Effect of Phospholipid on Pyrite Oxidation and Microbial Communities

May 27, 2015 - Menezes , J. C. S. S.; Silva , R. A.; Arce , I. S.; Schneider , I. A. H. Production of a poly-ferric sulphate chemical coagulant by sel...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/est

Effect of Phospholipid on Pyrite Oxidation and Microbial Communities under Simulated Acid Mine Drainage (AMD) Conditions Andro-Marc Pierre Louis,† Hui Yu,‡ Samantha L. Shumlas,† Benoit Van Aken,*,‡ Martin A. A. Schoonen,§ and Daniel R. Strongin*,† †

Temple University, Department of Chemistry, 1901 N. 13th Street, Philadelphia, Pennsylvania 19122, United States Temple University, Civil and Environmental Engineering, 1947 N. 12th Sreet, Philadelphia, Pennsylvania 19122, United States § Stony Brook University, Department of Geosciences, Stony Brook, New York 11794, United States ‡

S Supporting Information *

ABSTRACT: The effect of phospholipid on the biogeochemistry of pyrite oxidation, which leads to acid mine drainage (AMD) chemistry in the environment, was investigated. Metagenomic analyses were carried out to understand how the microbial community structure, which developed during the oxidation of pyrite-containing coal mining overburden/waste rock (OWR), was affected by the presence of adsorbed phospholipid. Using columns packed with OWR (with and without lipid adsorption), the release of sulfate (SO42−) and soluble iron (FeTot) was investigated. Exposure of lipid-free OWR to flowing pH-neutral water resulted in an acidic effluent with a pH range of 2− 4.5 over a 3-year period. The average concentration of FeTot and SO42− in the effluent was ≥20 and ≥30 mg/L, respectively. In contrast, in packed-column experiments where OWR was first treated with phospholipid, the effluent pH remained at ∼6.5 and the average concentrations of FeTot and SO42− were ≤2 and l.6 mg/L, respectively. 16S rDNA metagenomic pyrosequencing analysis of the microbial communities associated with OWR samples revealed the development of AMD-like communities dominated by acidophilic sulfide-oxidizing bacteria on untreated OWR samples, but not on refuse pretreated with phospholipid. FeS2 + 8H 2O + 14Fe3 + → 15Fe2 + + 2SO4 2 − + 16H+

1. INTRODUCTION Acid mine drainage (AMD) is a severe environmental problem impacting streams and rivers in areas where there are active and/or abandoned coal mining sites.1,2 A root cause of AMD is the exposure of mining waste (i.e., overburden, waste rock, and tailings), which typically contains 1−20% pyrite (FeS2) and lower amounts of other metal sulfides (e.g., marcasite and pyrrhotite), to air, water, and microorganisms.3,4 Both the abiotic and biotic oxidation of the metal sulfide, during and after mining activity, leads to the oxidation of the sulfur component to sulfuric acid (i.e., AMD). A composite reaction for the oxidation of pyrite by molecular oxygen under abiotic conditions can be written as FeS2 + 7/2O2 + H 2O → Fe2 + + 2SO4 2 − + 2H+

(3)

where the rate of Fe2+ oxidation (reaction 2) is accelerated in the presence of iron oxidizing bacteria.7 In essence, specific microbes can rapidly convert the ferrous iron product of reaction 3 back into ferric iron;6 a reaction that would be slow in the absence of microbial action at a pH near 4. Due to the detrimental environmental impact of AMD, there has been a significant amount of effort to find a viable solution to the problem. Current remediation schemes often include the use of alkaline reagents such as limestone (CaCO3), hydrated lime (Ca(OH)2), quick lime (CaO), and caustic soda (NaOH) to neutralize the acidity associated with AMD.8,9 This type of remediation can be costly and personnel intensive due to the continuous application of neutralizing agents in impacted areas, and often requires long-term management of sites.10 Overall, it is generally recognized that controlling the detrimental effects of AMD on the environment will require the use of multiple remediation strategies.10

(1)

