Rare-Earth Elements as Natural Tracers for In Situ Remediation of Groundwater

Abstract

The utility of rare-earth elements (REEs) as natural geochemical tracers for the analysis of groundwater remediation was examined in several example permeable reactive barriers (PRBs). The PRBs utilize zero-valent iron and organic carbon plus limestone mixtures for contaminant treatment. Zero-valent iron removed REEs from groundwater to below detection levels (2–4 ng/L) and subsequent rebound of REE concentrations in regions down-gradient of the treatment zones was not observed. In addition, REE concentrations within and down-gradient of an organic carbon/limestone PRB were significantly reduced to <1% of influent levels. Thus, REEs are sensitive tracers for evaluating the interaction of groundwater with materials placed in the subsurface for contaminant remediation. Analysis of geochemical tracers for understanding in situ remediation becomes important in situations where down-gradient contaminant concentrations fail to decrease within expected timeframes. The field data indicated that increased solid-phase partitioning of REEs occurred with increasing pH and heavy REEs were preferentially removed compared to light REEs in ZVI systems. In the organic carbon PRB, unexpected negative europium anomalies were observed, revealing new information about redox conditions within the treatment zone. REE concentrations and shale-normalized profiles can be used as natural tracers to better understand in situ technologies for groundwater remediation.

Graphical Abstract

INTRODUCTION

The rare-earth elements (REEs; La to Lu, excluding Pm) have been widely used as natural tracers in studies of regional groundwater migration and groundwater–surface water interactions.^1–8 The application of REEs as natural tracers takes advantage of their uniform trivalent charge, similar ionic radii, and within-group fractionation behavior that is controlled by aqueous complexation and reactions at the mineral–water interface.² REE patterns in groundwater were previously examined in the context of understanding problems such as regional mixing;^3,4,9–12 upward migration of groundwater through fractured rock;^13–15 upward migration of groundwater from shale gas development and underground carbon sequestration;^16–18 assessing radionuclide migration;¹⁹ and, for better understanding groundwater impacted by mine waste materials.^20,21

REE patterns (i.e., total abundance and shale-normalized profiles) in groundwater are the result of pH-redox conditions, solid-water interactions, and aqueous complexation reactions. Interpretations of REE patterns have mainly focused on differentiating between source-related and process-related controls.² Mobility of the REEs is enhanced in acidic water compared to near-neutral or alkaline water.^1,6,22–24 Furthermore, pH appears to exert the greatest control on the overall abundance of REEs in groundwater,⁶ although aquifer solid-phase composition and residence time are likely important factors governing REE concentrations. Attenuation processes of the REEs in groundwater systems include ion sorption to Fe, Mn, and Al oxyhydroxides. For example, the heavy REEs (HREE; Ho-Lu) tend to preferentially sorb to iron oxyhydroxides leading to light REE (LREE; La-Nd) enrichment in solution.^11,22,25,26 On the other hand, HREE enrichment in carbonate-dominated groundwaters can be caused by progressively stronger REE-carbonate complexation with increasing atomic number.⁴ Distinctive mineral precipitation or coprecipitation mechanisms for the REEs have generally remained elusive, possibly reflecting the variable solubilities of pure REE hydroxides and REE-carbonates as well as the challenges of detecting trace amounts of REE partitioning in aquifer solids. Redox conditions can influence the geochemical behaviors of Ce and Eu, providing another potential probe on subsurface conditions.

REEs have not been extensively used in the evaluation of groundwater remediation technologies. However, Grawunder et al.²¹ examined long-term REE patterns in shallow groundwater at a uranium leaching site following mine closure and remediation. Results from this previous study indicated that groundwater REE patterns could be used to detect features of contaminant sources as well as formation and dissolution of secondary phases in subsurface solids.²¹ A common question to answer during performance monitoring of in situ groundwater remediation is whether water has, in fact, passed through and interacted with the emplaced/injected reactive materials used for contaminant removal. This issue becomes especially important in situations where contaminant concentrations in regions hydraulically down-gradient of treatment zones fail to decrease within predicted timeframes. REEs provide a largely untapped tool for examining in situ remediation.

