Mechanisms of water-rock interaction and implications for remediating flooded mine workings elucidated from environmental tracers, stable isotopes, and rare earth elements

Abstract

Contamination from acid mine drainage affects ecosystems and usability of groundwater for domestic and municipal purposes. The Captain Jack Superfund Site outside of Ward, Boulder County, Colorado, USA, hosts a draining mine adit that was remediated through emplacement of a hydraulic bulkhead to preclude acid mine drainage from entering nearby Lefthand Creek. During impoundment of water within the mine workings in 2020, a diverse and novel dataset of stable isotopes of water, sulfate, and carbon (${\text{δ}}^2\text{H}$, ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$, ${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$, ${\text{δ}}^{34}\text{S}$, ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$), rare earth elements, and environmental tracers (noble gases and tritium) were collected to understand groundwater recharge and mixing, mechanisms of sulfide oxidation and water-rock interaction, and the influence of remediation on the hydrologic and geochemical system. Water isotopes indicate that groundwater distal from the mine workings has seasonally variable recharge sources whereas water within the workings has a distinctive composition with minimal temporal variability. Sulfate isotopes indicate that sulfide oxidation occurs both within the mine workings and in adjacent igneous dikes, and that sulfide oxidation may occur under suboxic conditions with ferric iron as the oxidant. Carbon isotopes track the neutralization of acidic waters and the carbon mass budget of the system. Rare earth elements corroborate stable isotopes in indicating groundwater compartmentalization, and additionally illustrate enhanced mineral weathering in the mine workings. Environmental tracers indicate mixing of modern and pre-modern groundwater and inform timelines that active remediation may be needed. Together these datasets provide a useful template for similar investigations of abandoned mine sites where physical mixing processes, sources of solute loading, or remediation timeframes are of importance.

1. Introduction

Acid mine drainage (AMD) results from the exposure and oxidation of sulfide minerals during and after mining and may cause contamination of surface water and groundwater (Nordstrom, 2011). Although acidity alone may cause ecological damage, much of the concern related to AMD is attributed to increased solubility of base metals under acidic conditions (Gammons et al., 2015). Geochemical effects of AMD may last for thousands of years (Pérez-López et al., 2010) and may require complex mitigation measures to reduce contaminant transport in surface-water and groundwater systems (Nordstrom, 2011).

Contamination from AMD is common throughout the state of Colorado and has caused negative effects to ecosystems and human populations (Cowie et al., 2014; Rodriguez-Freire et al., 2016). Numerous abandoned mines have contaminated surface water and groundwater in the Colorado Mineral Belt, stretching from the San Juan Mountains in the southwestern part of the state to Boulder County in the northeast (Colorado Department of Public Health and Environment, 2017). Contamination from natural weathering of sulfide minerals un-associated with mining in the Colorado Mineral Belt has also affected waterways and aquifers (Verplanck et al., 2009). Remediation of abandoned mine sites in Colorado may be beneficial to restore ecological functioning of streams and to provide groundwater and surface waters that are useable for agricultural and human/animal consumption.

There are a number of remediation strategies which may be applied to sites with AMD including interaction with carbonate material (Cravotta and Trahan, 1999), enhanced biogeochemical fixation of metals (Foote et al., 2007), or active pump-and-treat (Elliot and Younger, 2014). An additional strategy to control sources of metals from underground mine settings to surface waters is the use of structural or hydraulic bulkheads to impound water within mine workings (Walton-Day et al., 2021; Wolkersdorfer, 2008). Bulkheads generally decrease the discharge from draining mine adits and may decrease point-source loading of metals to streams (Walton-Day and Mills, 2015), though the watershed-scale effects on metal loading remain difficult to constrain (Petach et al., 2021; Walton-Day et al., 2021). Additionally, bulkheads may limit sulfide oxidation by limiting the ingress of atmospheric oxygen (Walton-Day et al., 2021; Wolkersdorfer, 2008), though sulfide oxidation can occur via multiple pathways with both atmospheric oxygen or ferric iron (${\text{Fe}}^{3+}$) as the oxidant (Hubbard et al., 2009).

Geochemical processes governing sulfide oxidation are commonly evaluated using geochemical models (Manning et al., 2013) and stable isotopes (Nordstrom et al., 2007). These methods provide insights on the rates and mechanisms of AMD generation, which may then be used to implement remedial activities to counteract solute mobilization. Subsequent transport of AMD in surface waters and groundwater may be quantified using mass balance approaches, which highlight discrete and diffuse sources of AMD (Glynn and Brown, 2012; Runkel et al., 2013). Using detailed mass-loading information, it is possible to target specific locations for remediation. Despite the wide range of approaches for evaluating effective remedial actions for AMD, remedial activities still only achieve marginal efficacy in many locations (Runkel et al., 2016). Decreased efficacy leads to longer timeframes of required treatment and increased costs. Remediation of draining adits with bulkheads is an example of a scenario potentially requiring long periods of active treatment (Walton-Day et al., 2021).

This research focused on mechanisms of water-rock interaction and hydrologic limitations on remediation strategies within flooded mine workings at the Captain Jack Superfund Site located near Ward, Colorado, USA (Fig. 1). To understand water-quality variability and constraints owing to water-rock interaction and local groundwater hydrology, a diverse dataset (Newman, 2022) was collected that included stable isotopes (${\text{δ}}^2\text{H}$, ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$, ${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$, ${\text{δ}}^{34}\text{S}$, ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$), rare earth elements (REE), and environmental tracers (tritium [${ }^3\text{H}$], noble gases). In contrast to metal concentrations, these constituents may provide mechanistic understanding of the processes leading to AMD. For instance, noble gases and tritium may identify physical subsurface controls and groundwater mixing in mining environments (Elliot and Younger, 2007; Manning et al., 2008; Wellman et al., 2011). The REE provide similar process-based understanding of potential water-rock interaction pathways (Gӧb et al., 2013; Pérez-López et al., 2010). Application to the Captain Jack site serves as an example of the utility of these datasets at other mine sites where long-term remediation is required.

Fig. 1.

Geologic map and monitoring sites near the Captain Jack Superfund Site in north-central Colorado, USA. The Big Five and Dew Drop workings are shown, the full extent of the Niwot workings are not shown. Only sites that were sampled as part of this study are labeled.

1.1. Study site

Mining in the vicinity of the Captain Jack Superfund Site focused on gold and silver, occurred between the 1860s and 1992, and was completed in a series of distinct mines and tunnel systems including the Big Five workings, Dew Drop workings, and Niwot workings. The primary source of AMD to surface waters in the vicinity of the mine site is the Big Five adit, which has historical discharge rates of 1.3–10 L per second (L/s; U.S. Environmental Protection Agency, 2017).

The study area is comprised primarily of crystalline igneous and metamorphic rocks ranging in age from Precambrian to Tertiary. The oldest rocks are biotite gneiss and schist of Precambrian age (1.7–1.8 billion years [Ga] before present) regionally known as the Idaho Springs Formation (Tweto, 1977) and the Silver Plume Granite of middle Proterozoic age (1.4 Ga; Gable and Madole, 1976). The Idaho Springs Formation and Silver Plume Granite underly most of the study site (Fig. 1). Tertiary igneous dikes are also present including quartz latite porphyry, quartz monzonite, and monzo-granodiorite porphyry (Gable and Madole, 1976). Intrusion of Tertiary igneous dikes into the Idaho Springs Formation and Silver Plume Granite was associated with regional igneous activity during the Laramide orogeny (Lovering and Goddard, 1950). Igneous dikes generally strike in a northwest-southeast orientation, and because mining developed along hydrothermal features, the Big Five mine workings follow a similar trend as the igneous dikes and associated hydrothermal alteration (Fig. 1; Lovering and Goddard, 1950; Wahlstrom, 1935).

Multiple phases of hydrothermal alteration and mineralization in the study area are related to the emplacement of Tertiary intrusive igneous rocks, specifically the White Raven quartz monzonite porphyry (Lovering and Goddard, 1950; Wahlstrom, 1935), named after the nearby White Raven mine. Hydrothermal alteration minerals at the White Raven mine include pyrite, chalcopyrite, sphalerite, galena, siderite, quartz, and calcite (Wahlstrom, 1935). This alteration assemblage is similar to that found in the Big Five, Dew Drop, and Niwot workings, which additionally includes fluorite, molybdenite, wolframite, chalcocite, and barite but does not include siderite or quartz (Lovering and Goddard, 1950).

The climate of the area is semi-arid with annual precipitation ranging from 0.40 to 0.95 m, the majority of which falls during the winter months as snow. Mean annual air temperature is approximately 3.7 °C. These climatic attributes were derived using the ClimateEngine online tool (Huntington et al., 2017). Several streams occur in the vicinity of the site including Lefthand Creek and an unnamed tributary to Lefthand Creek, which flows over the mine workings (Fig. 1).