Reaction 1, where dioxygen is the oxidizing agent, is relatively slow compared to the circumstance where ferric iron is the oxidizing agent. Acidophilic autotrophic bacteria commonly found in AMD environments, such as Acidithiobacillus ferrooxidans, significantly increase the rate of ferrous iron oxidation by oxygen, generating ferric iron, which in turn serves as the electron acceptor for pyrite oxidation:5−7 14Fe2 + + 7/2O2 + 14H+ → 14Fe3 + + 7H 2O © 2015 American Chemical Society

Received: Revised: Accepted: Published:

(2) 7701

November 3, 2014 May 26, 2015 May 27, 2015 May 27, 2015 DOI: 10.1021/es505374g Environ. Sci. Technol. 2015, 49, 7701−7708

Article

Environmental Science & Technology An alternate approach to the continual treatment of AMD afflicted environments is to develop chemical strategies to inhibit the oxidation and decomposition of pyrite that lead to acid generation. The realization of such a strategy would require the direct chemical modification of pyrite in the environment of interest. At least one advantage of such a strategy is that it would address the root-cause of AMD and potentially be a useful complement to remediation schemes that solely rely on treating AMD effluent resulting from pyrite oxidation. A disadvantage of such a targeted approach is that it would likely be limited to regions where pyrite-containing waste is amenable to chemical treatment. The development of such a strategy has been aided by extensive research over the past three decades that has been essential in bringing forward a molecular-level understanding of pyrite oxidation.11 Using the molecular-level picture developed in these prior studies, research in our laboratory showed that specific phospholipids can be used to bind to reactive regions of the pyrite surface to form robust hydrophobic coatings on pyrite that significantly reduce iron and sulfur oxidation under abiotic and biotic environments.12−16 Prior experiments, however, were conducted using pure pyrite as the reactive phase to simulate metal sulfide-containing minerals. In the present contribution, we extended our prior studies to investigate the effect of phospholipids on acid generation from actual coal mining pyrite-containing overburden/waste rock (OWR). A broader objective of the investigation is to test the hypothesis that the chemical modification of pyrite is a potentially viable scheme to suppress AMD chemistry for extended periods of time. To accomplish this objective, the pH, sulfate [SO42−], and iron [Fe2+/3+] of effluent evolving from test columns individually packed with OWR and phospholipid-treated OWR were compared. Two different phospholipids were studied: 1,2bis(10,12-tricosadiynoyl)-sn-glycerol-3-phosphocholine (23:2 Diyne PC) and a commercially available phosphocholine containing lipid formulation, Phospholipon 80H. In addition, metagenomic studies were conducted to characterize the microbial communities that were present on the OWR (both untreated and lipid pretreated materials) after 2−3 years of reaction. The results from these studies help to more fully evaluate the effect of phospholipids on the biogeochemistry of pyrite oxidation and its potential for suppressing AMD under simulated environmental conditions.

time that they were exposed to weathering conditions at the mining site. One of the samples used in this study was collected from a mining site in western Pennsylvania only days after its excavation. We refer to this fresh-OWR as FOWR throughout our study. The second sample was OWR that was collected after a time period that approached five years of weathering at the mining location. We refer to this aged-OWR as AGOWR. Pure pyrite was used as a reference material in some of our experiments and the sample used was purchased from Wards Scientific (#49-5884). 2.3. Characterization Techniques. 2.3.1. Brunauer− Emmet−Teller Surface Area Measurement (BET). BET was used to determine the specific surface area (SSA) of the different OWR samples (N2 was the adsorbing gas). The BET instrument was a Micromeritics, model ASAP 2020 V1.05. A mass of 100 mg of dry fine crushed pyrite, AGOWR or FOWR, was individually placed in a BET glass container for degassing and analysis. All samples were degassed (i.e., to desorb adsorbed water) at 100 °C for 600 min prior to any surface area measurements. The SSA for pyrite, AGOWR, and FOWR used in this study was measured to be 10.9 ± 0.8, 46.9 ± 0.6, and 2.93 ± 0.04 m2/g, respectively. We refer to the crushed AGOWR and FOWR that were not exposed to the column test environment as AGOWR/as-received and FOWR/as-received, respectively. 2.3.2. X-ray Fluorescence (XRF) Analysis. The elemental composition of AGOWR and FOWR samples was characterized with XRF using a nondestructive Niton XL3t XRF Analyzer Hand-held from Thermo Scientific. The analyzer was equipped with a 50 kV miniaturized X-ray tube and a scintillation detector. Detection limits for the major and trace elements were on the order of a few parts per million (ppm or mg/L). 2.3.3. X-ray Diffraction (XRD). OWR samples were characterized by XRD using a Bruker Kappa Apex II Duo single crystal X-ray diffractometer. Both FOWR and AGOWR samples were crushed into a powder, and then placed into 0.7 mm O.D. capillary tubes for data collection. Samples were analyzed with MoKα radiation (λ = 0.71073 Å) at nearly constant irradiation volumes in the 2θ range of 5−50° with a step size of 0.02° at a rate of 180 s/step at room temperature. Eva Software was used to process the XRD diffractograms. 2.4. Column Experiments. A schematic of a typical packed column used in our study is shown in the Supporting Information (SI), Figure 1S. Before adding any of the lipidpretreated or untreated OWR, the columns were loaded with precut disks prepared with geotextile Polyfiber fish filter pads and glass beads (∼10 mm diameter) at the bottom to prevent clogging and to maintain the water flow rate. The columns were 50 cm in length and 6 cm in diameter. 2.4.1. Preparation of Lipid Treated OWR. To prepare 80Htreated OWR (FOWR and AGOWR), 25 g of the OWR

