrologically stable for the long term, which may then allow mining permits to be relinquished. Modeling endeavors therefore are important to inform long-term closure of pit lakes. The conceptual models and numerical methods used to predict pit-lake water quality are complex and require substantial amounts of data and a variety of assumptions (Castendyk et al., 2015). One of the common modeling assumptions is that solute generation via acid mine drainage is eliminated (or at least decreased) by the filling of the pit lake and inundation of high walls and any in-pit waste rock.
However, few studies have reported on observations that support or refute this assumption. The water-quality records of existing pit lakes provide ideal datasets to test these assumptions.
Pit lakes commonly are somewhat unique limnologically, due to the different surface area to depth ratio (termed the relative depth) when compared to natural lakes (Castendyk, 2009). Hydrodynamic and geochemical conditions in pit lakes may lead to pronounced vertical stratification, wherein several geochemically distinct layers form that may not mix with one another. These layers may include the epilimnion (upper-most layer), metalimnion (transition layer), hypolimnion (lower-most layer that participates in mixing), and monimolimnion (semi-permanently stratified bottom layer). If the separate layers of a pit lake C. P. Newman et al. experience full mixing at least once annually that pit lake is termed holomictic, whereas a pit lake with a monimolimnion (which rarely or never fully mixes with the rest of the pit lake) is termed meromictic. There are various ways in which the limnologic behavior of pit lakes can be predictively evaluated (Jewell, 2009), although these are not always effective in predicting the limnology observed in pit lakes (Schultze & Boehrer, 2009). The difference between meromictic and holomictic pit lakes may be important from a regulatory perspective because different modeling and management assumptions are applicable to the different limnologic behaviors (Schultze et al., 2016).

Study Site
The former Sleeper mine is located in northwestern Nevada (Figure 1), and has Figure 1. Regional location of the sleeper pit lake and on-site groundwater monitoring wells. Journal of Geoscience and Environment Protection been inactive (in terms of ore extraction) since 1996. The mine extracted Au and Ag from the famous high-grade Sleeper vein and associated widespread stockwork mineralization (Nash et al., 1995). Water quality in the pit lake has been monitored since the cessation of dewatering in 1996. The current pit lake actually occupies three separate sub-pits, the Sleeper, Wood, and Office pits ( Figure 1). The Sleeper and Wood pits undergo regular water-column monitoring using a conductivity-temperature-depth (CTD) probe, and sampling for dissolved inorganic constituents at several depths (3, 40, and 100 m [~10, ~130, and ~330 ft. respectively] below lake surface). Groundwater quality in the vicinity of the pit lake is also monitored using a series of wells ( Figure 1).
During the initial closure of the Sleeper mine a novel technique at the time was utilized wherein nearby groundwater was pumped into the open pit to minimize the period of exposure of acid generating high walls near the bottom of the pit and waste rock that was backfilled into the pit (Dowling et al., 2004). This rapid-fill approach resulted in the pit lake nearing its final predicted stage elevation much more quickly than would have occurred under normal hydrogeologic conditions. The rapid-fill also included episodic lime dosing. Prior to the rapid fill, the pits were partially backfilled to cover unoxidized portions of the pit floors and walls with both oxidized and unoxidized waste rock, which was amended with lime in some instances. During initial infilling, the pit lake contained moderately acidic water with elevated concentrations of dissolved metals and metalloids such as As, Cd, Cu, and Mn. However, supplemental lime addition resulted in the pit lake rapidly (within approximately three years) attaining a circumneutral pH with a dramatic decrease in concentrations of the aforementioned constituents. The implemented closure strategy is likely responsible for the rapid and long-term stabilization of the pit-lake water quality.
Although the current pit lake overall has a relatively stable and good-quality water chemistry, seasonal changes are evident in the hypolimnion (assumed to be represented by water quality in the sample from 100 m [330 ft.] depth). These changes constitute fluctuations of pH, alkalinity and the dissolved concentrations of metals such as Zn and Mn. At present, these trends do not cause major changes in the closure plan or regulatory compliance, but they may be indicative of important geochemical processes occurring in the lake with possible future implications at the Sleeper mine and other pit lakes. The purpose of this paper is to explore three potential conceptual models that may cause these seasonal changes: A) Pyrite oxidation may be ongoing in the submerged backfill leading to fluxes of associated constituents (Zn, Fe, etc.) in the hypolimnion. These constituents may then be diluted by the less concentrated shallower waters of the epilimnion during turnover events; B) Redox processes may cause benthic fluxes to contribute dissolved metals seasonally to the hypolimnion waters; or, C) The pit lake may have a permanently stratified bottom layer (monimolim-C. P. Newman et al. Journal of Geoscience and Environment Protection nion), with elevated metal concentrations, that has never been sampled. During the stratified time period diffusion of dissolved solutes in the monimolimnion to the hypolimnion causes observed water quality changes, while these changes are masked by dilution during mixing events between the epilimnion and hypolimnion.