In November 2017, installation of a hydraulic bulkhead was completed approximately 300 m from the surface within the Big Five workings. Boreholes were also drilled into the mine workings upgradient from the bulkhead to allow for active treatment and recirculation of treated water between the adit and Dew Drop #3 (Fig. 1). Closure of the bulkhead to impound water within the mine workings began in May 2018. Within five months of closure of the bulkhead water levels within the mine workings had increased by approximately 30 m, whereas wells completed only 10 m from the mine had experienced almost no change in water level (Newman, 2023). In a detailed statistical study of correlations between water-level fluctuations in the mine workings and adjacent wells Newman (2023) found that wells completed in Silver Plume Granite generally were disconnected from changes within the workings (with the exception of MW #3), whereas wells completed in Tertiary igneous dikes were more closely connected with the workings. The disconnection between groundwater levels in the adjacent groundwater and mine workings indicates that the system is highly heterogeneous and compartmentalized. Rapidly increasing water-levels within the mine workings caused concern regarding available water treatment capacity and that highly contaminated AMD may begin to discharge from open boreholes drilled into the mine workings. Due to these concerns, the bulkhead was opened in September 2018, and water was allowed to discharge from the workings. During water impoundment and upon opening, water discharging from the mine workings had elevated metal concentrations and decreased pH when compared to historical water-quality data (Colorado Department of Public Health and Environment and U.S. Environmental Protection Agency, 2018). Increased metal concentrations and decreased pH are consistent with dissolution of soluble efflorescent sulfate salts (Elliot and Younger, 2014; Gzyl and Banks, 2007; Newman et al., 2019). These water-quality changes provided the impetus for the current study.

In September 2020, the bulkhead was once again closed to impound water within the mine workings. During the September 2020 closure, an active treatment system was also used to neutralize water within the workings using a combination of lime and carbonaceous material. Active treatment included piping a small volume of water from the bulkhead to a surface treatment facility, amending with a temporally variable mixture of lime (to neutralize acidity) and carbonaceous material (to promote reducing conditions), and then pumping the treated water to an injection well (Dew Drop #3; Fig. 1) upgradient from the bulkhead where the treated water was reintroduced into the mine workings.

Several questions arose during the initial closure of the bulkhead in 2018 that resulted in additional investigation during the 2020 closure. These questions include 1) what are the mechanism(s) of sulfide oxidation in the mine workings and do mechanisms change during flooding, 2) can contaminated water within the mine workings discharge to the adjacent aquifer during filling, and 3) what is the likely amount of time that active treatment and recirculation will be required to achieve stable geochemical conditions within the flooded workings. This study evaluates these questions by focusing on mechanisms of water-rock interaction and changing hydrologic controls during the second period of water impoundment and treatment, beginning in September 2020. Sampling was designed to capture temporal changes in the hydrologic system during water impoundment and treatment, and initial samples were collected in June 2020.

2. Methods

2.1. Water-quality data collection and analysis

Water-quality and solid mineral samples were collected from June through December 2020. Water-quality samples were collected according to standard methods of the U.S. Geological Survey (USGS) as documented in the National Field Manual (U.S. Geological Survey, variously dated). Water-quality samples were collected from the hydraulic bulkhead, groundwater wells, surface seeps/springs, and a single sample of stream water. Groundwater wells sampled are completed both within crystalline bedrock and within open and partially collapsed mine workings. The types of locations sampled are summarized in Table 1.

Table 1.

Naming conventions and site types of groundwater (wells completed in crystalline bedrock), mine water (wells completed in mine workings), spring/seep, and surface water monitoring sites sampled. USGS station IDs are referenced to the USGS National Water Information System (NWIS) database (U.S. Geological Survey, 2022). Well attributes including coordinates, completion lithology, and depth are available from the NWIS database (U.S. Geological Survey, 2022). Only locations that were sampled are included, not all locations illustrated on Fig. 1 were sampled.

USGS Station ID	Short Site Name	Monitoring Site Type	Lithology
400348105304401	Big Five Adit (sampled at hydraulic bulkhead)	Mine Water	Tertiary igneous dike
400351105305401	CDOT ROW #1	Groundwater	Idaho Springs Formation
400351105305402	CDOT ROW #2	Mine Water	Tertiary igneous dike
400354105305901	Dew Drop #1	Groundwater	Tertiary igneous dike
400354105310002	Dew Drop #3	Mine Water	Tertiary igneous dike
400356105310401	Midway #1	Groundwater	Tertiary igneous dike
400349105304401	MW #1	Groundwater	Silver Plume Granite
400347105304801	MW #2	Groundwater	Silver Plume Granite
400347105305401	MW #3	Groundwater	Silver Plume Granite
400346105304801	Seep-06	Spring/seep	Silver Plume Granite
400353105304701	Seep-08	Spring/seep	Till of Bull Lake
400346105304501	Seep-05	Spring/seep	Silver Plume Granite
400353105304702	Seep-07	Spring/seep	Till of Bull Lake
400351105310101	SW-03	Surface Water	Idaho Springs Formation

Water-quality samples were collected for a full suite of dissolved analytes including major cations and anions (alkalinity, Ca, Cl, F, K, Mg, Na, ${\text{NO}}_2-{\text{NO}}_3$, ${\text{SO}}_4$), trace elements (Ag, Al, As, B, Ba, Be, Br, Cd, Co, Cr, Cu, ${\text{Fe}}_{\text{total}}$, ${\text{Fe}}^{\text{II}}$ [ferrous Fe], Li, Mn, Mo, Ni, Pb, Sb, Se, Si, Sr, Th, Ti, Tl, U, V, Zn), rare earth elements (REE; Ce, Dy, Er, Eu, Gd, Ho, La, Lu, Nd, Pr, Sm, Tb, Tm, Yb), stable isotopes (${\text{δ}}^2\text{H}$, ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$, ${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$, ${\text{δ}}^{34}{\text{S}}_{\text{SO}4}$ [referred to hereafter as ${\text{δ}}^{34}\text{S}$], ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$), noble gases (Ar, He, Kr, Ne, Xe), and tritium (${ }^3\text{H}$). Data collected by USGS are available through the NWIS database (U.S. Geological Survey, 2022) by station number and are included in the USGS data release associated with this study (Newman, 2022). Data collection, including quality assurance blank and replicate samples, was conducted according to the USGS National Field Manual (U.S. Geological Survey, variously dated) and as described in detail in Supplementary Material Section S1.

Results presented in this study focus only on REE, stable isotopes, and environmental tracers (${ }^3\text{H}$, noble gases) because these constituents provide detailed understanding of solute sources, hydrologic connectivity and mixing, and efficacy of remediation strategies that may be difficult to constrain using trace elements. Analytical results for major ions and trace elements are not presented in detail in this study but are consistent with AMD throughout Colorado (Colorado Department of Public Health and Environment, 2017), with samples from mine workings having low pH (2–5) and elevated concentrations of metals such as Al, Cu, and Fe. Samples from groundwater wells screened in crystalline bedrock, seeps/springs, and surface water generally have circumneutral pH and more dilute metal concentrations. Brief discussion of water-quality characteristics and temporal changes are included in the results section and in the Supplementary Material Section S3.

Analysis of REE was conducted at the EPA Office of Research and Development laboratory by high resolution inductively coupled plasma-mass spectrometry (HR-ICP-MS) using methods and quality control procedures presented in Wilkin et al. (2021). Stable isotopes of water (${\text{δ}}^2\text{H}$, ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$) were analyzed at the USGS Reston Stable Isotope Laboratory (RSIL) by dual-inlet isotope-ratio mass spectrometry (IRMS) according to methods described in Révész and Coplen (2008a; 2008b) and are reported in units of per mil (‰) in reference to Vienna Standard Mean Ocean Water (VSMOW), with uncertainties of 2 and 0.2‰, respectively. Stable isotopes of sulfate (${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$, ${\text{δ}}^{34}\text{S}$) were analyzed by continuous flow IRMS at the RSIL according to methods described in Révész et al. (2012) and are reported in units of ‰ in reference to VSMOW and Vienna Canyon Diablo Troilite (VCDT), respectively, with uncertainties of 0.4‰. Stable carbon isotopes (${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$) were analyzed at the RSIL according to methods described by Qi et al. (2021) and are reported in ‰ in reference to Vienna Peedee Belemnite (VPDB) with uncertainty of 0.4‰. Noble gases were analyzed at the USGS Noble Gas Laboratory by magnetic-sector mass spectrometry and ultralow vacuum extraction line according to methods described by Hunt (2015) and are reported in units of cubic centimeters at standard temperature and pressure per gram of water (ccSTP/${\text{gH}}_2\text{O}$) and as ratios (e.g., ${ }^{20}\text{Ne}/{ }^{22}\text{Ne}$), respectively, with analytical uncertainties of 1% (He), 2% (Ne), 2% (Ar), 3% (Kr), and 3% (Xe). Tritium analysis was conducted at the USGS Menlo Park Tritium Laboratory using distillation and electrolytic enrichment followed by liquid scintillation and is reported in tritium units (TU) with analytical uncertainty ranging from 0.17 to 0.36 TU.

