Molybdenum Release Triggered by Dolomite Dissolution: Experimental Evidence and Conceptual Model.

The injection of oxygenated water into anoxic aquifers during managed aquifer recharge (MAR) can cause the mobilization of metal(loid)s. Here, we study the processes controlling MAR-induced molybdenum (Mo) release in dolomitic aquifers. Sequential chemical extractions and energy dispersive X-ray spectroscopy combined with scanning electron microscopy point to an association of Mo with easily soluble sulfurized organic matter present in intercrystalline spaces of dolomites or directly incorporated within dolomite crystals. The easily soluble character was confirmed by a batch experiment that demonstrated the rapid mobilization of Mo, dissolved organic carbon, and sulfur. The type and time of batch solution contact with the sulfurized organic matter impacted the release of Mo, as demonstrated by a 36% increase in Mo concentrations when shaking was intensified. Based on the experimental results, a conceptual model for the release of Mo was formulated, where (i) the injection of oxygenated water causes the oxidation of pyrite in the aquifer matrix, and (ii) the associated release of protons (H+) induces the dissolution of dolomite as a buffering reaction, which (iii) enhances the accessibility of the injectant to intercrystalline and incorporated sulfurized organic matter within dolomite, causing the release of Mo.

Introduction

The mobilization of metals and metalloids in groundwater environments due to anthropogenic disturbances of naturally prevailing aquifer conditions is a common occurrence.[1−10] Among those, metal mobilization in conjunction with the redox shifts that occur during the injection of oxygenated or nitrate enriched water into anoxic aquifers during managed aquifer recharge (MAR) activities has been widely reported.[11] The current prediction is that at least 10% of the world’s drinking water supply will be provided through MAR applications in the near future.[12]

The release of arsenic (As) during aquifer storage and recovery (ASR) operations has been described for many sites in Europe,[13] North America,[14] and Australia.[15,16] Similarly, MAR operations have been shown to cause Mo concentrations exceeding the WHO-recommended value for drinking water of 70 μg/L,[17−19] while Mo concentrations in most native freshwaters are significantly below 10 μg/L.[20] Compared to As, which is generally present in pyrite,[21,22] the primary source of Mo in sediments remains more controversial. Several studies described an association of Mo with iron sulfides.[23−26] For example, Vorlicek et al.[26] proposed a reaction pathway for the binding of Mo with pyrite. Bostick et al.[23] observed strong adsorption of tetrathiomolybdate (MoS42–) on pyrite and attributed it to strong inner-sphere complexes that persisted even at alkaline pH ranges. Prommer et al.[16] reported that the injection of highly treated excess water from coal seam gas extraction into a deep sandstone aquifer in Queensland (Australia) resulted in the joint release of Mo and As in conjunction with the oxidation of pyrite. On the other hand, Chappaz et al.[27] found 80 to 100% of Mo associated with the non-pyritic matrix of euxinic muds and shales. They concluded that pyrite is a nontrivial sink for Mo but not the primary host phase in most euxinic shales. Instead, several studies observed correlations between Mo and total organic carbon (TOC).[28−30] Parnell et al.[31] observed Mo associated with organic-rich laminae in shales. Studies by Tribovillard et al.[32] on Mesozoic geological formations indicated a positive correlation of sulfurized organic matter (OM) with Mo, and Dahl et al.[33] proposed Mo removal in natural sulfidic waters via particulate OM.

Data compiled by Fischler et al.[19] indicated that 8 of 13 ASR facilities in Florida showed elevated Mo concentrations. Among those, exceptionally high Mo concentrations were reported for the ASR facility in Orange County (Florida), where the injection of oxygenated drinking water into an anoxic dolomite aquifer resulted in Mo concentrations of up to 853 μg/L.[19] Given the strategic importance of ASR operations for seasonally storing excess waters to enhance overall drinking water availability in Florida and many other parts of the world, it is essential to develop a process-based understanding of the factors controlling elevated Mo concentrations. This understanding is critical for determining the long-term risks and underpinning the development and application of effective, engineered mitigation strategies, such as modifying the injectant treatment process.[34] A model-based analysis of field data in Lithia (Florida) suggested that the mineralization of OM by oxygen could be the governing process for the release of Mo in the Floridan Aquifer System.[5] However, studies by Pichler and Mozaffari[35] that investigated the carbonates of the Floridan Aquifer System showed the extraction of more than 70% Mo by sodium acetate at a pH of 8.1. They concluded that Mo might be weakly bound to mineral surfaces and OM.[35] The latter could imply that besides oxidation processes, perturbations of the ambient groundwater pH play an essential role in the release of Mo in the Floridan Aquifer System. However, to date, no rigorous investigation of the rapid solubilization of Mo has been undertaken to support this hypothesis.