Potential methods by which these end-member scenarios (or combinations thereof) may be assessed are detailed in the following section.

Methods and Data
Water-quality samples have been collected from various depths within the Sleeper pit lake for approximately the past 22 years. As such a very complete dataset exists consisting of quarterly compliance samples and CTD probe measurements As such it appears that the majority of metals in the pit lake are present in dissolved form. This study utilized all water-quality monitoring data and CTD data to evaluate the aforementioned conceptual models.
As described by Gammons (2009), oxidation of pyrite (FeS 2 ; the most common sulfide mineral in many ore deposits) can proceed with either oxygen (O 2 ) or ferric Fe (Fe III ) as the oxidant. The rate of oxidation by these two species can be calculated according to Equation (1) (for O 2 ) and Equation (2) where r represents the oxidation rate (mol/m 2 ·sec), square brackets represent the molal concentration of the given species, and Eh represents the oxidation-reduction potential (V) corrected to the standard hydrogen electrode. Although both pH and Eh are recorded by the CTD, the availability of CTD data is sporadic and Eh measurements are not available for the entire filling history.
Therefore, recent Eh measurements (from August 2018) were utilized in all calculations for the rate of oxidation by Fe III . This has the net effect of increasing uncertainty as to how this process impacts water quality. Also, Equation (2) is only applicable if the water is in equilibrium with respect to ferrihydrite (Gam-  mons, 2009). This requirement is met in the majority, but not all, of the sampling events utilized in these calculations. Of the 63 representative hypolimnion concentrations utilized in this analysis 58 displayed ferrihydrite saturation index (SI) values greater than 0 (indicating propensity for precipitation). In support of these calculations, the total acidity of the pit lake was also calculated using the methodology of Kirby and Cravotta (2005) shown in Equation (3), which allows assessment of the total acidity load in the pit lake to be quantified in addition to the observed pH changes (which only quantifies hydrogen acidity).
where A denotes acidity (mg/L as CaCO 3 ), pH is in standard units, and [Mn], [Fe], and [Al] respectively denote the concentration of Mn, Fe, and Al (all mg/L). Kirby and Cravotta (2005) differentiate between the acidity of ferrous Fe (Fe II ) and ferric Fe (Fe III ), which may be important in some systems. No Fe speciation analytical results are available for the Sleeper pit lake however. As such all Fe was considered as ferric in the calculations, which would tend to overestimate metal acidity from Fe. As detailed below in the results section, Fe is not the most abundant metal throughout the majority of the monitoring record. Therefore the effect of assuming ferric versus ferrous Fe is likely not consequential to the results.
To evaluate conceptual model B (that redox reactions and benthic fluxes impact the lake chemistry), the depth profiles of various constituents were compared with one another over time, and simple redox reactions were considered.
It is important to note however that benthic fluxes cannot be quantitatively evaluated without the use of in-situ measurements (e.g., using diffusion-controlled dialysis samplers, a.k.a. peepers; Martin & Pedersen, 2002). The analysis presented herein is therefore conceptual only. Eary and Castendyk (2013) summarize a variety of potential sediment biogeochemical processes that may occur in pit lakes, specifically with respect to the oxidation of organic carbon and pyrite, and the reduction of metal oxyhydroxides according to reactions shown in Equations (4)-(8), where C org represents organic carbon (simplified for this analysis), and solids are denoted using the subscript (s). Overall, these series of reactions can be combined (including multiplication to balance species) into the net Reaction (9).
This reaction conceptually represents the oxidation of pyrite with dissolved oxygen to form an Fe-oxyhydroxide (ferrihydrite is commonly assumed in these conditions), combined with the reduction of Mn-oxide (which could be pyrolusite or another phase) with organic carbon as an electron donor. This reaction is conceptual only, as no observations of such phase transformations are available for the Sleeper pit lake. Several aspects of the net reaction are noteworthy. First, the reaction as written requires four moles of pyrite for every mole of organic matter and Mn-oxide. Second, the reaction produces substantial amounts of sulfate (8 moles) and acidity (16 moles of hydrogen ions). Also, it could be relevant to discuss the reduction of Fe-oxyhydroxides (such as ferrihydrite), but reduction of these minerals occurs at lesser Eh values (more reducing conditions) than for Mn-oxyhydroxides (Appelo & Posta, 2005). Therefore reductive dissolution of Fe-oxyhydroxides was not considered in this analysis, although future evaluation of this process may be appropriate.