In addition to water-quality samples several solid samples were collected for isotopic characterization. Stable isotopes of solids at mine sites are useful for understanding controls on water quality because aqueous stable isotopes are commonly controlled by equilibrium with solids (Seal, 2003). Grab samples of secondary minerals were collected for analysis of ${\text{δ}}^{34}\text{S}$ from mine workings adjacent to the hydraulic bulkhead and on the surface of the Dew Drop waste rock dump (near Dew Drop wells; Fig. 1). Suspected efflorescent sulfate salts sampled from the mine workings were shown to be carbonate minerals upon laboratory analysis. The grab sample from the surface of the Dew Drop waste rock dump was confirmed to be an efflorescent sulfate salt. Grab samples of sulfides on the surface of the Dew Drop waste rock dump and from waste rock near the New California Raise (located near the CDOT ROW wells; Fig. 1) were collected for analysis of ${\text{δ}}^{34}\text{S}$. Grab samples of carbonaceous amendment material introduced into the mine during active treatment were collected for analysis of ${\text{δ}}^{13}\text{C}$. Analysis of ${\text{δ}}^{34}\text{S}$ in suspected efflorescent sulfate salts was conducted by the USGS Geology, Geophysics, Geochemistry (G3) Science Center by IRMS. Analysis of ${\text{δ}}^{34}\text{S}$ in sulfides was conducted by the G3 Science Center by IRMS according to methods described in Johnson et al. (2019). Analysis of ${\text{δ}}^{13}\text{C}$ was conducted by the RSIL by IRMS according to methods described in Révész and Qi (2006).

2.2. Environmental-tracer modeling

Environmental tracers including noble gases and ${ }^3\text{H}$ provide insight into physical and geochemical processes occurring in groundwater systems (Aeschbach-Hertig and Solomon, 2013; Cook and Böhlke, 2000). Subsets of these environmental tracers have been applied to mineralized areas (Cook et al., 2017; Manning et al., 2020; Sánchez-España et al., 2014), and several studies have sampled for environmental tracers directly within underground mines or from flowing adits (Elliot and Younger, 2007, 2014; Fey and Wirt, 2007; Manning et al., 2008; Parry et al., 2000; Wellman et al., 2011). Despite these previous applications, a comprehensive analysis utilizing environmental tracers and lumped parameter modeling has yet to be applied to draining mine adits and associated groundwater. In this study, environmental tracers were used to investigate groundwater recharge characteristics (such as temperature and excess air), calculate apparent groundwater ages (Suckow, 2014), and identify mixing of young and old groundwater (Kulongoski et al., 2008; Suckow, 2014).

Noble gas results were analyzed using the software DGMETA (Jurgens et al., 2020) to evaluate recharge conditions. Three models of excess air formation were considered: closed-system equilibration (CE), partial re-equilibration (PR), and unfractionated air (UA), as described by Aeschbach-Hertig and Solomon (2013). Inverse model simulations were evaluated using the chi-squared (${\chi }^2$) parameter, which accounts for model errors in each gas. Minimizing ${\chi }^2$ generally maximizes the calculated model probability. Model probabilities greater than 1 percent are generally acceptable for further interpretation (Aeschbach-Hertig and Solomon, 2013). Model probabilities produced in this study ranged from 25 to 99 percent. Additional details on noble-gas models are provided in the Supplementary Material Section S5.

Concentrations of ${ }^3\text{H}$, helium-4 (${ }^4\text{He}$), and tritiogenic helium-3 (${ }^3{\text{He}}_{\text{trit}}$) that were corrected for calculated recharge conditions were input into the software TracerLPM (Jurgens et al., 2012) to facilitate lumped parameter modeling. TracerLPM may be used to estimate groundwater ages and to fit environmental tracer data to conceptual models of groundwater flow, which allows for mechanistic understanding of physical boundaries and mixing relationships in the system (Cook and Böhlke, 2000). Several lumped parameter models were considered including the piston-flow model (PFM), dispersion model (DM), and binary mixing models (BMMs) of the PFM and DM. In addition to lumped parameter modeling, apparent groundwater ages were calculated using the ${ }^3\text{H}/{ }^3\text{He}$ method (Suckow, 2014) and He-isotope systematics accounting for accumulation of radiogenic ${ }^4\text{He}$ in the crust due to U–Th decay in the aquifer matrix (Kulongoski et al., 2008).

3. Results and discussion

3.1. Temporal geochemical changes during bulkheading

Temporal variations in geochemistry during flooding of the mine workings is illustrated in Fig. 2, which indicates the response of ${\text{SO}}_4$, Al, Cu, and pH to water impoundment during the May 2018 and September 2020 closures of the hydraulic bulkhead. Water levels measured within the mine workings (PB #1 and PB #2; Fig. 1) rapidly respond to closure of the bulkhead. Water levels in adjacent bedrock monitoring wells (CDOT ROW #1, MW #1, MW #3, Midway #1) and wells completed within the mine workings upgradient of the bulkhead (CDOT ROW #2, Dew Drop #3) showed variable responses to water impoundment. In the case of bedrock monitoring wells two locations (CDOT ROW #3, MW #3) indicated water-level response that was correlated to the response at PB #2, illustrating potential effects of fracture-flow connectivity. Water-level responses were noted in CDOT ROW #2 and Dew Drop #3 consistent with open workings between those locations and the bulkhead. A detailed analysis of physical hydrologic responses is provided in Newman (2023).

Fig. 2.

Temporal variation in water levels within the mine workings measured in well Dew Drop #3 (black lines) and (a) ${\text{SO}}_4$, (b) Al, (c) Cu, and (d) pH (S.U. denotes standard units) in CDOT ROW #2, MW #1, and the bulkhead (Big Five adit). Shaded blue areas indicate periods when the hydraulic bulkhead was closed, and water was being impounded within the mine workings. The black line on all panels shows the water-level elevation within the mine workings through time, measured in meters above National Vertical Datum of 1988 (NAVD88) on the secondary y-axis.

Concurrent with water-level increases in 2018, ${\text{SO}}_4$ concentrations discharging from the bulkhead (Big Five Adit) underwent nearly a 19-fold increase from 275 mg/L to 5120 mg/L (Fig. 2a) before falling back to near pre-impoundment concentrations in mid-2020. Concentrations of Al and Cu at the bulkhead underwent similar increases and decreases (Fig. 2b and c). Concentrations of ${\text{SO}}_4$, Al, and Cu in crystalline bedrock unconnected to the mine workings (MW #1) did not change during water impoundment. During May 2018 closure period, pH measured at the bulkhead decreased from approximately 6 SU to less than 2 SU, then increased again following the dewatering of the mine pool with an external active treatment system. During the September 2020 closure, pH did not show substantial declines (Fig. 2d) because of active amendment of the mine workings with neutralizing agents, but more frequent measurements at the bulkhead by onsite personnel indicated short-term pulses of low pH water during the impoundment event. Efflorescent salts, which are the presumed source of acidity and metals, may have been largely flushed from the mine workings during the May 2018 impoundment and did not have sufficient time to re-accumulate before the September 2020 closure.

Rapid water-quality changes in water within the mine workings caused by the water-level rise after bulkheading in May 2018 are consistent with the dissolution of efflorescent sulfate salts, which are likely to have accumulated within the mine workings. Efflorescent sulfate salts, such as melanterite $\left({\text{Fe}}^{\text{II}}{\text{SO}}_4\cdot 7{\text{H}}_2\text{O}\right)$, römerite $\left[{\text{Fe}}^{\text{II}}{\text{Fe}}_2^{\text{III}}{\left({\text{SO}}_4\right)}_4\cdot 14{\text{H}}_2\text{O}\right]$, copiapite $\left[{\text{Fe}}^{\text{II}}{\text{Fe}}_4^{\text{III}}{\left({\text{SO}}_4\right)}_6{(\text{OH})}_2\cdot 20{\text{H}}_2\text{O}\right]$, coquimbite $\left[{\text{Fe}}_2^{\text{III}}{\left({\text{SO}}_4\right)}_3\cdot 9{\text{H}}_2\text{O}\right]$, halotrichite $\left[{\text{Fe}}^{\text{II}}{\text{Al}}_2{\left({\text{SO}}_4\right)}_4\cdot 22{\text{H}}_2\text{O}\right]$, and pickeringite $\left[{\text{MgAl}}_2{\left({\text{SO}}_4\right)}_4\cdot 22{\text{H}}_2\text{O}\right]$, dissolve rapidly and can produce highly acidic solutions with elevated metal concentrations (Cravotta, 1994; Frau, 2000; Newman et al., 2019). Dissolution of accumulated efflorescent salts has been reported to cause episodic acidic discharges during flooding of other underground mine workings (Cravotta et al., 2014; Elliot and Younger, 2014; Gzyl and Banks, 2007).