The objective of this study was, therefore, to determine the source of Mo and its influencing factors based on more detailed geochemical investigations and to formulate a conceptual model for its release from the aquifer matrix during ASR operations and for similar geochemical settings at other sites. The Orange County (Florida) ASR facility was selected for our investigations (i) based on the elevated Mo concentrations and (ii) the extensive long-term data set documenting hydrochemical patterns, including Mo release, over several ASR cycles.

Material and Methods

Study Site

Our study site, an aquifer targeted by the Orange County ASR operations, is located in east-central Florida as part of the St. Johns River Water Management District (SJRWMD). It includes one ASR well for injection and two monitoring wells at a distance of 30.5 and 152.4 m from the injection well.[36] The storage zone extends between a depth of 318.5 and 362.7 m, therefore coinciding with the Eocene Avon Park Formation of the Lower Floridan Aquifer in the Floridan Aquifer System.[36−38] This formation was deposited under cyclic shallow open marine to tidal-flat conditions with sometimes restricted seawater circulation and arid climate.[39−41] In Orange County, the formation is dominated by dolomite and dolomitic wackestones that consist of dolomite, calcite, trace amounts of clay, natural OM, and pyrite in the form of framboids and euhedral crystals.[36,42] The dolomitization took place during the middle Eocene by normal to hypersaline seawater.[43,44] Gilboy[45] described the occurrence of salt marsh and swamp deposits of peat and carbonaceous material in the Avon Park Formation. Sediment samples were collected from core W-18722 at the Core and Cuttings Repository of the Florida Geological Survey in Tallahassee (Florida). This core was obtained during the drilling of the monitoring well at a distance of 152.4 m from the injection well. Three samples (43 OC, 55 OC, 63 OC) were selected for this study. The selection targeted samples showed concentrations of Mo above 150 mg/kg. While samples 43 OC (depth: 338.3 m) and 55 OC (depth: 354.6 m) were collected directly from the storage zone, sample 63 OC (depth: 365.9 m) was located below the storage zone.

Mineralogical Characterization

Samples 43 OC and 55 OC were freeze-dried and prepared for X-ray diffraction (XRD) according to Vogt.[46] The samples were measured from 3 to 85° 2θ with a step size of 0.017° 2θ and a step time of 100 s on a Philips X’Pert Pro MD X-ray diffractometer with a Cu tube (Kα, λ 1.541). Polished thin sections of the samples 43 OC, 55 OC, and 63 OC were examined using an optical microscope. Individual mineral phases and OM were analyzed in thin sections selected from samples 55 OC and 63 OC using a Zeiss field emission scanning electron microscope Supra 40 with a Bruker energy disperse X-ray spectroscopy (EDX) detector. The compositions of powellite and pyrite crystals in thin sections from samples 43 OC and 55 OC were measured with a CAMECA SX100 Electron Probe Microanalyzer.

Sequential Extractions

The distribution of Mo in samples 43 OC and 55 OC was investigated using two sequential extraction procedures for carbonate samples, which were based on Scheplitz et al.[47] In the first procedure (Table ), the amount of powdered sediment was doubled to 0.5 g, and equally, the reagents were doubled compared to the original method. Furthermore, one H2O rinse was applied after each step, and the wash solutions were added to the extract. The second procedure was further modified to ascertain the complete dissolution of dolomite. Therefore, step 2 of the sequential extraction procedure (Table ) was repeated 5 times before washing. This was deemed necessary because the procedure of Scheplitz et al.[47] was optimized for calcitic and aragonitic but not for dolomitic matrices. Triplicates of each sample were analyzed together with three procedural blanks. The concentration sums of each step were compared to the total concentrations determined by aqua regia digests that were carried out using an additional 0.5 g of each sample.