Permeable reactive barriers (PRBs) represent a class of in situ technologies that are implemented as site management tools at contaminated sites and data are available to support the long-term application of these systems for dealing with persistent contaminant plumes.^27,28 PRBs are in situ treatment zones created below ground using trench-and-fill, underground injection techniques, or hydraulic fracturing. They are designed to clean up contaminated groundwater using natural groundwater migration to transport contaminants to a defined treatment zone. Contaminants are removed from groundwater in the PRB via abiotic and/or biotic processes and treated groundwater passes through the permeable treatment zone; eventually, a “clean front” is created on the down-gradient side. In situations where treatment is not achieved within or down-gradient of the PRB along the prevailing groundwater flowpath, forensic studies are usually initiated for subsurface investigation such as more complete geochemical analyses, coring and solid-phase analysis, and/or geophysical surveys to understand the problem.²⁸ New methodologies are needed to help understand such remedy failures and to improve monitoring in the field. The objectives of this study were to examine REE behavior in groundwater at several PRB sites and to evaluate REEs as natural geochemical tracers for the analysis of groundwater remediation efforts.

MATERIALS AND METHODS

We measured concentrations of REEs dissolved in groundwater at three zero-valent iron (ZVI) PRB sites and one organic carbon (wood chip/manure) plus limestone PRB site. The ZVI PRBs were installed to treat groundwater contaminated with volatile organic compounds (VOCs; chlorinated ethenes) and hexavalent chromium²⁹ (Elizabeth City site), arsenic³⁰ (East Helena site), and a mixture of VOCs (New England site). The organic carbon (OC) plus limestone PRB was designed to treat acidity and metal contamination in groundwater (Delatte Metals site).³¹ Where possible, REE concentrations in groundwater were measured in up-gradient (i.e., hydraulically up-gradient of in situ treatment), in-PRB, and down-gradient regions to track concentration profiles along natural flow paths. Background information about the sites examined in this study is provided in the Supporting Information.

Groundwater Sampling and Analysis.

Groundwater samples were collected from 2015 to 2019 using peristaltic (Alexis) or submersible centrifugal (Proactive Mega Monsoon) pumps. Prior to sample collection, parameter measurements were made with a flow-through cell or overflow cell containing calibrated electrodes for pH, oxidation–reduction potential (ORP), specific conductance, and dissolved oxygen. REE concentrations were measured using Inductively Coupled Plasma-Sector Field-Mass Spectrometry (ICP-SF-MS; Thermo Element XR) in medium- and high-resolution modes to resolve potential isobaric interferences. Custom-made interference check standards containing mixtures of Pr, Nd, Eu, Sm, Gd, Tb, and Ba were analyzed with each sample set to evaluate mass resolution and confirm the accurate correction of oxide interferences, particularly BaO⁺ interference on the Eu isotopes (¹⁵¹Eu and ¹⁵³Eu) and the LREE and middle REE (MREE; Sm-Dy) oxide interferences on the HREEs (see the Supporting Information).^32,33 Rhodium (¹⁰³Rh; 1 μg/L) was used as an internal standard. Minimum detection limits for the REEs ranged from 2 to 4 ng/L. Samples were filtered (0.45 μm, Gelman Aquaprep; polysulfone) and preserved with nitric acid (Optima, Fisher Scientific) in the field at the time of sample collection. Thus, here we define dissolved REE concentrations as those passing through a 0.45 μm pore, but we acknowledge that this fraction could include a component of sub-micronsized particles. Duplicate field samples for analysis of reproducibility were randomly collected on about 15% of the samples. Groundwater samples were shipped on ice and stored in a laboratory refrigerator prior to analysis. Quality control samples in the laboratory included duplicates, blanks, matrix spikes, quantitation limit checks, calibration check standards, interference checks, second-source standards, and tertiary quality control samples. Results of field and laboratory quality control samples and sample data are provided in the Supporting Information. Concentrations of major and minor cations/anions, dissolved organic carbon (DOC), and methane were also measured using methods described in the Supporting Information (Table S1).