3.2. Stable isotopes

Stable-isotopic compositions of waters are illustrated in Fig. 3 and are plotted in comparison to the Rocky Mountain snow-dominated meteoric water line (MWL) of Anderson et al. (2016). Several samples of snow collected in December 2020 (Newman, 2022) plot close to the MWL, but with depleted (more negative) values characteristic of cold weather precipitation. Groundwater wells completed in crystalline bedrock and seeps/springs have a relatively wide range in isotopic composition reflecting groundwater recharge throughout the year and spatial variation in groundwater age and source. Ephemeral seeps (Seep-07 and Seep-08) plot closely to stream water. In contrast, mine-water samples plot in a distinctive cluster that does not overlap with other sample types. Waters in the mine workings have a depleted isotopic signature relative to nearby groundwater wells and plot more closely with the two ephemeral seeps/springs and stream water. The depleted isotopic composition of mine water is similar to the Dinero mine tunnel near Leadville, Colorado (Walton-Day and Poeter, 2009). There are several possible explanations for the depleted isotopic signature of mine waters including recharge at higher elevation, recharge during the winter (i.e., Jasechko et al., 2014), or focused recharge from stream water with a similarly depleted isotopic composition. A combination of these explanations is also possible and could be further evaluated by collection of multiple samples of stream water through time as opposed to the single sample collected in this study.

Fig. 3.

Stable isotopes of water including, (a) samples from monitoring sites and snow samples (collected December 2020), (b) detail view of water-quality sample collection locations. Also included is the Rocky Mountain snow-dominated MWL of Anderson et al. (2016), and (c) variation in ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$ through time (vertical line indicates when bulkhead was closed). Note that panels (a) and (b) have the same symbology, and panel (c) has a unique symbology.

Temporal evaluation of ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$ (Fig. 3c) indicates that most locations have relatively invariant ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$ compositions with the exceptions of Seep-05, Seep-06, and Midway #1. Seep-05 and Seep-06 have depleted isotopic compositions following spring snowmelt which become enriched during the summer, indicating a seasonally changing recharge source to these springs. Midway #1 displays a substantial isotopic depletion in December 2020, after the study area had received snow, possibly indicating a direct fracture-flow source of recharge to the well (e.g., Gleeson et al., 2009). The invariance of ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$ at all other locations indicates that recharge to these locations is less seasonal and occurs over longer timeframes which results in a smoother isotopic signal.

Stable isotopic compositions of aqueous ${\text{SO}}_4$ (${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$, ${\text{δ}}^{34}\text{S}$) and solid minerals (sulfides and efflorescent sulfate salts) provide information on sources of dissolved sulfate. Isotopic compositions of solids from the study site are summarized in Table 2. Sulfide minerals have depleted isotopic compositions typical of this class of minerals (Nordstrom et al., 2007). A sample of primarily galena (based on visual identification and mineralogy from Lovering and Goddard, 1950) has the most depleted ${\text{δ}}^{34}\text{S}$ (−12.13‰), whereas samples likely containing various mixtures of galena, chalcopyrite, and pyrite range from −5.16 to −7.19‰. The single efflorescent sulfate salt sample from the study area had an average ${\text{δ}}^{34}\text{S}$ composition of about −9.70‰ based on replicate analysis. The similarity of the sulfate salt with sulfide minerals indicates that sulfides are the sulfur source of this salt because oxidation of sulfide generally does not result in fractionation. Other studies have found substantial differences in ${\text{δ}}^{34}\text{S}$ between sulfides and sulfates near mine sites (Nordstrom et al., 2007). The similarity in ${\text{δ}}^{34}\text{S}$ between the sulfate salt and sulfides precludes the use of ${\text{δ}}^{34}\text{S}$ in water to differentiate between the influence of sulfate salt dissolution and sulfide oxidation within the mine workings, though this approach warrants assessment elsewhere in areas where sulfides and sulfates have substantially different isotopic compositions (i.e., Nordstrom et al., 2007), because transience in ${\text{δ}}^{34}\text{S}$ may be used to understand temporally varying sources of solutes.

Table 2.

Stable isotopic compositions of sulfide and sulfate minerals.

Sample ID	Sample Collection Location	Mineralogya	${\text{δ}}^{34}\text{S}$, in ‰b
CJ-2	Dew Drop waste rock dump	Galena	−12.13	-
CJ-5	Dew Drop waste rock dump	Chalcopyrite and pyrite	−6.17	−5.16
CJ-8	New California Raise waste rock dump	Galena and chalcopyrite	−7.19	-
CJ-11	New California Raise waste rock dump	Chalcopyrite and pyrite	−6.88	−6.76
CJ-15	Dew Drop waste rock dump (below ore chute)	Efflorescent sulfate salt (unknown mineralogy)	−9.66	−9.74

^aBased on visual inspection and ore deposit character (Lovering and Goddard, 1950; Wahlstrom, 1935);

^bMultiple values are listed where separate analyses were conducted on splits of the sample; – indicates not applicable.

Comparison of ${\text{δ}}^{34}\text{S}$ and corresponding ${\text{SO}}_4$ concentrations (Fig. 4) indicates that wells with elevated ${\text{SO}}_4$ concentrations tend to have depleted ${\text{δ}}^{34}\text{S}$ compositions indicative of sourcing from sulfide minerals or sulfate salts (Seal, 2003; Taylor and Wheeler, 1994). The pattern of elevated ${\text{SO}}_4$ concentrations having depleted ${\text{δ}}^{34}\text{S}$ values is observed in locations within the mine workings (Big Five Adit, CDOT ROW #2) as well as in groundwater near the workings in Tertiary igneous dikes (Dew Drop #1, Midway #1). These results indicate that sulfide oxidation products account for ${\text{SO}}_4$ concentrations exceeding approximately 50 mg/L. These results also indicate that these oxidation reactions are occurring both within the mine void and in fractures in crystalline igneous rocks, even where those fractures are not directly connected to the mine workings (e.g., Dew Drop #1).

Fig. 4.

Sulfate concentration versus ${\text{δ}}^{34}\text{S}$ of water samples. The range of ${\text{δ}}^{34}\text{S}$ in sulfide phases at Captain Jack Superfund Site is displayed in blue shading and the ${\text{δ}}^{34}\text{S}$ of sulfate salts is illustrated in the dotted line.

Stable isotopes of oxygen in sulfate and water may also be used to quantify the predominance of sulfide oxidation reactions and oxidants, assuming that all aqueous ${\text{SO}}_4$ is derived from sulfide oxidation. Sulfide oxidation is commonly assumed to occur via one of two simplified mechanisms (Hubbard et al., 2009; Nordstrom et al., 2007):

$$Fe{S}_2+3.5O_2+H_{2}O\leftrightarrow F{e}^{2+}+2S{O}_4^{2-}+2H^{+}$$ (1)

$$Fe{S}_2+14F{e}^{3+}+8H_{2}O\leftrightarrow 15F{e}^{2+}+2S{O}_4^{2-}+16H^{+}$$ (2)

Oxidation in Eq. (1) occurs with atmospheric oxygen as the oxidant whereas oxidation in Eq. (2) occurs with ferric $\text{Fe}\left({\text{Fe}}^{3+}\right)$ as the oxidant. Ferric iron generally only reaches appreciable concentrations under acidic conditions, which are observed in the mine workings (Fig. 2). The process in Eq. (1) is assumed to only be applicable where a readily available supply of atmospheric oxygen dissolved in water is present, whereas Eq. (2) may be applicable under suboxic conditions. Ferric Fe to drive sulfide oxidation (Eq. (2)) may be replenished via the following process:

$$F{e}^{2+}+H^{+}+0.25O_2\leftrightarrow F{e}^{3+}+0.5H_{2}O$$ (3)

In addition to oxidation driven by either free ${\text{O}}_2$ or free ${\text{Fe}}^{3+}$, pyrite oxidation may also occur under anoxic conditions in the presence of dissolving efflorescent sulfate salts, such as romerite as described by Cravotta (1994):

$$Fe{S}_2+7F{e}^{II}F{e}_2^{III}{\left(S{O}_4\right)}_4+14H_{2}O\leftrightarrow 22F{e}^{2+}+30S{O}_4^{2-}+16H^{+}+90H_{2}O$$ (4)

Eq. (3) also requires atmospheric oxygen (as does Eq. (1)), but only 7 percent (on a molar basis) of that required to drive sulfide oxidation with atmospheric oxygen as the oxidant, whereas Eq. (4) requires no oxygen but a supply of efflorescent salt to be dissolved. Thus, it may be possible for the processes in Eqs. (2)–(4) to occur at low oxygen concentrations where the process in Eq. (1) would be unlikely (Gammons, 2009). These sulfide oxidation mechanisms are important from a remediation perspective because the mechanism in Eqs. (2) and (3) may occur orders of magnitude more rapidly than the mechanism in Eq. (1), depending on other environmental conditions (Hubbard et al., 2009).

Aqueous compositions of ${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$ and ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$ may be used to evaluate the prevalence of different sulfide oxidation mechanisms because of observed fractionation factors between ${\text{δ}}^{18}\text{O}$ in aqueous sulfate, water bound, and atmospheric sources (Nordstrom et al., 2007; Seal, 2003). One commonly used model to estimate the source of oxygen in aqueous sulfate is the general isotope balance model (Hubbard et al., 2009; Taylor and Wheeler, 1994):

$${\text{δ}}^{18}{O}_{SO4}=X\left({\text{δ}}^{18}{O}_{H2O}+{\epsilon }_{SO4-H2O}\right)+(1-X)\left({\text{δ}}^{18}{O}_{O2}+{\epsilon }_{SO4-O2}\right)$$ (5)

where: X = fraction of sulfate-oxygen derived from water molecule; ${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$ = isotopic composition of oxygen in sulfate; ${\epsilon }_{\text{SO}4-\text{H}2\text{O}}$ = fractionation factor associated with the reaction that forms sulfate from a water-oxygen source; ${\text{δ}}^{18}{\text{O}}_{\text{O2}}$ = isotopic composition of oxygen in the atmosphere; ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$ = isotopic composition of oxygen in water; and ${\epsilon }_{\text{SO}4-\text{O}2}$ = fractionation factor associated with the reaction that forms sulfate from an atmospheric-oxygen source.