2. MATERIALS AND METHODS 2.1. Materials. Research grade 23:2 Diyne PC (Structure 1) was purchased from Avanti Lipids (Alabaster, Alabama). Commercially available Phospholipon 80H was purchased from Lipoid Corporation (Ludwigshafen, Germany). According to manufacturer specifications, the Phospholipon 80H (referred to as 80H) formulation contained a minimum of 60% hydrogenated phosphatidylcholine type lipid, a maximum of 10% lysophosphatidylcholine (containing 1 fatty acid group), and fatty acids (∼85% stearic acid and ∼15% palmitic acid). To better characterize the phosphatidylcholine component, we analyzed the 80H with high-pressure liquid chromatography/ mass spectroscopy (HPLC/MS). The major fraction was associated with a m/z value of 790.6 that could be associated with a molecular formula of C44H88NO8P, consistent with the molecular structure of 1,2-dioctadecanoyl-sn-glycero-3-phosphocholine lipid (Structure 2). 2.2. Mining Samples. Two types of mining OWR were used in this study: the difference between the two being the 7702

DOI: 10.1021/es505374g Environ. Sci. Technol. 2015, 49, 7701−7708

Article

Environmental Science & Technology

Figure 1. pH (a) and [SO42−] (b) versus time data for columns that were packed with AGOWR, AGOWR/80H, FOWR, and FOWR/80H. The OWR experienced a continuous flow of water for over 3 years.

extracted from approximately 250 mg of the samples using the PowerSoil DNA Extraction kit (MoBio, Carlsbad, CA). Partial 16S rDNA genes were amplified by PCR using barcoded universal prokaryotic primers, 519F (5′-CAGCMGCCGCGGTAATWC-3′) and 1086R (5′-CTGACGRCRGCCATGC-3′), combined with Roche 454 adaptors.19 After visualization by agarose gel electrophoresis, the PCR products flanked with adaptors were purified using the Agencourt Ampure PCR Purification system (Beckman Coulter, Beverly, MA) and quantified using a Nanodrop-2000 spectrophotometer (Nano Drop Technologies, Wilmington, DE). DNA extraction and PCR amplification were conducted in triplicate for each sample. A metagenomic library was constructed by pooling equimolar concentrations of all 16S rDNA products. The library was sequenced using a 454 GS Junior pyrosequencing system (Roche Diagnostics, Indianapolis, IN). Pyrosequencing data were filtered and analyzed to determine the bacterial diversity and the distribution of major taxonomic groups using the QIIME package (http://qiime.org/). The impact of various parameters, including pH, [SO42−], and [Fe] on the microbial community profile was studied by redundancy analysis (RDA) using CANOCO 4.5 software (Biometris, Wageningen, The Netherlands) with unrestricted Monte Carlo permutation test. For data processing and analysis, the replicate sequences of each sample were pooled together. For quality control of the results, the sequences of each replicates were compared by clustering analysis using the QIIME package.