Table 1

Sequential Extraction Procedure for Carbonate Samples, Based on Scheplitz et al.[47]

step	phase	reagent	reagent information	procedure
1	adsorbed/exchangeable	10 mL 1.0 M NH4CH3COO (pH 8.2)	reagent grade (Fisher Chemical, USA)	2 h leach, 5 mL H2O rinse
2	carbonates	40 mL 1.0 M NH4CH3COO (pH 5.0)	reagent grade (Fisher Chemical, USA)	2 h leach, 5 mL H2O rinse
3	hydrous iron oxides	10 mL 0.25 M NH2OH·HCl* in 0.25 M HCl**	*ReagentPlus grade (Sigma-Aldrich, Germany)	2 h bath at 60 °C, 5 mL H2O rinse
			**purified by sub-boiling from analytical grade acids (Merck, Germany)
4	crystalline iron oxides	15 mL 1.0 M NH2OH·HCl* in 25% CH3COOH**	*ReagentPlus grade (Sigma-Aldrich, Germany)	3 h bath at 90 °C, 5 mL H2O rinse
			**trace metal grade (Fisher Chemical, USA)
5	sulfides/organic material	10 mL aqua regia	7.5 mL HCl, 2.5 mL HNO3; purified by sub-boiling from analytical grade acids (Merck, Germany)	∼12 h bath (2 h at 120 °C)

To investigate the influence of oxygen on the release of Mo, extraction step 1 (Table ) was repeated once again under oxic and anoxic conditions. To reach anoxic conditions, the reactant was degassed with argon for 30 min, and the preparation for the extraction was done in a nitrogen-filled glovebox.

The easily soluble phase of Mo in extraction step 1 was further investigated for its dependence on pH and the presence of nanoparticles. This experiment was performed only with sample 55 OC because insufficient sediment material was available for sample 43 OC. To ascertain that the method was as close as possible to that applied by Pichler and Mozaffari,[35] the extraction was performed with 1 g sediment and 20 mL 1.0 M sodium acetate (NaCH3COO, ACS, reag. Ph Eur, Merck, Germany). Therefore, the influence of different reagents could also be tested. The effect of pH was assessed by adjusting the sodium acetate solution to pH 8.2, 7.5, and 7.0. One blank per pH value was included. The extracts were filtered with 0.2 and 0.015 μm membranes to check for the presence of nanoparticles. After each extraction, the remaining sample material was analyzed for total carbon (TC), TOC, and total sulfur (TS) with a Leco CS 744 instrument. The inorganic carbon was removed with 12.5% HCl to determine TOC. The contents of TC, TOC, and TS were also determined on a fresh sediment sample.

All extracts were analyzed for Mo, Ca, Mg, Fe, As, S, and Sr using a PerkinElmer Optima 7300 DV inductively coupled plasma–optical emission spectrometer (ICP–OES). The calibration standards were matrix-matched for each extraction step following Scheplitz et al.[47] An artificial multi-element standard indicated an error of precision below 10% for all analytes. Triplicates generally deviated by less than 10%. The average of the triplicates was used in the data interpretation.

Batch Experiment

The injection of oxygenated water into the aquifer was mimicked by a batch experiment with sediment material from sample 63 OC. Drinking water devoid of Mo and with known concentrations of Ca, Mg, Na, K, HCO3, SO4, and Cl were selected for the experiment based on its similarity with the injection water employed at the ASR facility. Both waters were classified as “magnesium bicarbonate” type. The drinking water was equilibrated with atmospheric oxygen for 3 days and adjusted to pH 8.0, with the same pH as the injection water,[48] before a sample was taken to determine the initial water composition.

A total of 14 batch vessels were filled with 1.5 g of freeze-dried, powdered sediment material, and 15 mL of drinking water. Two vessels were filled only with drinking water for procedural control. Half of the samples were constantly shaken at 3 rounds per minute (rpm) in an overhead shaker. The other half was intermittently shaken by hand every 30 min for the first 5 h and then every 24 h, to ascertain the influence of surface contact between drinking water and sediments. Samples were taken after 30 min, 1 h, 2 h, 5 h, 24 h, 3 days, and 7 days. The drinking water for procedural control was sampled after 24 h and after 7 days. The samples were measured for pH, dissolved oxygen (DO), electrical conductivity (γ), and temperature (T) with a Multi 3430 digital meter (WTW), centrifuged at 3000 rpm for 5 min, decanted, and subsequently flitered through a 0.45 μm membrane and stored for further analyses.