Values for variables in Eq. (5) are: ${\epsilon }_{\text{SO}4-\text{H}2\text{O}}=4.1$, ${\text{δ}}^{18}{\text{O}}_{\text{O2}}=23.8‰$, and ${\epsilon }_{\text{SO}4-\text{O}2}=-11.4$ (Hubbard et al., 2009). Completion of these calculations yields the fraction of oxygen in ${\text{SO}}_4$ that was derived from the water molecule, with the remainder being assumed to be derived from an ${\text{O}}_2$ (atmospheric) source. The water molecule occurs in both Eq. (1) and Eq. (2), but ${\text{O}}_2$ occurs only in Eq. (1). Therefore, a predominance of water-derived ${\text{SO}}_4$ oxygen indicates that the process in Eq. (2) is the dominant mechanism whereas a predominance of ${\text{O}}_2$-derived ${\text{SO}}_4$ oxygen would indicate that Eq. (1) is dominant. Two other similar frameworks exist for evaluating sulfide oxidation mechanisms, the stoichiometric isotope balance model and sulfite (${\text{SO}}_3^{2-}$)-water exchange model (Hubbard et al., 2009; Taylor and Wheeler, 1994; Seal, 2003). Neither of those models were utilized for this study because previous applications comparing models indicate they produce similar results (Hubbard et al., 2009). Additionally, these conceptual relationships of sulfide oxidation are greatly simplified when compared to the multi-step process occurring at the atomic level (Hubbard et al., 2009; Nordstrom et al., 2007; Seal, 2003). Nevertheless, the general isotope balance model may lend insight about geochemical reactions occurring at the study site by indicating the potential for sulfide oxidation to occur in suboxic conditions.

Results of general isotope balance model calculations for the study site are illustrated in Fig. 5. The fraction of sulfate-oxygen derived from the water molecule (X in Eq. (5)) has an apparent inverse relationship with ${\text{δ}}^{34}\text{S}$, and depleted ${\text{δ}}^{34}\text{S}$ values are indicative of active sulfide oxidation (Fig. 4). Several sites within the mine workings have values of X of approximately 1, indicating that nearly all ${\text{SO}}_4$ could be derived from sulfide oxidation mediated by ${\text{Fe}}^{3+}$ occurring in the absence of atmospheric ${\text{O}}_2$ (i.e., by the process in Eq. (2) or in Eq. (4)). Several wells completed in Tertiary igneous dikes also display this pattern (Dew Drop #1, Midway #1). This result has important implications for the study site because the remediation strategy of hydraulic bulkheading and water impoundment is predicated on the assumption that submerging sulfide materials in the mine workings will limit sulfide oxidation (Walton-Day et al., 2021; Wolkersdorfer, 2008). If sulfide oxidation in the mine workings can occur in the absence of atmospheric ${\text{O}}_2$ then water impoundment is unlikely to limit sulfide. It is worth noting that the applicability of the general isotope balance model to the process in Eq. (4) has not been studied, and cannot be discerned by the dataset presented herein because stable isotopes were not collected during the initial (2018) flooding of the mine workings, when most soluble salts would likely have been dissolved. Future studies should incorporate sampling and analysis of stable sulfate isotopes during initial mine flooding to further explore this process. Based on results in Fig. 5, it appears that water impoundment within the workings at the study site has been unsuccessful in precluding sulfide oxidation during the 2020 filling of the mine workings. These results are corroborated by the balance of ${\text{Fe}}^{\text{II}}$ and ${\text{Fe}}^{\text{III}}$ (Supplementary Material Section S4), and concentrations of DO that decrease through time in the workings after the closure of the bulkhead (Fig. S3b), even though ${\text{SO}}_4$ concentrations during this same time period increased (Fig. 2).

Fig. 5.

Results of general isotope balance model calculations versus ${\text{δ}}^{34}\text{S}$ of water samples. The range of ${\text{δ}}^{34}\text{S}$ in sulfide phases at the site is displayed in blue shading.

Stable carbon isotopes (${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$) assist in understanding redox geochemistry and efficacy of treatment amendments within the workings. Changes in ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ and alkalinity (Fig. 6) indicate that processes controlling carbon mass balance vary spatially across the study site and through time. Several locations appear to be in approximate isotopic equilibrium with a large carbonate reservoir based on substantial alkalinity (Dew Drop #1, MW #2, and MW #3) and ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ in the range −10 to −5‰, the range commonly observed for secondary carbonates Schulte et al. (2011). Wells with substantial alkalinity are also at approximate geochemical equilibrium with calcite, with mean calcite saturation indices (SI) of −0.77 (Dew Drop #1), −0.34 (MW #2), and −0.51 (MW#3). All SI values are provided in Newman (2022). The Midway #1 well and Big Five Adit (prior to water impoundment) also appear to be in isotopic equilibrium with carbonate based on ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$, but potentially with a small or limited reservoir based on the low alkalinity in these locations (Fig. 6). Mean calcite SIs for Midway #1 and the Big Five Adit are −1.43 and −4.17, respectively. Carbonate isotopic equilibrium control on ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ in these sites is consistent with the occurrence of secondary calcite associated with the ore deposit (Lovering and Goddard, 1950) and with previous limestone amendments (U.S. Environmental Protection Agency, 2017). Sites with more negative ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ of approximately −15 to −20‰ (Seep-05, Seep-06) are in approximate equilibrium with soil organic matter from C3 and C4 plants (Sharma et al., 2013; Schulte et al., 2011). Most locations have ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ that are invariant through time, with the exception of CDOT ROW #2 and the Big Five Adit, both of which show decreasing ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ and increasing alkalinity through time. Addition of treatment amendment (${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ of −13.8 to −11.2‰) began in early September 2020, but ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ in CDOT ROW #2 and the Big Five Adit become more isotopically depleted than treatment amendment, eventually reaching the most depleted compositions observed in the study (Fig. 6). Isotopic depletion of this magnitude may result from organic matter oxidation (Sharma et al., 2013). Organic matter oxidation likely operates with dissolved ${\text{O}}_2$ as the electron acceptor, based on decreasing ${\text{O}}_2$ concentrations through time as amendment was added (Fig. S3b). Despite apparent organic matter oxidation in the workings, ${\text{δ}}^{34}\text{S}$, ${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$ and ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$, ${\text{Fe}}^{\text{II}}:{\text{Fe}}_{\text{total}}$, and concentrations of DO (shown in Supplementary Material Section S4), all indicate concurrent and ongoing sulfide oxidation. It is possible that organic matter oxidation has slowed or mitigated some sulfide oxidation, but ongoing organic matter oxidation in the workings would require a long-term source of organic material to be introduced.

Fig. 6.

Samples of ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ versus alkalinity concentrations. Vertical shaded areas represent the measured ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ composition of amendment materials added to the mine workings and compositions of typical organic matter. Carbonate and typical organic matter ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ compositions are from Sharma et al. (2013) and Schulte et al. (2011) respectively. Arrows indicate the geochemical evolution of locations with temporally variable compositions.

3.3. Rare earth elements

The REE are useful for understanding geochemical reactions and groundwater mixing at mine sites (Gӧb et al., 2013; Verplanck et al., 1999). Concentrations of dissolved REE normalized to the chondrite composition (Anders and Grevesse, 1989) are illustrated in biplots in Fig. 7. Biplots use normalization and heavy versus light REE to conceptualize compositions in two dimensions (Noack et al., 2014). For this analysis, the following classifications of REE were used; light rare earth elements (LREE) = La, Ce, Pr, and Nd; middle rare earth elements (MREE) = Sm, Eu, Gd, Tb, and Dy; and heavy rare earth elements (HREE) = Ho, Er, Tm, Yb, and Lu. The chondrite composition was used for normalization because the bedrock in the study area is igneous, thus making shale normalization less applicable. Previous studies of REE in igneous rocks in the area have also used the chondrite composition (Stern et al., 2018).

Fig. 7.

Rare earth element normalization diagram for all monitoring sites. The approximate REE compositions of igneous rocks in the vicinity of the site, including Silver Plume Granite and Tertiary igneous dikes from Stern et al. (2018), is illustrated in gray shading.