samples were individually dispersed into 250 mL of water containing ∼4 g of 80H for 30 min. These OWR samples are referred to as FOWR/80H and AGOWR/80H, respectively. Control experiments were set up in a similar way except that the FOWR and AGOWR were not pre-exposed to phospholipid. Exposure of FOWR to 23:2 Diyne PC was carried out differently than with the 80H lipid. In this case, 25 g of FOWR was placed in a 200 mL suspension containing 4 g of 23:2 Diyne PC. Unlike the exposure process used for the 80H lipid, the FOWR/23:2 Diyne PC suspension was stirred and simultaneously exposed to the UV light emitted by a 900 W Xe lamp (Schoeffel Instrument-Corp) for approximately 3 h. This sample is referred to as FOWR/23:2 Diyne. Prior studies from our laboratory suggested that the exposure of this lipid, which contained diacetylene groups in the tails, to UV light when adsorbed on pyrite resulted in the cross-linking of the lipid tails.13 Columns for these experiments were set up under continuous water conditions (pH 7) for exposure times up to 3 years. The objective in using this sample was to determine whether cross-linked lipid produced a coating that was more effective, with regard to long-term stability, in suppressing acid generation than 80H (containing no cross-linked lipid). 2.4.2. Analysis of Column Effluent. During column test experiments, water having a neutral pH was continuously flowed through the sample columns at a flow rate of ∼1.5 mL/ min. Aliquots of ∼10 to 30 mL were periodically collected in glass vials and analyzed for pH, [SO42−], and [Fe]Tot. [Fe]Tot (sum of Fe3+ and Fe2+) was measured in the column effluent using the Stookey Ferrozine method.17,18 Samples used for this analysis were first acidified with ascorbic acid to reduce Fe3+ to Fe2+ prior to the introduction of ferrozine. The absorbance of the Fe2+−ferrozine complex was monitored with UV−vis spectroscopy (Evolution 201, Thermo Scientific) at 560 nm. [Fe]Tot was determined using a standard calibration curve. [SO42−] was determined with ion chromatography (IC). Sample aliquots (∼1.5 mL) were filtered with 0.22 μm poly(ether sulfone) membrane filters and injected with a 1 mL syringe. Effluent samples that were not analyzed immediately were kept in airtight glass vials and stored in a refrigerator. 2.5. Metagenomic Analysis of the Microbial Communities. At the end of the column tests (2−3 years depending on the experiment), samples of the OWR material were individually collected under sterile conditions and centrifuged at 6000g for 30 min. Aliquots of the original, OWR/as-received, were also collected. Total DNA was then

3. RESULTS 3.1. Chemical Analysis and XRD. Major constituents of both OWR samples included Fe, S, Si, Al, and Ca (SI, Table 1S). XRD analysis of both OWR samples were performed and compared to a pure pyrite standard (SI, Figure 2S). The diffractogram for FeS2 showed diffraction lines associated with d-spacing values of 3.13, 2.71, 2.42, 2.21, 1.92, 1.63, and 1.04 Å. XRD results for the FOWR/as-received sample were similar to those of the FeS2 reference material, suggesting that pyrite was the predominant crystalline phase in FOWR. A notable difference between the two patterns, however, is the presence of weak and broader lines observed in the 2θ° range of 5−12°, indicating that FOWR likely contained mineral phases with a low degree of crystallinity in addition to pyrite. Diffraction data associated with the AGOWR/as-received exhibits less intense pyrite-related XRD reflection lines and the appearance of new reflection lines at 2θ° values 7.2 × 10−2), except for the FOWR sample, which showed inter-replicate distances between 2.1 × 10−2 and 7.3 × 10−2. SI Figure 5S presents a phylogeny-based comparison between the samples using clustering analysis (Unifrac metric, QIIME). The results show that all three replicates of each sample cluster closely together, indicating consistency of the metagenomic analyses. In addition, the original, OWR/asreceived samples (FOWR/as-received and AGOWR/asreceived), the untreated samples (FOWR and AGOWR), and the treated samples (FOWR/80H, AGOWR/80H, and FOWR/23:2 Diyne) each present a distinct bacterial profile. Among treated samples, FOWR/80H shows a distinct profile from FOWR/23:2 Diyne and AGOWR/80H. Based on the Shannon diversity indexes, the microbial diversity in the OWR/as-received samples was rather low, with indexes ranging from 3.3 to 3.4. The diversity increased significantly in all samples after incubation in the OWR/lipid and OWR packed columns, reaching indexes from 5.25 to 8.0. The diversity indexes were higher in the lipid-treated samples than in the control samples, with the exception of the lipid/UVtreated FOWR sample (i.e., FOWR/23:2 Diyne), which showed a lower diversity than the FOWR control (SI, Table 3S).