To compare REE patterns, raw REE concentrations were normalized using the North American Shale Composite (NASC).³⁴ NASC-normalized Ce-anomaly values (Ce/Ce*), Eu-anomaly values (Eu/Eu*) and ratios of HREE to LREE and MREE to LREE+HREE were calculated using the equations provided in the Supporting Information. Theoretical aqueous and mineral equilibrium speciation diagrams were created using the ACT2 module of the Geochemist’s Workbench release 10.0 (Aqueous Solutions LLC) and the Lawrence Livermore National Laboratory (LLNL) thermodynamic database (thermo.com.v8.r6+). The SPECIATE module was used to calculate the distribution of REE complexes with inorganic ions using the LLNL database and updated thermodynamic constants^35–37 (see Supporting Information; Table S2).

RESULTS AND DISCUSSION

REE Behavior at ZVI Sites.

Influent groundwater to the Elizabeth City PRB had slightly acidic pH (5.81 ± 0.15; n = 7), low total dissolved solids (TDS; 329 ± 45 mg/L; n = 7), and sodium-sulfate-type water composition. Redox conditions in the influent favored mobility of hexavalent chromium; groundwater contained low levels of dissolved oxygen (<1 mg/L), no detectable ferrous iron (<0.05 mg/L), and had low concentrations of DOC (mean = 1.6 mg/L). Monitoring wells were screened in 15 cm segments over the depth range from 3.9 to 6.9 m below ground surface (BGS). With increasing depth, pH decreased about one standard unit (from 6.11 to 5.06) and TDS values decreased slightly (351–248 mg/L). Total dissolved REE (∑REE) concentrations ranged from 0.360 to 1.657 μg/L and ∑REE correlated with depth (R² = 0.76). Two depth-dependent clusters were apparent: a deep zone at 4.9–6.9 m BGS (∑REE = 1.21–1.66 μg/L; Figure 1a) and a shallow zone from 3.9 to 4.5 m BGS with lower REE concentrations (∑REE = 0.36–0.43 μg/L) and higher pH. Both groups showed a HREE-enrichment pattern (Figure 1a). The HREE/LREE ratio increased from 1.3 to 5.8 with decreasing depth BGS. In detail, this site also showed a slight positive Gd anomaly and this feature was apparent in multiple sampling events (Figures 1a and S2).

Figure 1.

NASC-normalized patterns of dissolved REE concentrations in groundwater up-gradient of the Elizabeth City PRB (A, screen depth is indicated) and the East Helena PRB (B). Calculated aqueous speciation of REEs is shown as percent distribution of carbonate species, sulfate species, and free cations from Elizabeth City (C) and East Helena (D) (plots show mean values ± standard error).

Speciation calculations showed that REEs were mainly present as sulfate complexes and as free cations in solution (Figure 1c). LREEs were more associated with sulfate compared to MREEs and HREEs. The sulfate-associated percentage decreased from 53.2 ± 11.1 for La to 40.4 ± 11.7 for Lu (Figure 1c). The HREEs tended to form more carbonate complexes than the LREEs; the percentage of REEs as carbonate complexes ranged from 5.0 ± 2.5 for La up to 19.2 ± 10.2 for Lu. The percentage of free REEs stayed about constant in each well at 40.0 ± 7.1. REE complexes with Cl and O anions were consistently low (<1.0%) in all seven up-gradient wells. Results from the Elizabeth City site are summarized above for a sampling event in May 2019; however, consistent patterns in dissolved REE concentrations were observed in annual sampling events from 2017 to 2019 (Supporting Information; Table S3 and Figure S1).

Influent and side-gradient (i.e., locations adjacent to hydraulic flow paths intercepted by the PRB) groundwater at the East Helena PRB site had near-neutral pH (7.06 ± 0.68), moderate TDS (960 ± 135 mg/L) and displayed sodium-bicarbonate-sulfate composition. Groundwater at this site was characterized by an iron-reducing redox signature; dissolved iron concentrations ranged from 0.63 to 3.11 mg/L. DOC concentrations were low (2.5 ± 0.9 mg/L). Dissolved REE (∑REE) concentrations ranged from 0.68 to 11.13 μg/L (Figure 1b). Most groundwater samples from the East Helena site showed HREE-enrichment (HREE/LREE ranged from 2.2 to 3.5) with positive Ce/Ce* values (1.3–3.2). In monitoring well EPA07, we observed distinct changes in the REE pattern between two sampling events in 2015 and 2016 (Figure 1b). This observation was unique. In most cases where monitoring wells were sampled over time, generally consistent REE patterns were observed (Figure S2). Groundwater restoration activities have led to an overall decrease in water levels at the East Helena site and this could be a factor in controlling REE patterns through time.³⁸ The dominant REE species found in East Helena groundwater were sulfate and carbonate complexes (Figure 1d). LREEs were more associated with sulfate and HREEs tended to associate with carbonate. Due to the higher solute concentrations, free REE cations made up <5% of the REE species (Figure 1d). The average percentage of La present as a free species was 5.2 ± 2.6 (n = 4) and for Lu it was 0.78 ± 0.25 (n = 4).