Concentrations of Mo, Ca, Mg, Fe, As, S, and Sr were measured by ICP–OES. The measurement was performed with a 2% HNO3 matrix and a dilution of 1:3. The precision was checked with an artificial standard and showed errors below 3 % for all analytes. The accuracy for Mg, Ca, and S was further checked with an internal standard, and errors were below 4%. Acid blanks showed no contamination during the analyses. The sulfate concentration was determined by ion chromatography (IC) using a Metrohm 883 Basic IC plus. The accuracy and precision of the measurement were checked with an internal standard and had errors below 10%. The concentration of bicarbonate (HCO3–) was determined using the flow-injection method described by Hall and Aller[49] and Lustwerk and Burdige.[50] The sample was injected into a carrier stream with 30 mM HCl, which converted inorganic carbon to carbon dioxide (CO2). Employing an exchange cell with a Teflon membrane, the CO2 finally entered the receiver stream with 5 mM NaOH, thereby converting it to carbonate ions. Quantification was performed by measuring the conductivity in comparison to prepared standards. The analysis of dissolved organic carbon (DOC) in the water samples, here operationally defined as the organic fraction smaller than 0.45 μm,[51] was performed using a Shimadzu TOC analyzer TOC-V CPN (Shimadzu Corporation). The accuracy and precision of the measurements were checked with a certified Total Organic Carbon Standard 50.0 mg/L (Aqua Solutions) and had errors below 6%.

Results

Characterization of the Rock Matrix

The XRD measurements identified dolomite as the dominant mineral in samples 43 OC and 55 OC (Figures S1 and S2, Supporting Information), with semiquantitative concentrations determined at 98 and 96%, respectively. OM was heterogeneously distributed within the dolomite matrix (Figure A). Some larger OM inclusions were observed (Figure B), which contained up to 7 wt% sulfur (Table S1, Supporting Information).

Figure 1

Backscatter images of the samples 55 OC and 63 OC. (A) Transition from dolomite to a mixture of dolomite and OM. (B) OM inclusion in the dolomite matrix. (C) Pyrite framboids. (D) Powellite next to OM.

Pyrite mainly occurred as framboids in small gaps of the dolomite (Figure C). The framboids were often smaller than 5 μm, which limited the electron probe microanalyses to larger pyrite crystals (Table S2, Supporting Information). The average concentrations are listed in Table . Arsenic made up the largest fraction among the minor components of pyrite, with average concentrations of 550 and 2518 mg/kg in samples 55 OC and 43 OC, respectively (Table ), and a maximum concentration of 5130 mg/kg (Table S2, Supporting Information). While the Mo concentration was below the detection limit of 16 mg/kg in almost all pyrite measurements, a maximum concentration of 110 mg/kg was measured in sample 43 OC (Table S2, Supporting Information). Powellite (CaMoO4) was identified by optical microscopy combined with EDX and occurred occasionally between dolomite crystals and sometimes next to OM (Figure D). The presence of powellite was corroborated by its unequivocal identification by XRD in similar samples from the Avon Park Formation.[52] Electron probe microanalyses indicated, among others, As, Fe, and S as minor components (Table ) with maximum concentrations of 17780, 8890 and 37,370 mg/kg, respectively (Table S3, Supporting Information).

Table 2

Average Concentrations or Single Measurements for Pyrite and Powellite in the Samples 55 OC and 43 OCa

pyrite	Fe [wt %]	S [wt %]	As [mg/kg]	Mo [mg/kg]	Mn [mg/kg]	Co [mg/kg]	Ni [mg/kg]	Cu [mg/kg]	Zn [mg/kg]	Ag [mg/kg]	Sb [mg/kg]	total [wt %]
55 OC (n = 1)	45	52	550	<dl	<dl	<dl	<dl	<dl	<dl	<dl	<dl	97
43 OC (n = 4)	44	51	2518	110	410	<dl	<dl	<dl	<dl	<dl	<dl	95

The abbreviation “n” describes the number of measurements. The concentrations of Mo and Mn in pyrites of sample 43 OC correspond to the maximum concentration, because all other measurements were below the detection limit. The individual measurements, detection limits, and standard deviations are listed in the Supporting Information (Tables S2 and S3).