associated with aluminosilicates and/or mixtures of iron oxyhydroxide and metal sulfate phases. 3.2. FOWR and AGOWR Columns with and without Lipid. 3.2.1. Treatment of FOWR and AGOWR with 80H Lipid. Data presented in Figure 1a show the pH of the effluent associated with FOWR/80H and AGOWR/80H as a function of time. Inspection of the data shows that the pH of the effluent associated with FOWR/80H at the earliest collection times was ≥6. In contrast, the pH of the effluent associated with AGOWR/80H was near 4 at the earliest collection times, but rose rapidly to a pH near 7 after 4 weeks of water flow. After one month, the pH associated with both FOWR/80H and AGOWR/80H was essentially circumneutral over the course of the 3.5 year experiment: the average pH of the effluent was calculated to be 6.9 and 7.1, for FOWR/80H and AGOWR/ 80H, respectively (SI, Table 2S). In contrast, the effluent associated with FOWR or AGOWR, which had not been pretreated with 80H, had an average pH of 3.7 and 3.5, respectively, over the same period of time. Sulfate data shown in Figure 1b are generally consistent with the pH data. At the earliest measurement times, [SO42−] measured for the lipid-treated samples was similar to the untreated samples. As the reaction time progressed, however, there was a reduction in [SO42−] in the effluent associated with FOWR/80H and AGOWR/80H. The average [SO42−] in the effluent, averaged over the entire experimental run, was reduced by over 70% (SI, Table 2S) when the OWR was pretreated with lipid (i.e., FOWR/80H and AGOWR/80H) relative to lipidfree OWR. 3.2.2. Treatment of FOWR with 23:2 Diyne PC. Figure 2a and b exhibit data for the pH and [SO42−] associated with the effluent for FOWR/23:2 Diyne lipid (with UV treatment) and FOWR that was not pretreated with lipid. The pH of the effluent for FOWR/23:2 Diyne remained in the pH ∼6−7 range over the 3-year run. However, the average pH was 6.2 for the FOWR/23:2 Diyne system, lower than the average pH of 6.9 for FOWR/80H. 3.2.3. Fe Concentration in the Effluent of Column Experiments. The [Fe]Tot in the effluent of AGOWR and FOWR calculated as an average [Fe] value (i.e., [Fe]Avg) for the entire time of the column test was elevated as compared to the respective lipid-pretreated samples. For example, [Fe] Avg associated with FOWR and FOWR/80H is 3.17 and 0.81 mg L−1, respectively (SI, Table 2S). This result is consistent with the lower pH and [SO42−] associated with lipid-treated OWR, relative to untreated OWR. We emphasize, however, that a plot of [Fe]Tot versus time (SI, Figure 3S) over the course of a 7704

DOI: 10.1021/es505374g Environ. Sci. Technol. 2015, 49, 7701−7708

Article

Environmental Science & Technology

Figure 3. Abundance of major bacterial phyla detected in the column samples and as-received samples. The phyla containing less than 1% of the total microbial species were not represented. FOWR/as-received and AGOWR/as-received are FOWR and AGOWR samples, respectively, that were not subjected to the test column environment.

Figure 4. Abundance of AMD-related bacteria families detected in the column samples and as-received samples.