Although these phase transformations have not been investigated in the Sleeper pit lake using sediment sampling or analysis of pore waters, the plausibility of the overall reaction can be evaluated by quantifying the changes in Mn concentration in relation to sulfate and pH, precisely because of the stoichiometry discussed above. Additionally, this reaction is consistent with the oxidation of pyrite while maintaining low dissolved Fe concentrations, as is observed in the pit lake and discussed below in the results section.
Finally, to evaluate conceptual model C (diffusion of water from a thin monimolimnion of unknown composition or from sediment pore-waters), an analytical form of Fick's law of diffusion may be used according to: where q z represents diffusive flux in the z dimension (mol/m 2 ·sec), D represents the diffusion coefficient for the given species (m 2 /sec), and dC/dz represents the concentration gradient in the z dimension (mol/m). To calculate the total mass of diffusive transport, the area over which diffusion acts must be known. For this analysis this area was calculated as the approximate area of the bottom of the pit lake, where diffusion from the saturated sediments or the monimolimnion would conceptually take place. Diffusion coefficients (D) vary between sources.
For this analysis the online reference Aqion (2018) was utilized.

Seasonal Geochemical Fluctuations and Limnologic Behavior
The Sleeper pit lake is circumneutral. Seasonal fluxes of various constituents are observed by plotting the water quality of the hypolimnion and epilimnion waters through time (Figure 2). These seasonal changes include decreasing alkalinity  Seasonal changes in water quality may also be evaluated with respect to the total acidity of the lake (i.e., metal acidity plus hydrogen acidity). Figure 3 illustrates the changes in acidity in both the epilimnion and hypolimnion over time.    (Figure 3(c)). This acidity was rapidly neutralized and the epilimnion had little initial acidity. The elevated dissolved metal concentrations in the hypolimnion were likely due to flushing of the in-pit waste rock and enhanced oxidation during initial filling. Illustrated in Figure 3(b) are seasonal fluctuations in calculated acidity in more recent years. These reflect the changing pH of the water column during the summer (Figure 2(b)) as well as seasonal loading of Mn.
Seasonal patterns are clearer in the hypolimnion, although some long-term and seasonal changes are also observed for the epilimnion. It should be noted that the seasonal changes observed in Figure 3 Although it is commonly assumed in pit-lake predictive studies that all pit lakes in Nevada are holomictic (i.e., they experience full turnover of the water column at least once per year), and the Sleeper pit lake has been observed to turnover annually in the past (Dowling et al., 2004), current observations illustrate that the pit lake is meromictic in some years (i.e., it remains stratified for periods of greater than one year). If a given Nevada pit lake does experience full mixing, this process commonly occurs during either the fall and/or spring seasons when wind stress is at a maximum and thermal inversion serves to destabilize the existing stratification (Balistrieri et al., 2006).
The limnologic behavior of the Sleeper pit lake over various timescales is illustrated in Figure 4. While Figure 2 indicates clearly that the pit lake has been largely holomictic in recent years, Figure 4   while the lower portion of the lake may receive relatively little groundwater inflow due to lower hydraulic conductivity volcanic rocks (Dowling et al., 2004).
The influent groundwater from the basal alluvial gravel aquifer is generally high in alkalinity, with pH, total alkalinity concentrations, and total dissolved solids Although only a snapshot in time, this analysis suggests that the Sleeper pit lake may have variable limnologic behavior, with some periods characterized by holomixis ( Figure 2) and others characterized by meromixis (Figure 4). It is also interesting to note that even though both major pits of the original Sleeper mine are occupied by one contiguous pit lake, the limnologic behavior of these pits is not always linked. Figure

Solute Generation in the Hypolimnion
The rate of subaqueous pyrite oxidation (by both O 2 and Fe III ) was calculated according to Equations (1) and (2)  One possible series of reactions responsible for the generation of Mn, without appreciable Fe, is described in Equations (4)-(9). One way to evaluate the applicability of this series of reactions is by quantifying changes in the molar ratios of all constituents over time. Most notably the net reaction produces 8 moles of sulfate and 16 moles of H + for each mole of Mn. However, the calculated acidity changes over time ( Figure 3) and variability in pH (Figure 2(b)) are much less than this reaction would predict. Also, quantification of the molar ratios over time of these constituents show that these ratios are not maintained in the actual pit lake. Taken together these results show that the exact balanced reaction in Equation (9) is not occurring in the saturated backfill in the Sleeper pit lake, or other additional processes are also involved. The ratios of constituents to one another would be different if a different series of reactions were utilized, and it is important to note that Mn release through reductive dissolution is not dependent on pyrite oxidation (Equation (4)), which may explain the observations.