REEs at both ZVI sites were essentially removed from groundwater within and down-gradient of the PRBs. At the Elizabeth City PRB, ∑REE values were <0.01 μg/L in 21 samples collected within the ZVI PRB from 2017 to 2019; whereas, during the same period dissolved ∑REE concentrations in up-gradient groundwater were 1.15 ± 0.58 μg/L (n = 21), indicating over ~2 orders of magnitude reduction in the dissolved REE load (Figure 2a). Note that the sampling period for REEs was 21 to 23 years post installation of the PRB; presumably REEs have been captured in the PRB over the entire history of operation. Coherent REE patterns were not present in groundwater that passed through ZVI; instead, only sporadic detections of individual elements were observed. Single-element REE detections within the PRB at Elizabeth City were LREEs (La, Ce, Pr, and Nd). Furthermore, REE concentrations in groundwater within 2 m (~5 to 8 days of travel time) of the down-gradient edge of the PRB remained low (range <0.002–0.064 μg/L; median 0.016 μg/L). Importantly, there was no apparent rebound effect of increasing ∑REE observed in groundwater down-gradient of the PRB (Figure 2a). This suggests that REE patterns can be used as a near-field probe to determine if groundwater has interacted with ZVI emplaced in the subsurface. Similar trends were also observed at the East Helena PRB. The median ∑REE concentration within the ZVI at East Helena was 0.03 μg/L compared to the median up-gradient concentration of 0.78 μg/L (n = 10; Table S3).

Figure 2.

Relationship between total dissolved REEs and pH. (A) Elizabeth City zero-valent iron PRB: up-gradient groundwater (pink shaded region); within PRB groundwater (tan shaded region); down-gradient groundwater (green shaded region). (B) Delatte Metals organic carbon plus limestone PRB: up-gradient groundwater (pink shaded region); within PRB and down-groundwater (tan shaded region). The arrows trace the relative direction of groundwater flow.

∑REE versus pH trends for the ZVI systems showed a decreasing pattern over the pH range from 5.5 to about 11. Inverse relationships between pH and ∑REE were documented in previous surface water and groundwater studies, e.g., refs 1, 22, 24, 39. Dissolved REE levels were between about 0.001 and 0.1 μg/L within and down-gradient of the ZVI PRBs (moderately alkaline pH), whereas up-gradient groundwater had dissolved REE levels >1 μg/L (Figure 2a). Increased complexation with carbonate ions at higher pH has been proposed to explain the pH-dependent behavior of REEs.^10,35 As pointed out by Noack et al.,⁶ stronger carbonate complexation would favor concentration increases in the MREE and HREE compared to the LREE and increasing ∑REE with pH; however, these relationships are not usually apparent in groundwater. ZVI has been shown to remove REEs from aqueous systems in laboratory studies via iron corrosion in water and REE sorption to mineral corrosion products.⁴⁰ In this study, detections of LREEs were more common than the HREEs within the ZVI media, which is consistent with the experimental observation of removal of HREEs in preference to LREEs in ZVI systems.⁴⁰