(11 to 23%).20−23 AMD-related species are here defined as species belonging to families commonly detected in AMD environments, including members of Acidimicrobiaceae, Acidithiobacillaceae, Acidobacteriaceae, Acetobacteraceae, Clostridiaceae, Gallionellaceae, Holophagaceae, Hydrogenophilaceae, Leptospirillaceae, and Syntrophobacteraceae (Figure 4). It is important to note that these families may also contain species that are not specific to AMD. Nontreated OWR contained many autotrophic iron and/or sulfur oxidizing AMD-related species (21−39%), which were almost absent in the lipidtreated OWR samples (∼0%) (Figure 4). In contrast, lipidtreated samples showed enrichment in members of the families, Caulobacteraceae, Comamonadacae, Holophagaceae, and Xanthomonadaceae, which contain for the most part heterotrophic species, suggesting that microorganisms in

The relative abundance of bacterial groups in the OWR/asreceived samples showed a predominance of Proteobacteria, Firmicutes, Actinobacteria, and Bacteriodetes (Figure 3). The relative bacterial abundance in all column samples shows that Proteobacteria was the predominant phylum, representing 43− 80% of the classified sequences. Also, Proteobacteria, Acidobacteria, Actinobacteria, Firmicutes, and Nitrospirae were the dominant phyla across all samples. Relatively higher proportions of Proteobacteria, Acidobacteria, and Firmicutes were observed in all lipid-treated OWR samples, while higher proportions of Actinobacteria and Nitrospira were observed in the untreated OWR samples (Figure 3). Examination of the abundance of the specific bacterial families previously detected in OWR samples revealed that nontreated OWR contained much higher proportions of AMDrelated species (42−45%) than the lipid-treated OWR samples 7705

DOI: 10.1021/es505374g Environ. Sci. Technol. 2015, 49, 7701−7708

Article

Environmental Science & Technology

Based on prior studies from our laboratory, phospholipids (which contain two fatty acid chains) with suitable fatty acid chain lengths suppress the oxidation of pyrite. The ability of the lipid to form a bilayer structure on the surface of the pyrite, which depends on the chain length, appears to be a prerequisite for the suppression.12 Prior spectroscopic studies suggest that the phosphate group of the phospholipid binds strongly to the Fe-component of pyrite.24 Lipids containing only a single fatty acid chain, such as stearic acid (or presumably lysophosphatidylcholine lipid), which the 80H formulation contains, would not have the ability to suppress pyrite oxidation based on our prior studies.24 XRD results (SI, Figure S2) show that FOWR/as-received has a larger fraction of pyrite than AGOWR/as-received. This result is reasonable considering that FOWR had been collected soon after excavation at the mining site and would not have undergone as much AMD generating chemistry as AGOWR. The possibility that phospholipid is suppressing AMD chemistry by binding to the pyrite component of FOWR (and AGOWR) is probably best shown by returning to further analyze the XRD diffractograms that are associated with FOWR and FOWR/80H after >3 years of reaction. In particular, reflections due to pyrite are still observable for the FOWR/80H sample whereas they are absent in the FOWR diffractogram. In short, lipid protected pyrite from decomposition reactions over the 3-year period. Perhaps not surprisingly, there were significant changes between the FOWR/as-received and FOWR that underwent reaction for 3 years in the column experiment. In general, many additional reflections appeared in the diffractogram associated with the 3-year reacted sample compared to the FOWR/as-received sample. The FOWR sample after 3 years looks very similar to the AGOWR sample (SI, Figure S2). Exposure to water, oxygen and bacteria for the 3 years likely resulted in many dissolution and reprecipitation reactions that formed various crystalline phases. We are currently conducting experiments to identify some of these new crystalline phases. We believe a second significant result is that the lipidinduced suppression of pyrite oxidation chemistry has a dramatic effect on the composition of the microbial community structure. This result implies that the intentional modification of mine tailings, for example, in a coal mining site to inhibit AMD will change the nature of the microbial makeup of the site. This contention is consistent with prior field-study research that has shown that the composition of microbial communities associated with AMD-impacted environments change after such areas underwent remediation by limestone treatment.25 We distinguish our study from this prior type of investigation in the sense that we have intentionally modified OWR by selectively binding a biomolecule (i.e., phospholipid) to the active-AMD generating phase in the OWR. Clustering analysis based on bacterial phylogeny (SI, Figure 4S) reveals that the OWR/as-received samples presents a bacterial profile very distinct from the ones of the column samples, regardless the application of AMD-mitigation treatment, which indicates that prolonged wet incubation in the columns modified the bacterial community to a greater extent than aging in the field. As expected, the untreated (control) column samples, FOWR and AGOWR, also showed a profile distinct from FOWR/80H, AGOWR/80H, and FOWR/23:2 Diyne, which likely reflects the abundance of AMD-related species in the untreated samples and/or the abundance of heterotrophic species in the treated ones (see below).

