- Email: [email protected]
Spatial variations in the fate and transport of metals in a mining-influenced stream, North Fork Clear Creek, Colorado
Science of the Total Environment 407 (2009) 6223–6234
Contents lists available at ScienceDirect
Science of the Total Environment j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s c i t o t e n v
Spatial variations in the fate and transport of metals in a mining-influenced stream, North Fork Clear Creek, Colorado Barbara A. Butler 1, James F. Ranville ⁎, Philippe E. Ross 2 Division of Environmental Science & Engineering, Colorado School of Mines, Golden, CO, 80401 USA
a r t i c l e
i n f o
Article history: Received 7 May 2009 Received in revised form 9 August 2009 Accepted 27 August 2009 Available online 3 October 2009 Keywords: Metal speciation Sorption Acid-mine drainage Iron oxyhydroxides Manganese oxyhydroxides Visual-MINTEQ
a b s t r a c t North Fork Clear Creek (NFCC) receives acid-mine drainage (AMD) from multiple abandoned mines in the Clear Creek Watershed. Point sources of AMD originate in the Black Hawk/Central City region of the stream. Water chemistry also is influenced by several non-point sources of AMD, and a wastewater treatment plant (WWTP). In-stream conditions immediately downstream from point-source inputs result in a visual and rapid precipitation of hydrous iron oxides (HFO). Hydrous manganese oxides (HMO) are seen to coat rocks further downstream during some seasons. Synoptic spatial sampling was used to assess the fate and transport of Cu, Fe, Mn, and Zn during different years and hydrological seasons. Visual-MINTEQ was used to compare observed and model-calculated percentage particulate Cu and Zn as influenced by sorption to both HFO and HMO and aqueous complexation with dissolved organic carbon (DOC). Over distance, Cu and Fe were transported predominantly in the particulate phase, Mn in the dissolved phase, and Zn was intermediate in its distribution, with generally about 50% being in each phase. Under higher flows, a larger fraction of the total metals was present in the dissolved phase, along with a lower total suspended sediment (TSS) concentration. This is consistent with the source of TSS being predominantly in-stream precipitation of metals, which might be kinetically limited under higher flows. Modeling results most closely represented observed percentage particulate Cu under lower flows; a strong seasonal trend was not evident for Zn. Model over-predictions of percentage particulate Cu suggest non-equilibrium with sorbent phases or that something in addition to DOC was keeping a portion of the Cu in solution; under-predictions for Zn suggest an additional sorbent. Differences between observed and modeled particulate varied significantly between sites and seasons; ranging from 1 to 54% for Cu and 1 to 34% for Zn overall. © 2009 Elsevier B.V. All rights reserved.
1. Introduction The transport of metals in mining-influenced waters (MIW) can be very complex, and may vary over both hydrologic season and space. In-stream mechanisms contributing to the fate and transport of metals include those shown in Fig. 1. Precipitated or sorbed metals can be transported downstream or aggregate and settle to the streambed. Desorption and dissolution may occur concurrently with sorption and precipitation in the water column. Metals can also be input from the watershed, as either dissolved or particulate. Scouring and settling are flow dependent, with higher flows capable of carrying larger sized particles in higher quantities and settling occurring
⁎ Corresponding author. Department of Chemistry & Geochemistry, Colorado School of Mines, 1500 Illinois Street, Golden, CO 80401 USA. Tel.: +1 303 273 3004; fax: +1 303 273 3629. E-mail addresses: [email protected] (B.A. Butler), [email protected] (J.F. Ranville). 1 Current address: U.S. EPA, National Risk Management Research Laboratory, Cincinnati, OH 45268 USA. 2 Posthumous. 0048-9697/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2009.08.040
mostly under lower flows. Synoptic spatial sampling affords a better understanding of the fate and transport processes occurring over the course of a stream reach, and may assist in remediation efforts. Geochemical modeling allows for confirmation and/or prediction of metal behavior and processes thought to be occurring in MIW. It is well known that metals form surface complexes with organic and inorganic particulates, such as particulate organic carbon (POC), hydrous iron oxides (HFO), aluminum oxyhydroxides (ALO), and hydrous manganese oxides (HMO) (Davis and Leckie, 1978; Dong et al., 2003; Jenne, 1968; Mahony et al., 1996; Turner et al., 2004). Geochemical modeling programs that include surface complexation reactions with these phases have been applied successfully for understanding metal partitioning behavior in MIW (e.g., Smith et al., 1998; Tonkin et al., 2002; Balistrieri et al., 2007). It also is well known that dissolved organic carbon (DOC) will complex with metals (Bryan et al., 2002; Shi et al., 1998; Tipping et al., 2002), maintaining them in solution. Models that include both aqueous DOC complexation and surface complexation are especially useful for understanding metals' fate and transport in these systems (e.g. Paulson and Balistrieri, 1999; Balistrieri et al., 2003). This research examined the fate and transport of dissolved (operationally defined as passing a 0.45-μm filter) and particulate
6224
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Fig. 1. Schematic representation of metal transport in a stream (Butler, 2005).