Further evaluation of the detailed reactions and processes involved may require additional data collection, possibly including evaluation of sediment cores and pore water analyses.

Monimolimnion Formation and Chemical Diffusion
The observed CTD profiles of the Sleeper pit lake indicate that a monimolimnion may form in some years, although the stratification is sometimes destabilized promoting full mixing of the pit lake. It is important to note that the monimolimnion may not have been subject to water-quality monitoring in the past.
The deepest water-quality sample collected from the pit lake to date is approx-  (10)) was used to evaluate this potential for diffusion from the monimolimnion or sediment pore waters.  Figure 4(a). The calculations are defined such that the z dimension is positive with increasing depth in the pit lake. Therefore a negative flux indicates upward diffusion in the water column, and vice versa.
The results of the calculations ( Figure 6) illustrate an upward diffusive flux of H + (essentially the upward migration of lower pH conditions) from the sediment-water interface in September of 2005, when the lake is believed to have been meromictic. This upward flux of H + continues nearly to the surface of the lake. It is worthwhile to note that this calculation mirrors the observed gradient in pH, and therefore the calculation is not independent of those observations.
The calculation merely quantifies the mass-transport of the variable pH zones.
Because the calculations are driven by observations, diffusion calculated using monitoring data from other dates would arrive at different results. Also, the surface area of the bottom of the lake (estimated as ~117,000 m 2 ) and the diffusion coefficient are sources of potential error in this calculation. While the diffusion results illustrated in Figure 6 may not be indicative of long-term behavior, the main result of the calculations remains valid. This conclusion is that a diffusive flux of H + was emanating from the lower lake or the sediment-water interface on the date shown in Figure 6.
Although these results indicate that diffusion of some chemical species could be controlling dissolved concentrations in the lake, it is interesting to note that the process of double diffusion is not observed in the Sleeper pit lake. Double diffusion is a physical process involving variable diffusive fluxes of heat and dissolved solutes that has been attributed with various water quality changes observed in some meromictic pit lakes von Rohden et al., 2010). This process is generally indicated by a stair-stepped appearance in the conductivity and temperature profiles of the lake, which is not typically observed

Conclusion
The Sleeper pit lake is somewhat unique among Nevada pit lakes in that it is apparently meromictic in some time periods (fall 2005 to fall 2006 for example), while it undergoes full mixing (holomixis) in other periods. The cause for the variable behavior is not known but warrants further investigation. This analysis has clearly demonstrated subaqueous geochemical changes in the Sleeper pit lake, despite not providing a complete understanding of the mechanisms leading to these observed changes. Several processes were evaluated including subaqueous pyrite oxidation, redox reactions, and diffusion of H + ions from the monimolimnion. Although none of these phenomena can be implicated with certainty, the observed water-quality trends are likely partially the result of a combination of these processes. Several important lines of additional characterization could be completed to provide insight on these processes including sampling of backfill pore waters and the sediment column, and sampling water quality near the bottom of the lake to evaluate monimolimnion chemistry (when applicable). Regardless of the lack of concrete evidence as to mechanistic processes, these findings have important implications for ultimate closure planning in pit lakes: 1) If necessary, sulfide material may be backfilled into a pit that will form a lake, but precautions need to be taken to minimize the oxidation that can occur prior to hydrologic stabilization of the lake. Also, this closure method is likely only applicable under certain geochemical and physical conditions (i.e., in deep stratified pit lakes with substantial influent alkalinity in groundwater and low Fe concentrations to minimize subaqueous pyrite oxidation by Fe III ).
2) Assumptions regarding annual turnover (i.e., that pit lakes will be holomictic in all years) may not be well-founded. The modeling and management strategies for meromictic lakes may be different than for holomictic lakes (e.g., Schultz et al., 2016). The possibility of long-term stratification, and the impact of this process on mine closure, should also be assessed.
3) Subaqueous generation of solutes is evident in the Sleeper pit lake (e.g., seasonal changes in Figure 2), although this does not change the current management strategy for the lake due to site-specific factors including high-alkalinity groundwater inflow and the large lake volume. However, in smaller lakes with less volume for dilution or without good-quality water sources, predictive modeling should evaluate the impact that long-term generation of solutes would have on water quality. If these processes are ignored in modeling and permitting studies, proponents risk underestimating the concentrations of important solutes (e.g., sulfate and metals) in future lakes, which may increase the time and cost of