these samples could use the lipids as a carbon and energy source. 3.4. Redundancy Analysis. Redundancy analyses (RDA) were conducted to estimate the effect of critical AMD parameters, including pH, [SO42−], and [Fe], on the structure of the microbial communities in the test column samples. As expected, the ordination diagram shows an inverse correlation between [SO42−], [Fe], and pH (SI, Figure 6S). The diagram also shows that bacterial phyla containing most AMD-related species, including Acidobacteria, Actinobacteria, and Nitrospirae, were present in the lipid-free OWR samples, with low pH and high levels of SO42− and Fe. On the contrary, Proteobacteria and Firmicutes were found in lipid-treated OWR samples, with relatively low [SO42−] and [Fe], and higher pH, as summarized in the ordination diagram (SI, Figure 6S). Analysis of the results shows that microbial communities typical of AMD environments developed on the untreated OWR, whereas these AMD-related communities did not develop on the OWR that were pretreated with phospholipid.

4. DISCUSSION This section will focus primarily on two experimental observations. First, the exposure of OWR to phospholipid suppresses chemistry that results in the oxidation of metal sulfide and the production of waters with low pH and elevated concentrations of metal (i.e., iron). Prior studies from our laboratory showed that phospholipids had a similar effect on pure pyrite model samples, but the experiments were shorttermed (e.g., 1 month), while the tests presented here were conducted for up to 3 years. While FOWR and AGOWR are rather complex mixtures of soil and rock, the results do suggest that the mechanism by which the phospholipid decreases AMD chemistry is due to binding to the iron sulfide component. Also, we believe, based on our prior studies, that the suppression of metal sulfide oxidation chemistry on FOWR/80H and AGOWR/80H is due to the phospholipid component of the 80H formulation. The large reduction in the rate of SO42− released in the presence of lipid-treated OWR, as well as the higher pH of the effluent relative to the untreated OWR indicates that adsorbed 80H phospholipid significantly suppresses the oxidation of the sulfur component of OWR that leads to the production of acid. Similar to 80H, pretreatment of FOWR with 23:2 Diyne PC (exposed to UV) showed a resistance to AMD chemistry. As mentioned before, 23:2 Diyne PC contains diacetylene groups in the hydrocarbon tails and exposure of this lipid to UV light when adsorbed on pyrite results in the cross-linking of the lipid tails.13 Prior research showed that the oxidation of pyrite treated in this way was greatly suppressed (compared to untreated pyrite) and the resulting cross-linked hydrophobic coating showed resistance to both an autotrophic and heterotrophic bacterium (A. ferrooxidans and Acidiphilium acidophilum, respectively).14 The present study shows that 23:2 Diyne PC has a similar ability to suppress AMD-like chemistry on pyrite-bearing OWR. Interesting to us is that the laboratory-based biotic studies just mentioned suggested before beginning the experiments that the cross-linked 23:2 Diyne PC would show better stability than 80H. This presumption was not supported by our experiments and suggests that modifying the phospholipid layer by photochemical or chemical means is not necessary. This result bodes well for a potential scenario where phospholipid can be directly applied to OWR in the environment without any further treatment. 7706