The concentrations measured after the sequential extractions are shown in Figure . The sums of concentrations from the extraction steps were in good agreement with those obtained by aqua regia digests, with deviations being generally below 10%, except for S and Fe (Tables S4 and S5, Supporting Information). The distribution of the analytes was similar for samples 43 OC and 55 OC (Figure ). During extraction step 1, more than 50% of the originally sediment-bound Mo and more than 40% of the As were dissolved, together with 30 to 50% of the sediment-bound sulfur. Ca, Mg, and Sr were more or less completely dissolved when extraction step 2 was repeated 5 times (Figure A). In the cases where extraction step 2 was executed only once, Ca, Mg, and Sr were still present in steps 3 and 4 (Figure B). Similarly, the dissolution of Mo, S, and As was also shifted to extraction steps 3 and 4 (Figure B). In contrast, when extraction step 2 was repeated 5 times, Mo and As were already mostly dissolved after steps 1 and 2, from both samples (Figure A). Iron was mainly extracted in steps 3 and 5, while S was extracted together with Mo and As in steps 1 and 2 and together with Fe in step 5 (Figure A).

Figure 2

Portions of Ca, Mg, Sr, Mo, As, S, and Fe in steps 1 to 5 relative to the sum of all steps during the sequential extraction with the 5 times (A) and 1 time (B) execution of extraction step 2.

The results for extraction step 1 for different combinations of redox conditions, pH values, solvents, and filter sizes are listed in Table S6 (Supporting Information). The extraction with sodium acetate at a pH of 7.0 and 7.5 resulted in 0.3% less TC in the sediment residues than the extraction at pH 8.2. Both TOC and TS concentrations showed no significant differences between the pH values. The concentrations of Mo (Figure ) differed by less than 10% compared to the conditions during the sequential extractions (oxic, NH4 acetate, pH 8.2, 0.45 μm). Similar results were observed for S with a maximum deviation of 4% (Figure ).

Figure 3

Concentrations of S and Mo under different redox conditions, pH values, solvents, and filter sizes determined in the repeated extraction step 1. The numbers above the columns indicate the relative standard deviation to the conditions during the initial sequential extractions (oxic, NH4 acetate, pH 8.2, 0.45 μm filter size).

During the first 30 min of the batch experiment, the pH increased to 8.6 for both shaking procedures and decreased after 24 h, together with DO (Table S7, Supporting Information). Once the measured pH reached 8.2, there was a slight increase in both Ca and Mg concentrations. The Mo, DOC, and S concentrations showed a similar release pattern during the first 24 h for both evaluated shaking procedures (Figure A). The highest release occurred for all of them during the first 30 min, and the concentrations of released Mo were indeed linearly correlated with the concentrations of the released DOC and S, respectively (Figure B). The results obtained for samples that were intermittently hand-shaken showed up to 36% lower Mo concentrations compared to the samples that were constantly shaken at 3 rpm (Table S7, Supporting Information).

Figure 4

(A) Concentration of dissolved OM (DOC), sulfur (S), and molybdenum (Mo) during the first 24 h of the batch experiment when intermittently hand-shaken and constantly shaken at 3 rpm. (B) Linear correlation of Mo with DOC and S in intermittently and constantly shaken samples during the first 24 h.