The REE normalization diagram in Fig. 7 indicates that there are several distinct sample groupings with no compositional overlap. Samples from within the mine workings plot in a grouping close to the composition of igneous rocks in the area (derived from Stern et al., 2018), whereas monitoring wells completed in crystalline bedrock and perennial seeps/springs plot in distinct groupings that are farther from the composition of igneous rocks. Perennial seeps/springs samples also display negative Ce anomalies consistent with oxidizing conditions (Gӧb et al., 2013; spider diagrams illustrating Ce anomalies are shown in the Supplemental Material Section S6). The similarity of mine-water samples with the REE geochemistry of igneous rocks likely reflects enhanced water-rock reaction under acidic and high ionic strength conditions (Gӧb et al., 2013; Pérez-López et al., 2010) and may also be due to REE enrichments in the mineralized rock. Mine water samples from this study display slight LREE enrichments (evident by plotting in the lower-left quadrant in Fig. 7 and spider plots in Supplemental Material Section S6), somewhat distinctive from previous research on REE in Colorado which indicated MREE enrichment (Verplanck et al., 1999). Samples from Rue and McKnight (2021) in a mineralized area in central Colorado also are enriched in the LREE (La and Ce), but the results of Rue and McKnight (2021) are not chondrite or shale normalized, so it is difficult to compare their patterns to other studies where normalization was used.

Also noteworthy is that bedrock monitoring wells show minor variation through time and do not plot along a mixing line with water in the mine workings, despite the close proximity of several wells to the mine (i.e., CDOT ROW #1 and Dew Drop #1; Fig. 7). This lack of mixing is consistent with large hydraulic gradients between the mine workings and adjacent bedrock that indicates that the bedrock groundwater system is geochemically compartmentalized from the mine workings, similar to those found in other bulkheaded mines (Walton-Day and Poeter, 2009). This application of REE in understanding physical mixing patterns is aided by the data for REE in solids from Stern et al. (2018), which illustrated the elevated REE signature of Tertiary igneous dikes within the study area. Other studies of AMD that apply REE would be aided by understanding the range in REE compositions within host rocks and hydrothermal alteration signatures.

3.4. Environmental tracers and groundwater age

Environmental tracers are useful for understanding timescales and physical controls on groundwater flow and transport (Cook and Böhlke, 2000; Manning et al., 2020), but have been sparingly used at mine sites to quantifying timescales of remediation. Samples were collected from groundwater wells, seeps, and the Big Five Adit in June and December 2020 to quantify temporal changes. The second sample from the bulkhead, collected in December 2020 after bulkhead closure, could not be analyzed for noble gases because of substantial instrument interferences. Interferences likely resulted from a possible excess of high-mass hydrocarbons in the sample, which may have been derived from flooding of the mine workings and addition of treatment amendments.

Noble gases and ${ }^3\text{H}$ may together be used to qualitatively evaluate mixing relationships and quantitatively estimate groundwater age. All samples contained detectable ${ }^3\text{H}$ (Table 3) consistent with a component of recharge since the 1950s and above ground nuclear weapons testing (Suckow, 2014). The sample of Midway #1 (adjacent to the mine workings) contains the second lowest ${ }^3\text{H}$ concentration in this study, potentially indicating the presence of old groundwater higher (further from the adit) in the igneous dike that is inflowing to the workings. Noble gases were not collected from this well however meaning more quantitative groundwater age estimates cannot be completed.

Table 3.

Concentrations of environmental tracers and results of TracerLPM (Jurgens et al., 2012) modeling, model results available in Newman (2022). LPM = lumped parameter model; PFM = piston-flow model; BMM = binary mixing model of piston-flow model and dispersion model; – = not applicable. Mixing of young and old groundwater was only calculated for samples fitting the BMM. Apparent ${ }^4\text{He}$ ages were calculated using site specific ${ }^3\text{H}/{ }^4\text{He}$ ratios derived according to methods in Kulongoski et al. (2008) but were only calculated where terrigenic ${ }^4\text{He}$ concentration was at least 1 × 10⁻⁸ cc/g ${\text{H}}_2\text{O}$ at STP.

Short Site Name	Sample Date	$\text{R}/{\text{R}}_{\text{a}}$	${ }^3\text{H}$, in TU	${ }^3{\text{He}}_{\text{trit}}$, in TU	${ }^3\text{H}/{ }^3\text{He}$ age, in years	Terrigenic ${ }^4\text{He}$, in cc/g ${\text{H}}_2\text{O}$ at STP	Apparent ${ }^4\text{He}$ age, in years	Results of TracerLPM Modeling
								Selected LPM	Total mean age, in years	Mean age of young groundwater, in years	Fraction of young groundwater in mixture	Mean age of old groundwater, in years
Big Five Adit	6/25/2020*	0.745	6.35	2.25	5.4	2.16 × 10−8	518	BMM	82	10	0.85	500
	12/8/2020	-	5.87	-	-	-	-	-	-	-	-	-
MW #2	6/25/2020*	0.353	3.10	31.50	42.9	2.45 × 10−7	4341	BMM	3773	49	0.16	4500
	12/10/2020	0.348	4.58	28.93	35.3	2.46 × 10−7	4366	BMM	3420	49	0.24	4500
MW #3	6/25/2020*	1.542	6.98	61.56	40.6	5.50 × 10−8	976	BMM	334	42	0.70	1000
	12/8/2020	1.559	6.15	57.08	41.4	4.90 × 10−8	869	BMM	251	39	0.75	900
CDOT ROW #1	6/25/2020*	1.133	6.64	11.17	17.5	1.28 × 10−8	1264	BMM	923	15	0.23	1200
	12/9/2020	1.146	6.83	10.99	17.0	1.20 × 10−8	1187	BMM	896	16	0.26	1200
CDOT ROW #2	12/9/2020	-	5.55	-	-	-	-	-	-	-	-	-
Midway #1	12/9/2020	-	3.60	-	-	-	-	-	-	-	-	-
Seep-05	7/3/2020*	1.024	8.42	4.2	7.2	6.24 × 10−9	-	PFM	-	-	-	-
	12/8/2020	1.004	7.87	0.38	0.8	4.41 × 10−9	-	PFM	-	-	-	-
Seep-07	6/23/2020*	1.004	8.70	0.76	1.5	5.49 × 10−9	-	PFM	-	-	-	-
Seep-08	6/23/2020*	0.981	10.02	0.84	1.4	7.09 × 10−9	-	PFM	-	-	-	-

^*indicates sample collected prior to bulkhead closure in September 2020.

Ratios of ${ }^3\text{He}$ to ${ }^4\text{He}$ in samples (R) over the same ratio in air (${\text{R}}_{\text{a}}$) versus the abundance of ${ }^4\text{He}$ over ${ }^{20}\text{Ne}$ are useful for conceptually evaluating mixing of groundwater ages (Fig. 8) because $\text{R}/{\text{R}}_{\text{a}}$ tends to increase because of mixing with mantle sources or decay of ${ }^3\text{H}$ to ${ }^3{\text{He}}_{\text{trit}}$, the latter of which occurs on timescales of decades, whereas ${ }^4\text{He}$ concentrations increases when compared to ${ }^{20}\text{Ne}$ concentrations on a timescale of hundreds to thousands of years due to U–Th decay. These plots therefore conceptually indicate which groundwaters are expected to be old versus young and mixing of old and young groundwater in individual samples (McMahon et al., 2019). Seepage samples have compositions close to equilibrium with air-saturated water (ASW) for the study site, indicating near-equilibration with the atmosphere in the shallow groundwater system. Samples of CDOT ROW #1 and MW #3 have $\text{R}/{\text{R}}_{\text{a}}$ values greater than one indicating decay of ${ }^3\text{H}$ to ${ }^3{\text{He}}_{\text{trit}}$, a qualitative indication of modern (<65-year-old) groundwater recharge. Contrastingly, samples of MW #2 and the Big Five Adit have $\text{R}/{\text{R}}_{\text{a}}$ less than one and elevated ${ }^4\text{He}/{ }^{20}\text{Ne}$ ratios, indicating that groundwater at these locations has a component of groundwater with residence times on the order of hundreds to thousands of years (Fig. 8). Isotopic compositions do not change substantially through time for any sampled waters, indicating that the system is in approximate steady state on the timeframe of the measurements. It was hypothesized that impoundment of water within the mine workings would cause perturbations to the groundwater-age structure. Based on data collected 3 months after closure of the bulkhead steady state conditions still existed. Future sampling at intervals representing a greater proportion of the mean groundwater age in each location would be useful to test if the system is still in steady state.

Fig. 8.

Ratio of ${ }^3\text{He}$ to ${ }^4\text{He}$ in samples (R) over the same ratio in air (${\text{R}}_{\text{a}}$) versus the abundance of ${ }^4\text{He}$ over ${ }^{20}\text{Ne}$. The composition of air-saturated water (ASW) for the site was calculated according to methods described in Kipfer et al. (2002).

Using qualitative understanding of groundwater mixing from noble gases and ${ }^3\text{H}$ lumped parameter models were used to provide quantitative estimates of groundwater ages and mixing relationships (Table 3). Apparent ${ }^3\text{H}/{ }^3\text{He}$ ages (Suckow, 2014) range from approximately 1 to 43 years. Seeps have the youngest apparent ${ }^3\text{H}/{ }^3\text{He}$ ages whereas wells MW #2 and MW #3 have the oldest. Apparent ${ }^3\text{H}/{ }^3\text{He}$ age is generally similar between June and December samples at each location, with the exception of MW #2 and Seep-05. In these two locations, the December samples were younger, potentially indicating the influence of winter recharge through fractures, consistent with stable isotopic shifts at Seep-05 (Fig. 3) and studies at other fractured bedrock sites (Gleeson et al., 2009). The apparent ${ }^3\text{H}/{ }^3\text{He}$ age of water discharging from the bulkhead is approximately 5 years. It is important to note several limitations of apparent ${ }^3\text{H}/{ }^3\text{He}$ ages. These limitations include only being representative of the young component in a mixed water (young bias; McCallum et al., 2015) and assuming a simple model of piston flow in the aquifer (Suckow, 2014).