Possibilities for REE removal by ZVI include surface adsorption onto granular iron and/or the neo-formed minerals that precipitate in ZVI, co-precipitation, or precipitation of REE minerals, such as REE hydroxides and carbonates. Using La as a geochemical probe, equilibrium modeling revealed that the solids La(OH)₃ and La₃(CO₃)₃·8H₂O were undersaturated in groundwater within the moderately alkaline ZVI systems (pH 9.1 to 10.5; Figure 3). Saturation indices [SI = log₁₀ (ion activity product/K_sp)] for La(OH)₃ and La₃(CO₃)₃·8H₂O from the Elizabeth City and East Helena sites ranged from −5.1 to −6.4 and −2.9 to −5.5, respectively, indicating that precipitation of a pure hydroxide or carbonate was not likely the removal mechanism for La. Therefore, although further study is needed, REE removal in ZVI systems is possibly related to adsorption and/or capture by neo-formed minerals in ZVI that form because of geochemical shifts in pH (e.g., carbonate precipitation),^41,42 microbial activity (e.g., sulfide precipitation), and/or iron corrosion reactions (e.g., oxide and oxyhydroxide precipitation). It should be noted that geochemical environments within the ZVI treatment zones were moderately alkaline and reducing, supporting both sulfate-reducing and methanogenic activity.²⁸

Figure 3.

Groundwater up-gradient (red circles), within and down-gradient of the Elizabeth City PRB (black squares) was undersaturated with respect to La(OH)₃ and La₂(CO₃)₃·8H₂O, indicating that lanthanum hydroxide or lanthanum carbonate precipitation was not the controlling uptake mechanism for La. Formation of a mixed precipitate would be expected to form a more insoluble solid compared to pure La(OH)₃. The stability field for La(OH)₃ is shaded purple. Solubility trend for La₂(CO₃)₃·8H₂O is shown with the dotted line for 20 ppm of total inorganic carbon (IC).

REE Tracer Application at an Injected ZVI PRB.

To further examine the utility of REE tracer analysis for understanding PRB behavior at ZVI sites, groundwater samples were analyzed from a New England site. At this site, a thin 100% ZVI PRB was installed using hydraulic fracturing methods. Several years of quarterly monitoring (starting March 2015) showed concentration profiles that were inconsistent with the expected degradation of chlorinated hydrocarbons. Because of the installation method used to construct the PRB, direct monitoring or sampling of groundwater from within the treatment zone was not possible. For example, direct monitoring within wider PRBs is conducted to evaluate whether the typical geochemical signature of ZVI–groundwater systems is present, such as elevated pH, reduced oxidation–reduction potential, decreased specific conductance, and reduction in contaminant concentrations relative to up-gradient locations.²⁷

In August 2017, discrete multilevel samples were collected from 8 locations along the ~110 m long PRB. Groundwater samples were collected in pairs along the PRB to provide corresponding up-gradient and down-gradient positions across the expected treatment zone. Analyses of REE concentrations indicated no statistical difference (nonparametric; Kruskal-Wallis) between the up-gradient and down-gradient locations (p-value = 0.44; Figure 4a). In all cases, ∑REE concentrations on the down-gradient side of the PRB were >0.24 μg/L and were as high as 8.89 μg/L. Inspection of individual up-/down-gradient sample pairs provided additional context. Along the southern and central regions of the PRB, ∑REE was lower on the up-gradient side in all cases but one (Figure 4b). These data indicated that down-gradient groundwater along this region of the PRB did not significantly interact with ZVI, which is consistent with the lack of observed contaminant degradation. Although several locations along the northern section of the PRB showed decreases in ∑REE from up-gradient to down-gradient positions, the net REE decreases were inconsistent with extensive ZVI–groundwater interactions based on results shown in Figure 2a for other functioning ZVI PRBs.

Figure 4.

Standard box-and-whiskers plots showing up-gradient versus down-gradient ∑REE concentrations at a thin ZVI PRB in New England (A) and paired up-gradient/down-gradient samples along regions of the PRB (B). REE concentration patterns indicate that groundwater at this site has not effectively interacted with ZVI.