sulfate (SO4), manganese (Mn), zinc (Zn), iron (Fe), and copper (Cu) loads and water chemistry parameters in a stream influenced by acidmine drainage (AMD), specifically North Fork Clear Creek (NFCC), Colorado. Variations over distance for a given sampling date and between seasons were examined. Synoptic sampling assumes that the water being collected from a downstream site is essentially the same water as was sampled at the previously sampled upstream site and that any changes in the water quality between the two sites, collected within a short time of one another (less than 1 h), are due to in-stream processes. The purposes for the synoptic spatial studies were to 1) determine the extent to which metals are either transported or retained in the stream, 2) examine seasonal trends in spatial transport, and 3) use equilibrium geochemical speciation modeling to evaluate geochemical processes. It was hypothesized for this work that a simple geochemical speciation model would represent closely the percentages of particulate copper and zinc using the primary processes expected to be occurring, namely complexation with DOC and sorption to HFO and HMO solids. 2. Materials and methods 2.1. Study site NFCC is a high-gradient stream with discharge directly related to AMD input, snowmelt, and strong seasonal storms, with snow accumulation and melt in spring dominating the hydrology of the watershed. The watershed is mountainous with steep-walled canyons and narrow valley floors, ranging in elevation from approximately 2100 to 3380 m above mean sea level and is located in the Central City–Idaho Springs Mining District in the Colorado Mineral Belt (CMB). The geology includes Precambrian crystalline rocks interlaid with gneiss, granite, schist, and pegmatite. Tertiary intrusives are the sources of sulfide ores that contain deposits of precious metals (gold and silver) and base metals. The vein forming minerals are quartz (SiO2), pyrite (FeS2), chalcopyrite (CuFeS2), tennanite (Cu3AsS3), enargite (Cu3AsS4), sphalerite (ZnS), and galena (PbS) (Wildeman et al., 1974). The headwaters of NFCC are located approximately 48 km west of Denver, CO, it is approximately 29 km long, and it joins Clear Creek about 16 km west of Golden, CO. Extensive mining in the late 1800s (Cunningham et al., 1994) has resulted in several point sources of AMD entering the stream near the towns of Black Hawk and Central City, which are located near the middle of the watershed. Non-point sources, primarily seepage from waste and tailing piles along the stream reach and from Russell Gulch (Fig. 2) during spring runoff, also are important. The primary metals of concern in the stream include Cu, Fe, Mn, and Zn. Conditions in the stream immediately downstream from the main AMD inputs result in a visual and rapid precipitation of HFO that both travels downstream as particulate and forms loose coatings on rocks. HMO is seen to form a hard coating on rocks further downstream and aluminum oxyhydroxide (ALO) precipitation is observed below Russell Gulch during some seasons. The locations of water sample collection are indicated on Fig. 2, following the nomenclature of the Colorado Department of Public Health and Environment (CDPHE). AMD inputs are italicized; the