Although all sampled waters contain a component of modern groundwater, mixing of groundwater with a residence time on the order of hundreds to thousands of years is indicated by ${ }^4\text{He}$ to ${ }^{20}\text{Ne}$ ratios (Fig. 8) and elevated terrigenic helium-4 (${ }^4{\text{He}}_{\text{terr}}$) concentrations (Table 3). Concentrations of ${ }^4{\text{He}}_{\text{trit}}$ and apparent ${ }^4\text{He}$ age were calculated according to methods described in Kulongoski et al. (2008) as described in the Supplemental Material Section S5. Apparent ${ }^4\text{He}$ ages range from approximately 500 to 4500 years (Table 3) and are greatest in well MW #2 (Fig. 8). The apparent ${ }^4\text{He}$ age of the waters in the mine workings (Big Five Adit) is approximately 500 years, indicating that a component of the water discharging from the mine workings predates mining (earliest mining operations ~160 years ago). Mine workings alter local groundwater flow paths and may capture flow from older and/or inactive regions of aquifers (Walton-Day and Poeter, 2009). The Big Five and associated workings have captured a portion of flow from deep within the fractured bedrock aquifer as indicated by the presence of old groundwater and potentiometric gradients at the study site (Newman, 2023). Like apparent ${ }^3\text{H}/{ }^3\text{He}$ ages, apparent ${ }^4\text{He}$ ages also are limited in that they consider only the age of the old water component in the mixture, and therefore do not provide an understanding of the mixing of young and old groundwater.

Tracer-tracer plots for ${ }^3\text{H}$ versus ${ }^3{\text{He}}_{\text{trit}}$ and ${ }^3\text{H}$ versus ${ }^4\text{He}$ (Fig. 9) illustrate the compositions of samples in comparison to several physical models of groundwater flow: the PFM, DM, and a BMM composed of mixing between the PFM and the DM. Lines indicate where samples would fall if they fit one of these models exactly, where the groundwater age varies along the line. Samples falling off the model lines indicate mixing complexity or other processes not considered by the model (Jurgens et al., 2012). Compositions of ${ }^3\text{H}$ versus ${ }^3{\text{He}}_{\text{trit}}$ (representing the young groundwater component; Fig. 9a) plot near the PFM, consistent with simple flow in the shallow groundwater system. Groundwater wells are fit to varying degrees by the PFM (MW #3) or DM (CDOT ROW #1), but no models adequately fit MW #2 in ${ }^3\text{H}$ versus ${ }^3{\text{He}}_{\text{trit}}$ space. The Big Five Adit sample plots below the PFM line indicating flow complexity even for the young component. Comparison of samples to models in ${ }^3\text{H}$ versus ${ }^4\text{He}$ (which conceptualizes flow paths and mixing for the old groundwater component; Fig. 9b) illustrates that all samples have ${ }^4\text{He}$ concentrations greater than explainable by the simple PFM. Most samples plot close to the BMM representing variable mixtures of PFM and DM flow paths.

Fig. 9.

Tracer compositions in samples and the PFM, DM, and BMM in (a) ${ }^3\text{H}$ versus ${ }^3{\text{He}}_{\text{trit}}$ and (b) ${ }^3\text{H}$ versus ${ }^4\text{He}$. The binary mixing zone in (b) displays the range of compositions described by variable mixing between old and young groundwater, as bounded by the range in end-member age estimates of both components.

Based on apparent mixing relationships (Fig. 9) quantitative modeling of groundwater mixing was conducted using TracerLPM (Jurgens et al., 2012). The age of young and old endmembers was defined based on apparent ${ }^3\text{H}/{ }^3\text{He}$ and apparent ${ }^4\text{He}$ ages, respectively. The total mean age of the mixed samples and mixing fractions were then calculated based on minimizing model misfit to observations. Mixing fractions of young groundwater sampled from wells ranges from 0.16 to 0.75 (Table 3), indicating highly spatially variable mixing patterns likely due to complex bedrock fracturing. The fraction of young groundwater discharging from the bulkhead is 0.85, and bulkhead water has a mean age of 82 years. Mean ages are applicable to contaminant transport because they represent the possible interaction of fast and slow flow paths in the aquifer matrix that could each contribute to solute concentrations (McCallum et al., 2015). Mean ages of groundwater wells all predate mining operations (>160 years old) consistent with low ${\text{SO}}_4$ concentrations in these wells, even though CDOT ROW #1 has a ${\text{δ}}^{34}\text{S}$ signature consistent with sulfide minerals (Fig. 4). Groundwater ages and tracers are an effective means of characterizing hydrologic transience (Massoudieh, 2013), and thus the lack of substantial change in mean ages or tracer concentrations through time indicates that substantial transient changes were not imposed on the groundwater system in the crystalline bedrock by impoundment of water within the workings. It is possible that transience is not yet indicated in groundwater age samples because the time period elapsed since closure of the bulkhead (approximately 3 months in the December samples) is a small fraction of the mean ages (ranging from 82 to 3773 years; Table 3). Transience may only be evident after more time has elapsed.

3.5. Conceptual model and remediation implications

The primary goal of hydraulic bulkheads for mine remediation, and specifically at the Captain Jack Superfund Site, is to control AMD discharge from adits to streams or other water bodies. In this goal, the approach of bulkheading has been successful in precluding AMD discharge to Lefthand Creek. There are other ancillary long-term remediation goals that may be desired from bulkheading however, including (1) submergence of sulfide minerals and (2) attainment of stable biogeochemical conditions that do not require active water treatment operations. Insights gained from stable isotopes and environmental tracers in this study allow a conceptual model to be built (Fig. 10) that provides information pertinent to groundwater mixing and transport, sulfide oxidation mechanisms, and timeframes of remediation.

Fig. 10.

Conceptual model of the site (not to scale) based on hydrologic and geochemical data. Piston flow and dispersion flow are denoted using different symbolized arrows. Groundwater residence time (${\text{τ}}_{\text{GW}}$) and residence time in the workings (${\text{τ}}_{\text{W}}$) are differentiated. A conceptual cross-sectional plot of relative intensity of sulfide oxidation, physical groundwater connectivity, and groundwater age mixing is shown on the bottom.

Two distinct groundwater-flow systems are present on the site, a shallow system likely present in the weathered cap of crystalline bedrock and thin overlying alluvium and a deeper system present in crystalline bedrock (Silver Plume Granite, Idaho Springs Formation, and Tertiary igneous dikes; Fig. 1). The shallow groundwater system discharges as ephemeral (Seep-07, Seep-08) and perennial (Seep-05, Seep-06) seeps and springs. The shallow groundwater system is in approximate equilibrium with the atmosphere based on noble gases (Fig. 8) and has seasonally variable recharge based on stable isotopes (Fig. 3). Groundwater residence times in the shallow system are on the order of months to several years (Table 3), not long enough for substantial water-rock interaction based on REE compositions plotting far from equilibrium with igneous rocks (Fig. 7). There is no evidence of AMD migration into the shallow groundwater system. In the crystalline bedrock groundwater system, there are two distinctive compartments which have minimal mixing: groundwater within Tertiary igneous dikes and mine workings and groundwater in Silver Plume Granite and Idaho Springs Formation. Based on distinctive REE (Fig. 7) and stable-isotopic compositions (Fig. 3), these groundwater compartments have minimal interaction, and contaminated groundwater has apparently not discharged laterally from the mine workings into nearby wells despite their close proximity. This is consistent with high lateral hydraulic gradients on the site indicating compartmentalization of the bedrock system (Newman, 2023).

Geochemical processes have changed through time, and several geochemical processes likely operate simultaneously within the flooded mine workings during the 2020 bulkhead closure. During the initial flooding of the workings in 2018, it appears that dissolution of accumulated efflorescent sulfate salts contributed to substantial increases in metal concentrations and decreasing pH (Fig. 2). Flooding of the mine workings during this 2018 closure appears to have removed much of the soluble reactive mass that had accumulated over the preceding decades, as the second bulkhead closure in 2020 produced a smaller flush of ${\text{SO}}_4$ and metals (Fig. 2). During the second bulkhead closure in 2020, sulfide oxidation appears to be occurring both within the mine workings and in adjacent Tertiary igneous dikes based on ${\text{δ}}^{34}\text{S}$ vs. ${\text{SO}}_4$ concentrations (Fig. 4) and models of ${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$ and ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$ (Fig. 5). Sulfide oxidation under suboxic conditions, as indicated by the general isotope balance model (Fig. 5), limits the efficacy of the hydraulic bulkhead in precluding future AMD generation. Conceptual models of sulfide oxidation have substantial uncertainty however (Hubbard et al., 2009), and long-term sampling for analytes providing insight on this process warrants consideration when hydraulic bulkheads are used. Despite isotopic evidence of suboxic sulfide oxidation, ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ indicates that organic matter oxidation was occurring during the initial flooding of the workings (Fig. 6), which would tend to decrease the intensity of sulfide oxidation because oxidation of carbonaceous material would represent an alternative electron donor in lieu of submerged sulfides. Continued organic matter oxidation requires a source of organic material however and is unlikely to continue without active amendments to the mine workings, meaning that the treatment strategy is unlikely to become passive.