Discussion

Roles of Iron Sulfides, Powellite, and Organic Matter

Several previous studies suggested an association of Mo with iron sulfides.[23−26] However, Mo was not detected in step 5 of our sequential extractions, which was specifically designed to target sulfides (Figure A). The chemical composition of individual pyrite crystals indicated a maximum of 110 mg/kg Mo in sample 43 OC (Table S2, Supporting Information). Even if all the Fe extracted by the aqua regia digestion of 43 OC would have been exclusively present in pyrite, only 19 μg/kg of the total 320 mg/kg Mo in the sample were released from pyrite (Table S4, Supporting Information). Instead, the results were more consistent with those obtained by Chappaz et al.[27] for euxinic muds and shales, which also showed a low association of Mo (0 to 20%) with pyrite. Similarly, Tribovillard et al.[32] did not find a significant Mo association with the abundance of pyrite for their Mesozoic formations but showed a correlation of Mo with sulfurized OM instead. The occurrence of powellite in the Avon Park Formation (Figure D; Table ) was confirmed by XRD results provided by Koopmann and Pichler.[52] The mineral was discarded as a relevant source of dissolved Mo, because powellite was observed to dissolve in step 3 of the applied extraction procedure,[53] while no Mo was detected in that step during the sequential extractions (Figure A). The formation of powellite was assumed to be the result of intermittent oxic conditions during the diagenesis of the Avon Park Formation.[52] The release of Mo due to the oxidation of Mo-containing sulfurized OM was thought to cause the supersaturation of powellite.[52] This was also observed and modeled for the occurrence of powellite in the Suwannee Limestone and Hawthorn Group of the Upper Floridan Aquifer.[35]

The up to 7-times higher amount of S than Fe in extraction step 5 observed during the sequential extraction of sample 55 OC (Table S4, Supporting Information) and the measurement of up to 7 wt % S in the OM (Table S1, Supporting Information) pointed to the presence of sulfurized OM in the Orange County samples. Sulfurized OM probably formed during intermittently restricted seawater circulation[40,41] under anoxic to euxinic conditions with an increased supply of OM and reduced sulfur.[52,54−56] The release of 30 to 50% S from samples 43 OC and 55 OC during extraction step 1 (Figure A,B), together with the observed co-dissolution of DOC and S during the batch experiment (Figure A), indicated the presence of highly soluble sulfurized OM within the aquifer matrix.

The occurrence of thin peat layers in the Avon Park Formation[45] points to the presence of humic substances in the samples. Humic substances, which typically constitute a significant fraction of OM in soils and sediments, can be operationally divided into fulvic acids, humic acids, and humins, depending on their solubility.[57−59] While sequential extractions separate phases based on their operational binding,[60] we hypothesize that our sequential extraction procedure similarly has separated the different types of humic substances. Fulvic acids, which are water-soluble at all pH values, and humic acids, which are water-insoluble at pH < 2 but soluble at high pH, were probably dissolved by extraction step 1, while humins, which are water-insoluble at all pH values, were likely dissolved in later steps.[57−59] Thus, the water solubility of sulfurized OM under alkaline conditions in step 1 of the sequential extraction and during the batch experiment probably suggests the dissolution of humic and fulvic acids, which often represent a large fraction of the OM in marine sediments.[61] Incorporation of S into humic substances has previously been described for salt marsh sediments,[62] which also occur in the Avon Park Formation,[45] estuarine sediments,[63] and marine sediments from the Jervis Inlet on the coast of British Columbia.[64] The linear correlation of DOC and S with Mo observed during the first 24 h of the batch experiment (Figure B) and the release of more than 50% Mo with S in extraction step 1 (Figure A,B) point to a joint release of Mo with sulfurized OM. Experiments by Helz et al.[65] showed that sulfurization of humic acids makes them effective Mo scavengers, and Gustafsson and Tiberg[66] described Mo binding by Suwannee River fulvic acid. Consequently, both operational groups of OM could be associated with the release of Mo in extraction step 1 and during the batch experiment.

Controlling Factors for the Release of Mo

Similar results at different filter sizes during step 1 of sequential extraction indicated that Mo is not linked to nanoparticles > 0.015 μm (Figure ). However, studies on Suwannee River fulvic and humic acid showed organic components with hydrodynamic diameters below 15 nm.[67−70] Because mechanisms linking Mo to OM are still debated, mainly due to the complexity of OM,[28,71] further investigations will be necessary to elucidate how the release and association of Mo and sulfurized OM operate at the molecular level.