wastewater treatment plant (WWTP) is site NCC-SW-15A; sites NCCSW-27, NCC-SW-20, and NCC-SW-17 are the primary AMD inputs; and NCC-SW-13 and NCC-SW-7 are lesser AMD inputs. Since this study, the WWTP has been moved to approximately 8 km downstream from its original location. During the June 2005 sampling, the stream was piped below the WWTP to minimize impacts from construction activities, which included removal of a hillside during late summer into early fall of 2004. An additional site (NCC-SW-15C) was sampled on June 3, 2005, located just below the WWTP and just above where the water entered the pipe. Not all sites indicated on Fig. 2 were sampled on all dates, but specific sites sampled are specified on the graphs in the Results section. 2.2. Field methods and sample collection Synoptic water sampling was conducted at multiple sites along NFCC on four sampling dates: May 21, 2002, June 3, 2004, September 14, 2004, and June 3, 2005. Water sampling methods are discussed in Butler et al. (2008a), but also are briefly described here. All sample containers were either new or were acid-washed and triple rinsed with DI water before use. Additionally, the sampling containers were rinsed three times with stream water before the sample was obtained. Grab samples were obtained in high-density polyethylene (HDPE) bottles and samples for total organic carbon (TOC) were collected in borosilicate amber glass vials. Sub-samples were obtained for total (acid-soluble) and dissolved metals (filtered at 0.45 μm). Samples for total suspended sediment (TSS) were obtained by vigorously shaking the water sample and immediately filtering a known volume (50– 250 ml) through pre-weighed membrane filters (Butler et al., 2008b); TSS was not measured during the May 2002 sampling event. Dissolved ferrous iron (Fe2+), temperature, and pH were measured in the field. Fe2+ was measured using the Hach™ Method 8146 (DR/800 spectrometer) in 2005 and the Hach™ comparator method (model IR-18C) in 2004 (both available from www.hach.com); each method involves the measurement of color generated from complexation of 1,10 Phenanthroline with the ferrous ion. For the comparator method, a minimum of two individuals read the value independently and the average was recorded. Fe2+ was not measured in May 2002. Temperature and pH were measured in-stream with a 3-point calibrated field probe and meter. Alkalinity was measured via sulfuric acid titration using the Hach™ alkalinity test kit, Model AL-DT. All probes and containers used for water chemistry measurements were triple rinsed with DI water between samples. Discharge was provided by the CDPHE using the velocity/area method from flow measurements obtained using a Marsh-McBirney flow meter at each site (or estimated from upstream and input sites) and provided by the operator for the wastewater treatment plant (NCC-SW-15A). There may be a 10% error associated with the flow measurements using a Marsh-McBirney flow meter (Ron Abel, CDPHE, personal communication); thus, reported loads may differ from true values by ±10%. QA/QC included replication at NCC-SW-14 in May 2002; at NCCSW-28 and NCC-SW-9 in June 2004; at NCC-SW-19 and NCC-SW-3 in September 2004; and at NCC-SW-19 in June 2005, and a DI blank for each date, treated in the same manner as the site samples. QA/QC requirements were considered met if field replicates were within ±20% relative error and field blanks had concentrations of analytes of interest at least an order of magnitude below their lowest concentration in the samples. 2.3. Analytical methods Water samples for total (acid-soluble) and dissolved metals (<0.45 μm) were acidified with concentrated trace-metal grade HCl to pH < 2 and analyzed on a Perkin-Elmer Optima-3000 ICP-AES. QA/ QC for the ICP method included an internal standard (scandium), instrumental triplicates with a CV of less than ±10%, and a calibration
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
6225
Fig. 2. North Fork Clear Creek study area and sampling site locations following the Colorado Department of Public Health and Environment (CDPHE) designations [stream ID–surface water–number] (personal communication Ron Abel CDPHE). AMD inputs are shown in italic.