The Elizabeth City and East Helena ZVI PRBs are examples of in situ remediation systems in which contaminated groundwater effectively interacts with the emplaced reactive media. This interaction between groundwater and the PRBs results in desired contaminant treatment as well as modified groundwater geochemistry, such as shifted pH, redox, and major ion patterns.^27,29,30 For example, chromium concentrations were not detected within or down-gradient of the PRB at the Elizabeth City site.²⁷ Removal of trichloroethylene (TCE) has ranged from 81 to >99% with evidence of decreased treatment efficiency after 18 years of operation,²⁸ yet over this same period REEs were effectively removed from groundwater as the plume passed through the PRB. In contrast, the New England site showed negligible contaminant treatment and unchanged REE patterns along the groundwater flowpath crossing the presumed location of the PRB. These observations suggest that groundwater has not interacted with the emplaced ZVI at the New England site.

REE Behavior at an Organic Carbon/Limestone PRB.

The Delatte Metals PRB was installed in 2003 to treat acidity and elevated concentrations of lead (Pb), cadmium (Cd), and nickel (Ni) in groundwater at a former lead battery recycling facility.³¹ Influent groundwater to the PRB in transects 1, 2, 5, and 6 (see Supporting Information; Table S3) had elevated ∑REE (mean = 973 ± 360 μg/L; n = 8) due to low groundwater pH (3.1–3.4; Figure 2b). Speciation calculations indicated that REEs occurred mainly as sulfate complexes (>90%) and free-ion species in the groundwater up-gradient of the PRB. For acidic and sulfate-rich groundwater, it is important to enter aluminum and iron in the input basis for speciation calculations because these elements can be present in significant concentrations (Al as high as 250 mg/L; Fe as high as 124 mg/L). Thus, Al and Fe can have an overall effect on the outcome of speciation calculations by impacting the aqueous partitioning of sulfate ions.⁴³

Normalized patterns of influent groundwater showed MREE enrichment (MREE/LREE + HREE = 1.43 ± 0.03; n = 8) and positive Ce/Ce* values (1.48 ± 0.06; n = 8). MREE enrichment is frequently observed in acid-mine drainage environments; yet, the general cause of this enrichment in natural waters has remained puzzling.^10,43–46 The primary uncertainty is whether the MREE patterns are inherited from source materials (i.e., weathering of metal sulfides or secondary minerals), derived from mineral–water reactions, or related to aqueous complexation. Because the low-pH groundwater, in this case, is not linked to mine water or caused by pyrite-oxidation reactions,⁴⁶ it is possible that MREE enrichment is related to acid dissolution or desorption from primary or secondary aquifer solids.⁴⁷ It is reasonable to also consider whether REE patterns reflect the strength and variation of the stability constants of the dominant aqueous species across the REE series.⁴⁸ The stabilities of the REE–sulfate complexes vary slightly across the lanthanide series and show a convex-up pattern,³⁷ so it is possible that the formation of REE–sulfate complexes contributes to the development of MREE enrichment in these acidic and sulfate-rich waters. The relative changes in the stability constants are small but comparable to the magnitude of MREE enrichment (generally <0.1 log units; ~1.3×); however, the maximum stability constants are apparent for Eu in the experimental data, whereas the field data generally indicate maximum normalized concentrations for Gd.

The positive Ce/Ce* values observed in low-pH groundwater up-gradient of the PRB are unusual. Negative Ce anomalies have been suggested to result from oxidation of Ce(III) to Ce(IV) and removal of Ce(IV) due to the solubility limits of CeO₂.⁴⁹ Positive Ce/Ce* values, on the other hand, indicate an enhanced tendency of Ce to remain or be present in solution compared to its REE neighbors, La and Pr. Positive Ce/Ce* values have been documented, for example, in alkaline carbonate-rich waters and are thought to be related to stabilization of Ce(VI)–carbonate complexes.⁵⁰ However, for this study, Ce(VI) was not likely to be present in low-pH groundwater at the measured E_H conditions, from about 0.3 to 0.5 volts (Supporting Information; Figure S3). One possibility for the positive Ce/Ce* values relates to acid dissolution and reduction of previously formed CeO₂ precipitates or coprecipitates contained in the aquifer solids in the presence of dissolved Mn(II) and/or Fe(II)

$$2{\text{CeO}}_{2\left(\text{s}\right)}+{\text{Mn}}^{2+}+4{\text{H}}^{+}=2{\text{Ce}}^{3+}+{\text{MnO}}_{2\left(\text{s}\right)}+2{\text{H}}_2\text{O}$$ (rxn 1)

$${\text{CeO}}_{2\left(\text{s}\right)}+{\text{Fe}}^{2+}+{\text{H}}^{+}={\text{Ce}}^{3+}+{\text{FeOOH}}_{\left(\text{s}\right)}$$ (rxn 2)