Extraction step 1 with sodium acetate at different pH values led to a decrease of 0.6 to 0.9% TC in the remaining sample material compared to the initial value of the untreated sediment (Table ). According to the concentrations of Ca and Mg in the extracts (Table S6, Supporting Information), only 0.1 to 0.2% of the decrease in TC can be assigned to dolomite dissolution during the extraction. Thus, the remaining portion was likely related to the dissolution of TOC during the extraction. However, no significant differences in TOC were observed in the remaining sample material (Table ). Froelich[72] emphasized that acidification of carbonate-rich sediments can lead to a loss of organic carbon. According to the operational definition of fulvic acids and humic acids (see above), treatment with 12.5% HCl before TOC analysis could have resulted in the dissolution of fulvic acids, whereas humic acids should have remained in the sediment. Consequently, the solution of humic acids during the extraction should have led to a decrease in TOC, while the solution of fulvic acids would not have been recognizable from the TOC value. Thus, the constant value of TOC after the extraction could be an indication of the dissolution of fulvic acids with Mo during the extraction. This was also supported by the observation that the different pH values during the extractions with sodium acetate did not significantly affect the release of S and Mo (Figure ). While the dissolution of humic acids requires high pH values, fulvic acids are water-soluble at all pH values.[57−59] Consequently, a release of humic acids together with Mo in extraction step 1 would have been inhibited at pH 7.0.

Table 3

Concentrations of TC, TOC, and TS of Sample 55 OC after Extraction Step 1 with Sodium Acetate and Different pH Values Compared to Initial Values of the Untreated Sediment

	pH 8.2	pH 7.5	pH 7.0	initial
TC [%]	13.4	13.1	13.1	14.0
TOC [%]	0.9	0.8	0.8	0.9
TS [%]	0.1	0.1	0.1	0.2

The dissolution of Mo, S, and DOC during the batch experiment happened rapidly, with the highest increase during the first 30 min (Figure A). While no oxygen depletion was observed during the first 5 h for both the intermittently and constantly shaken samples, except for the hand-shaken sample after 30 min, an increase of up to 6.1 mg/L Mo was observed (Table S7, Supporting Information). Together with a maximum deviation of 7 % for extraction step 1 under oxic and anoxic conditions (Figure ), the oxidation of sulfurized OM was unlikely to be the principal process for the release of Mo. This was consistent with observations during leaching experiments of Arthur et al.[42] that also showed no response of Mo to changing concentrations of DO. Although Pichler and Mozaffari[35] concluded the oxidation of OM as a possible process for the release of Mo, the proposed dissolution of Mo in conjunction with sulfurized OM seemed to be independent of that process for the sediments from Orange County. The different solvents also had no effect on the dissolution of Mo (Figure ), which was consistent with observations of Scheplitz et al.[47] during the sequential extraction of carbonate samples. However, constantly shaking the samples at a speed of 3 rpm during the batch experiment resulted in up to 36% higher Mo concentrations compared to the samples that were intermittently shaken by hand (Table S7, Supporting Information), suggesting that the character and increased exposure time of the sulfurized OM surface to the drinking water had an impact on the dissolution of Mo.

Role of Dolomite

Ingalls et al.[73] described the preservation of organic material in intercrystalline spaces between mineral crystals of shallow-water carbonate sediments. This shields OM from any rapid dissolution until the surrounding crystalline matrix is dissolved. The process of carbonate dissolution as a source of dissolved OM has been described by Zeller et al.[74] for a seagrass meadow in southern Florida. For our samples, the location of OM as inclusions in the dolomite matrix (Figure A,B) may have had a protective effect by partially preventing the contact between water and sulfurized OM. Destruction of the initial rock matrix by dolomite dissolution in the aquifer or by grinding the samples for the experiments probably increased the accessibility of the intercrystalline sulfurized OM with Mo.

Besides its presence in intercrystalline spaces, OM can also be incorporated into the crystal structure of carbonates.[73,75−77] The possible association of the release of sulfurized OM and Mo together with dolomite dissolution became evident when extraction step 2 was repeated 5 times, which caused up to 30% S and 50% Mo to be dissolved (Figure A). If extraction step 2 was not repeated, the dissolution of dolomite shifted to extraction steps 3 and 4 (Figure B), and the same shift was observed for Mo and S (Figure B). Thus, in Orange County, the dissolution of dolomite seemed to affect both intercrystalline and incorporated OM and probably led to the joint release of sulfurized OM with Mo.