Groundwater age is commonly used to assess sustainability of groundwater resources. Although this approach is simplified and other constraints also affect sustainability (Ferguson et al., 2020), groundwater age is a useful metric for understanding how anthropogenic activities may affect groundwater quantity and quality. Groundwater age is applicable to remediation strategies because effects of remediation will be dampened in older or mixed waters (Sanford and Pope, 2013). Water in the mine workings is a mixture of 85 percent young groundwater and 15 percent old groundwater. The mixing of ages means that old groundwater has been interacting with naturally occurring sulfides for several hundred years, predating the mining at the site, whereas young groundwater has been interacting with mine tunnels (likely with increased reaction rates based on oxygenation and increased surface areas) for a minimum of 5 years. Frequent monitoring during recent bulkheading and active remediation has indicated sustained inflow of acidic and concentrated water into the mine workings, despite additions of neutralizing materials and submergence of sulfidic material (meant to preclude sulfide oxidation in the workings and adjacent bedrock). Sustained acidic inflow is consistent with groundwater inflows that are older than the time period of active treatment and indicate that older groundwater is a constant source of solutes and acidity to the workings. Based on a simple assumption of the PFM age for water discharging from the adit (Table 3), it may require 5 years of active treatment to achieve stable geochemistry at the outlet of the mine workings. Because the mine water is composed of mixed waters with more complex flow paths it may require longer than 5 years for remedial actions to manifest within the workings.

An alternative means by which to conceptualize the timeframe required for active treatment is to estimate the residence time of one pore volume of underground mine workings, the flushing of which is needed to remove all reactive mass and to replace older mineralized groundwater with newer treated water. This timeframe may be estimated using the volume of the mine workings and either the rate of recirculation or the discharge rate from the mine workings. Based on geotechnical investigations (Deere and Ault, 2006), the width of the underground workings ranges from 2.5 to 5 m, and the height ranges from 1.8 to 6 m, providing a range in cross sectional area of 4.5–30 m². The primary unknown in calculating volumes of underground mine workings is their extent and connectivity. The lower range estimate for the workings volume between Dew Drop #3 and the Big Five Adit (the length over which active remediation occurs; 450 m) is between 2000 and 14,000 m³. Although the workings between the injection point and the adit are subject to active treatment, the full interconnected workings may include up to 64 km of interconnected tunnels based on evaluation of mine records compiled by Walsh (2008). The upper range estimate of workings volume is therefore between 14,000 and 1,930,000 m³. The estimate of the timeframe needed to recirculate one pore volume of the water between Dew Drop #3 and the adit is calculated using the average recirculation rate of 10.5 m³/day (MineWater, 2021), resulting in a range of residence time in the workings (${\text{τ}}_{\text{W}}$) of 0.53–3.5 years. This indicates that somewhat stable conditions within that section of mine workings may be achieved relatively rapidly. The estimate of the timeframe needed for one pore volume of water from the entire interconnected workings to discharge is based on the full workings volume and the observed average of discharge from the adit before bulkheading of 6.7 m³/day (Newman, 2023), resulting in ${\text{τ}}_{\text{W}}$ of the full mine system ranging from 5 to 786 years. This range is largely in agreement with the observed groundwater ages in and near the mine workings and consistent with continued acidic water within the workings.

It is important to differentiate between the time period of active remediation that may be required and the efficacy of the remediation itself, and how these are related to the variable residence times across the study site. Differences between residence times in the groundwater system (${\text{τ}}_{\text{GW}}$) and residence times within the open mine workings (${\text{τ}}_{\text{W}}$) are conceptually illustrated in Fig. 10, as well as the factors that control each. Groundwater has long residence times, which has allowed substantial interaction with sulfide minerals over years to millennia. Remedial actions will require long periods of time to be manifested in lower influent concentrations to the workings from groundwater (Sanford and Pope, 2013). Water within the limited length of the mine workings subject to active treatment however likely has a much shorter residence time of months to years, and thus active treatment could effectively neutralize water within the workings and reduce effluent metal concentrations. Together these considerations indicate that the active remediation has been effective in the short term but is unlikely to produce long-term changes in the water that flows into the workings.

The long-term goal of remediation is to create a self-regulating geochemical system that can function with no or minimal upkeep (i.e., a passive remediation strategy; Wolkersdorfer, 2008). Groundwater ages and mixing relationships indicate that a self-regulating geochemical system and passive remediation strategies may require years, or decades, to become feasible. Inflow of old groundwater that has interacted with naturally occurring sulfides may continue. These considerations assume steady-state conditions in the mine workings and adjacent groundwater system. Based on data from the crystalline bedrock, it appears that the groundwater system is in steady state, but instrumental limitations did not allow quantification of noble gases in the Big Five Adit sample following flooding, and thus it is unknown if groundwater ages and mixing relationships are changing in the mine workings. If the hydrologic system in the mine workings is not in steady state, it is possible that groundwater ages are decreasing, which may mean that active remediation strategies could be effective over shorter timeframes. Additional environmental tracer samples for this study site could inform ongoing evaluation of remedial strategies.

4. Conclusions

The Captain Jack Superfund Site in northcentral Colorado has been the focus of remediation activities including the emplacement of a hydraulic bulkhead and introduction of carbonaceous material into the impounded mine water. These remediation activities are aimed at precluding AMD discharge into Lefthand Creek and reducing dissolved metal concentrations in water discharged from the bulkhead. In order to provide information relevant to remediation at the study site, a novel and diverse set of geochemical constituents was collected, including stable isotopes, REE, and environmental tracers (noble gases, ${ }^3\text{H}$). The conclusions of this study are:

1. Based on stable isotopes of water (${\text{δ}}^2\text{H}$, ${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$) and REE, groundwater at the site is highly compartmentalized, consistent with statistical analysis of groundwater-level records through time (Newman, 2023). Compartmentalization means that AMD is not affecting the nearby crystalline bedrock aquifer. Clear definition of distinct groundwater systems would have been difficult or impossible using only major ions and trace elements because of the substantial overlap of data in those constituents. These analytical parameters could be used in other studies of AMD when physical groundwater/surface-water mixing and connectivity are primary study questions.

2. Stable isotopes of sulfate (${\text{δ}}^{18}{\text{O}}_{\text{SO}4}$ and ${\text{δ}}^{34}\text{S}$), water (${\text{δ}}^{18}{\text{O}}_{\text{H}2\text{O}}$), and carbon (${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$) track the mass balance of solutes within the flooded workings by indicating that sulfide oxidation may occur under suboxic conditions, even in the presence of treatment amendment material which imparts a unique ${\text{δ}}^{13}{\text{C}}_{\text{DIC}}$ signature through time. These isotopic systems indicate that active treatment is successful in reducing metal concentrations, but that a shift to passive treatment may not be as effective due to the prevalence of suboxic sulfide oxidation. These analytical parameters could be used in other studies of AMD when the mechanisms of solute loading and attenuation are of interest.

3. Groundwater age tracers and stable isotopes of water reveal that water within the flooded workings and adjacent bedrock has a large range of residence times, and most importantly that a portion of the groundwater predates mining at the site. Groundwater residence times may be used to understand the time period active treatment may be required before the system reaches an approximate geochemical equilibrium. These analytical parameters could be used in other studies of AMD when planning for the transition of active to passive treatment of a site.

These results have substantial implications for the remediation of draining mine tunnels through the use of hydraulic bulkheads and illustrate studies that could be undertaken in the future to further constrain processes occurring at similar sites. Both the presence of old groundwater and likelihood of suboxic sulfide oxidation place constraints on the use of hydraulic bulkheads in achieving rapid conditions that do not require treatment. Old groundwater within the mine workings is consistent with unique stable isotopic measurements of other draining mine adits (Walton-Day and Poeter, 2009), and should be further explored on a regional basis to understand the hydrologic and geochemical processes governing the composition of mine waters.

The dataset collected and analyzed in this study highlights the use of stable isotopes and environmental tracers for informing remediation at mine sites, where their use has been historically limited (Elliot and Younger, 2007, 2014; Fey and Wirt, 2007; Manning et al., 2008; Parry et al., 2000; Wellman et al., 2011). In addition to informing remedial activities, environmental tracers may be useful in assigning solute loading to potentially responsible parties (Mahlknecht et al., 2017) and in defining background water-quality conditions useful as remediation goals (Nordstrom, 2015). Because of these benefits, investigations of environmental tracers at other mine sites where remediation is being considered may be useful to evaluate the likelihood of successful remediation.

Data availability

Data available at: https://doi.org/10.5066/P9ZE4872