Up-gradient groundwater was elevated in both Mn (0.98–10.0 mg/L) and Fe (0.25–123.9 mg/L) and the measured redox and pH conditions were consistent with the presence of the divalent forms of these metals. Field measurements using the 1,10-phenanthroline method showed that 37–98% of the total Fe was present as Fe(II). According to Le Chatelier’s principle, these redox reactions (rxn 1 and rxn 2) could explain increases in groundwater Ce concentrations with increases in Fe and Mn. Taken in total, MREE enrichment and positive Ce/Ce* values suggest that the up-gradient distribution of REEs may be controlled by acid dissolution and enhanced mobilization of Ce and MREE, i.e., related to the main source contribution of REEs from aquifer solids.

Groundwater is neutralized in the PRB by reactions involving dissolution of limestone. In addition, alkalinity and dissolved sulfide are produced by microbial sulfate reduction.^31,51,52 The increased alkalinity, pH, and production of sulfide have led to the removal of metal contaminants from groundwater, such as Pb, Cd, and Ni.³¹ As the groundwater pH increased, ∑REE levels decreased to <1% of the influent (mean = 4.0 μg/L; n = 8). ∑REE levels remained low down-gradient of the PRB (mean = 2.2 μg/L; n = 16; Figure 2b). This reveals that groundwater had in fact passed through the reactive material and that the down-gradient REE signature was governed by processes occurring within the PRB. At this site, the residence time of groundwater recorded in the furthest down-gradient wells sampled is expected to be ~150 d based on the distance between the PRB and the wells (~9 m) and an estimated groundwater flow velocity of ~6 cm/d.³¹ Within and down-gradient of the PRB, sulfate complexes dominated the speciation of La and Ce; however, carbonate complexation became more important (Supporting Information; Figure S5). For example, for the HREEs Lu and Yb, sulfate and carbonate complexes made up about 16 and 80% of the dissolved species, respectively. The pattern of HREE enrichment in groundwater down-gradient of the PRB was possibly caused by carbonate-dominated speciation and the fact that Fe-oxyhydroxides were largely dissolved by interactions with reducing groundwater exiting the PRB.

The groundwater composition within and down-gradient of the Delatte Metals PRB is complex. For example, concentrations of DOC ranged from 4.2 to 33.3 mg/L (mean = 11.2 mg/L; n = 24). At higher concentrations of DOC, it is expected that organic matter complexes may have been the dominant form of dissolved REE.^15,50,53 DOC-REE complexation processes have been described in other studies and they may have a strong effect on REE speciation at DOC levels >2 mg/L, largely from HREE–DOC interactions. DOC–REE interactions represent an important aspect of utilizing REEs as tracers in remediation systems that rely on OC-driven processes, i.e., mainly to reduce redox potential for contaminant degradation (i.e., chlorinated solvents) or sequestration via precipitation (metals).

Following interaction with the organic carbon/limestone PRB, groundwater sometimes showed a negative Eu anomaly (Figure 5; Supporting Information, Figure S6). Eu/Eu* values in groundwater ranged from 1 to about 0.2 within and down-gradient of the PRB. Eu/Eu* values <1 were associated with low Eu concentrations, low groundwater E_H, and high dissolved methane and sulfide concentrations, suggesting Eu reduction (Figure S6). The negative anomaly indicates that Eu was preferentially removed from groundwater relative to the adjacent trivalent REEs (Sm and Gd). In this case, there is a confidence that the Eu feature is not source-controlled because negative anomalies were not observed in up-gradient groundwater. E_H-pH considerations indicated that groundwater conditions up-gradient of the PRB were within the Eu³⁺ stability field. Further, in-wall and down-gradient groundwater remained within the Eu³⁺ field (Supporting Information, Figure S4). Thermodynamic considerations predict that the likelihood is low for the presence of Eu²⁺ in natural waters because the stability field of Eu²⁺ is restricted to a narrow pH range at very reducing conditions⁵⁴ (see Figure S4). Yet, previous studies have reported unusual behavior of Eu that could be related to reduction.³⁹ Samples containing the negative Eu anomaly coincide with the predicted pH range for Eu²⁺ (6–8; Figure S4); however, the estimated E_H values from ORP measurements did not predict Eu²⁺. Moreover, calculated E_H values based on the CH₄/bicarbonate and H₂S/sulfate redox pairs suggest slightly more reducing conditions than were apparent from the ORP measurements, but still not within the predicted Eu²⁺ field (Figures S4 and S6). It is possible that more reducing mineral or organic surfaces may have allowed the conversion from Eu³⁺ to Eu²⁺ and redox conditions were not accurately depicted with measurements of ORP or other redox pairs in groundwater.⁵⁵