Conceptual Model

Overall, our experimental results indicated that the joint release of Mo and sulfurized OM was likely caused by dolomite dissolution during ASR at the Orange County facility. This was further supported by the linear correlation of TOC with Mg and Ca (Figure ) that occurred together with an increase of 193 μg/L Mo during a storage phase of the ASR operations.[48] The field-observed decrease of oxygen and the pH decline from 7.7 to 7.4[48] were consistent with the data collected in our batch experiment, where a decrease in oxygen and pH occurred after 24 h (Table S7, Supporting Information).

Figure 5

Linear correlation of TOC and Mg (left) and TOC and Ca (right) between 387 and 415 days since the start of the ASR operation. Data from FDEP.[48]

The observation of pyrite framboids in sediment samples from Orange County by Arthur et al.,[42] which was confirmed in this study (Figure C), suggests that pyrite oxidation and the associated release of acidity (Figure A) most likely caused the dissolution of dolomite (Figure B). This is consistent with earlier studies[78] that found the oxidation of sulfide minerals, and the subsequently induced carbonate dissolution, impacted groundwater quality during ASR operations in karstic carbonate aquifers. In our study, the dissolution of dolomite likely resulted in an increased exposure of the injectant to intercrystalline and incorporated sulfurized OM, thus triggering the release of Mo (Figure B). Several studies assumed that sulfurized OM in the sedimentary record is well preserved under anoxic conditions.[79−81] However, the results of this study indicate that oxygen introduced due to MAR can trigger the dissolution of sulfurized OM and the release of Mo. Oxidation of OM was suggested to cause elevated Mo concentrations in groundwater in Central Florida.[1,5] While the results suggest the initial release mechanism for Mo, the mobility could be further influenced by subsequent processes, which can and will be tested in a broader data analysis via reactive transport modeling. One possible process that needs to be investigated is the adsorption of the released Mo onto iron oxides like ferrihydrite,[82] which are formed during the oxidation of pyrite.[83]

Figure 6

Conceptual model for the release of Mo in Orange County. Pyrite oxidation led to a decrease in oxygen and pH (A). The reduction in pH was followed by the dissolution of dolomite and the release of intercrystalline and incorporated sulfurized OM and Mo (B).

Implications

Carbonate aquifers provide approximately 20 to 25% of the world’s water supply,[84,85] and their often karstic nature can lead to rapid transport of dissolved contaminants over long distances.[86,87] Disturbance of naturally stable geochemical conditions can lead to the release of Mo and increase the risk for groundwater quality deterioration, as observed in Florida. Because Mo is particularly enriched in marine sediments under euxinic conditions,[20] a release from carbonates is also conceivable at other locations worldwide. Romaniello et al.,[88] for example, reported elevated Mo concentrations in organic-rich carbonate mud from the Bahamas carbonate platform. Molybdenum concentrations of up to 34.7 mg/kg were observed in bituminous limestones from North Jordan, resulting in Mo concentrations of up to 650 μg/L in groundwater.[89] The fact that Florida is such a prominent example is likely because, in contrast to other locations, Mo analysis was part of the monitoring program during ASR operation.[17] Indeed, other than in paleo proxy studies, Mo is rarely measured in sedimentary rocks, at least partially due to the current lack of regulations for its monitoring and reporting.[17,20]

Climate change and groundwater overuse can induce decreasing groundwater levels, as already observed for many aquifers worldwide.[90,91] These hydrological changes can lead to similar shifts of the redox conditions and thus cause an extensive Mo release. Thus, processes that lead to the release of Mo need to be well understood.

The results of our study indicate that not only the oxidation of OM or pyrite but also the dissolution of carbonates can trigger Mo mobilization.[5,35] This has implications for developing suitable mitigation options, such as using deoxygenation as a pre-treatment for the injectant.[16] Other factors such as maintaining the natural pH in the aquifer need to be considered. The conceptual model of Mo release developed in this study will be the basis for field-scale reactive transport modeling (i) to validate further the conceptual model and the factors affecting Mo mobility in carbonate aquifers after its release and (ii) to develop the ability to systematically explore potential mitigation strategies.[34,92]