DOI: 10.1021/es505374g Environ. Sci. Technol. 2015, 49, 7701−7708

Article

Environmental Science & Technology

iron as an electron acceptor26 and have been detected in AMDimpacted environments.26,27 We mention that the biochemical oxygen demand (BOD) by heterotrophic bacteria for the decomposition of lipid could lead to the lowering of the steady state concentration of solution oxygen and contribute to the experimentally observed decreased rate of pyrite oxidation. Based on the amount of lipid in the column and the influent O2 concentration, we estimate that if reduction of all dissolved O2 were supported by lipid oxidation, then consumption of the lipid could occur in little more than a year (see SI). The column experiment results suggest that the use of phospholipid provide long-term suppression of AMD-chemistry within a time frame of at least 3 years. While the focus of this study was to determine the effect of phospholipid on the biogeochemistry of pyrite (contained within OWR) oxidation for an extended period of time, the results do suggest that the direct application of phospholipid offers the possibility of inhibiting the root cause of AMD. Related to this inhibition is that the presence of the phospholipid shifts the composition of the microbial community away from microbes common to AMD environments that use electron donors such as ferrous iron. More heterotrophic bacteria are associated with the phospholipid coated samples, but the stability of the phospholipid appears to be excellent for the extent of our trial. It is important to mention that bacterial cells are susceptible to UV-treatment, and this may explain the difference in microbial distribution observed between UVtreated and nontreated samples (Figures 3 and 4). Future experiments are underway to further evaluate the potential of phospholipids in the context of suppressing AMD in the environment. Scaling up a potential technology, conceived from laboratory-based experiments, to its application in the field is always an important issue. In this particular study the weight ratio of phospholipid to OWR was 0.16. This ratio was only chosen to stay consistent with prior laboratory-based studies that investigated the effect of different phospholipids on the oxidation of pure pyrite.12 While prior studies did not examine the minimum amount of phospholipid that could be used to suppress pyrite oxidation, it was shown that a weight ratio of phospholipid to pyrite of ∼0.0059 was effective.12 It is also mentioned that pyrite only constitutes a fraction of typical OWR (e.g., 1−2%) so even lower weight ratios should be effective in suppressing AMD chemistry in OWR. Addressing this issue with experiment is of current interest in our laboratory.

Interestingly, FOWR/23:2 Diyne showed a bacterial profile closer to AGOWR/80H than to FOWR/80H, which may be explained by the higher sulfate content in the two formers (i.e., 70−96 mg L−1) than in the later (i.e., 16 mg L−1). Sulfate likely was the dominant electron acceptor for the microbial metabolism in aged column samples and it may have largely affected the bacterial profiles. Alternatively, this observation may indicate that the 23:2 Diyne PC/UV treatment is less effective than the 80H treatment, resulting in FOWR resembling AGOWR after the in-column incubation. With regard to our microbial-based results, bacterial species detected in all the samples were distributed into 7 distinct phyla (the phyla containing less than 1% of the total microbial species were not considered). Five out of these 7 phyla are known to contain most AMD-related species: Proteobacteria, Acidobacteria, Actinobacteria, Firmicutes, and Nitrospirae.20−23 The OWR/as-received samples were characterized by a rather low microbial diversity. AMD microbial communities typically contain a limited number of distinct species that are capable to thrive under the extreme AMD conditions (low pH, high metal content) and to use specific electron donors available in these environments (reduced iron and sulfur).20,23 More specifically, genera of acidophilic bacteria typically associated with AMD environments20,21 were detected almost exclusively in the nontreated samples (FOWR and AGOWR) exhibiting low pH; these genera included the iron- and/or sulfur-oxidizers, Acidimicrobium, Acidithiobacillus, Leptospirillum, and Sulfobacillus, and the iron-reducer, Acidiphilium. The iron-oxidizing genus, Gallionella, was also detected primarily in nontreated samples. On the other hand, the iron-oxidizing genus, Thiobacillus, was abundantly represented in both treated and nontreated samples. A relatively higher proportion of Proteobacteria (mainly beta- and alpha-Proteobacteria) was observed in all lipid-treated OWR samples, suggesting enrichment in heterotrophic species using lipids as both carbon and energy sources. On the contrary, a higher proportion of Actinobacteria and Nitrospira were observed in the control samples (lipid-free), suggesting that bacterial communities in nontreated samples were dominated by autotrophic bacteria using reduced iron and/or sulfur. Although the OWR/asreceived samples contained a low proportion of strains typically found in AMD (