Figure 5.

NASC-normalized REE patterns from the Delatte Metals organic carbon plus limestone PRB (transect 5). High REE concentrations are present in the acidic up-gradient groundwater (red symbols; with positive Ce/Ce* values and MREE enrichment). REE concentrations were reduced within (black symbols) and down-gradient of the PRB (purple symbols). Negative Eu/Eu* values are apparent within and down-gradient of the PRB. The asterisk depicts the analytical detection level for Eu.

Implications for In Situ Remediation.

Treatment efficiency and longevity of in situ groundwater remedies is tracked by monitoring reductions in contaminant concentrations through time and in space using sampling transects that cross the expected position of subsurface treatment zones. For ZVI systems, there are several geochemical proxies that can be used to assess whether ZVI corrosion reactions, and presumably contaminant transformation/removal reactions, are taking place in the subsurface, such as shifts in pH and redox conditions and changes in dissolved gas compositions (i.e., production of methane, hydrogen, ethene, ethane).^27,28,56,57 Organic carbon systems are also evaluated using geochemical proxies, such as increases in alkalinity, DOC, and methane.⁵⁸ The primary goal of monitoring programs is to understand groundwater flow in three dimensions, to ensure that contaminant plumes are intercepted for treatment, and to demonstrate that contaminant transport beneath, around, or above the treatment system does not occur.⁵⁹

In situations where monitoring programs for tracking in situ remedies fail to show effective contaminant treatment in predicted timeframes, follow-up studies to understand the remedy failure (or remedy delay) are required. One of the first questions to answer in these situations is whether the contaminant plume has in fact interacted with reactive materials emplaced within the subsurface environment. Monitoring data showing ineffective treatment can result if contaminant plumes by-pass the reactive medium or if contaminant sources are present in the aquifer on the down-gradient side of the treatment zone. We propose that REE concentrations and patterns can be used as tracers to probe the key question of whether groundwater has interacted with reactive materials emplaced using subsurface injection or trench-and-fill methods. In two field examples of performing ZVI PRBs, REE concentrations were reduced to levels below detection (<4 ng/L). REE concentrations were also significantly reduced at an organic carbon/limestone PRB that successfully treats an acidic plume containing Pb, Cd, and Ni. In these cases, REE concentrations and patterns point to the fact that contaminated groundwater passed through the treatment zones. In one example of a nonfunctioning ZVI PRB (i.e., no indicated contaminant reduction), REE levels and patterns were unchanged across the flowpath presumed to intercept the treatment medium. The implication at this site is that the ZVI was not appropriately placed to intercept and treat the contaminated groundwater plume. This finding has led to subsequent site investigations to better understand subsurface conditions.

The abundance of REEs in groundwater influent to the PRBs investigated in this study was largely controlled by pH with lower pH favoring higher REE concentrations. Thus, the pH of up-gradient/contaminant-bearing groundwater is an important control on whether these natural tracers will be informative. Results indicated that greater contrast in REE concentrations between influent and effluent groundwater should be expected in situations where contaminated groundwater plumes are acidic to near-neutral in pH. For example, in moderately alkaline groundwater from ZVI systems, REE concentrations were below levels of detection by ICP-SF-MS using medium- and high-resolution detection modes for resolving spectral interferences. Thus, for moderately alkaline to alkaline groundwater (pH > 8.5), analytical detection levels may become a limiting factor for using the REEs as natural tracers.