check standard and check standards run in 20-sample intervals to evaluate instrument drift with acceptability of less than ±10% and less than ±20% relative error from the known values for each analyte, respectively. Standards were prepared from High-Purity Standards (obtained from Fisher Scientific). Standard deviations of the total and dissolved metals were determined through three instrument replicates of each sample for the 2004 and 2005 sampling events. May 2002 data were obtained from EPA Region VIII and instrumental replication data were not available. Additionally, May 2002 sulfate data were obtained via ion chromatography (IC), while for all other dates, sulfate was obtained by multiplying the ICP sulfur concentration by three. TOC sample vials were shaken and then allowed to settle for approximately 20 min, so that any solids present would not be included in the sample to potentially clog the instrument. Preparation in this way also allowed for the assumption that measured TOC was approximately equivalent to DOC for use in modeling. After acidification to pH 3 with 6 N HCl, the samples were analyzed via a Shimadzu TOC 5000A analyzer. QA/QC for the TOC method included replicate samples, DI blanks, and standards of known concentration. Standards were required to be within ±10–20% relative error from the known value, and laboratory replicates were required to be within ±20% of one another. Filters for TSS were dried at room temperature for a minimum of 24 h and massed. The masses of the filters prior to use were subtracted to obtain the masses of suspended sediment per volume of water filtered. All QA criteria were met. Standard deviations of the total and dissolved metals were propagated through calculations for particulate metals and for field replicate means. Detection limits (DLs) for the Fe2+ methods were 0.5 mg/l and 0.05 mg/l for the comparator method and the spectrophotometer method, respectively. The DL for
alkalinity was 2 mg/l as CaCO3. DLs for Cu, Fe, Mn, and Zn were 6, 29, 7, and 13 μg/l, respectively. 2.4. Visual-MINTEQ modeling The Visual-MINTEQ (ver. 2.3.0) modeling program was obtained from Gustafson (2004); modeling was conducted as described in Butler et al. (2008a), but is summarized here. Data inputs included pH, temperature, alkalinity, Al3+, Ca2+, Cu2+, Fe2+, Fe3+, K+, Mg2+, 2+ Mn2+, Na+, Ni2+, SO2− . Complexation with DOC was 4 , and Zn modeled using the Stockholm-Humic sub-model and surface complexation was modeled using the double-layer surface complexation submodels for both HFO (ferrihydrite) and HMO (birnessite) (Dzombak and Morel, 1990; Gustafson, 2001, 2004; Tonkin et al., 2004). Particulate Fe concentrations were calculated from the difference between total acid-soluble Fe and dissolved Fe3+, and were assumed entirely HFO (Fe2O3·H2O, MW 177.7 g/mol) and converted to a mass of HFO for sorption modeling. Dissolved Fe3+ was assumed equal to the difference between the total dissolved (filterable) Fe from ICP measurements and the field Fe2+ measurements, where Fe2+ was present. Where Fe2+ was not present in the field, the ICP measured dissolved Fe was assumed Fe3+. Particulate Mn concentrations were calculated from the difference between total acid-soluble and dissolved Mn, and were assumed entirely HMO (MnO2·nH2O, 119 g/ mol) and converted to mass of HMO for sorption modeling (Tonkin et al., 2004). Measured TOC was assumed to represent DOC for modeling. Because the model uses the input DOC as being 100% active fulvic acid, the assumption that TOC is equivalent to DOC may cause over-estimates of DOC metal complexation. Sampling sites where observed percentage particulate Zn and/or Cu had a coefficient of variation (% CV) greater than 30% were not included in
6226
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
modeling. Additionally, some sites did not contain quantifiable particulate metals. 3. Results 3.1. Observations Total and dissolved loads of Cu, Fe, Mn, Zn, and SO4; discharge, pH, and TOC and TSS concentrations are shown in Fig. 3 for May 2002; Fig. 4 for June 2004; Fig. 5 for Sept. 2004; and in Fig. 6 for June 2005. Differences between the total and dissolved lines represent particulate metal. An inset table in each figure presents these data for the point and non-point sources of AMD and the WWTP. Instrument and replicate standard deviations, propagated through mathematical calculations, are shown on each of the graphs, but in many cases the error is less than the size of the symbol and not easily seen. Spring runoff occurred during the May 2002, June 2004, and June 2005 sampling events. The region experienced a severe drought in 2002 and a lesser drought in 2004; thus, data presented for June 2005 are more representative of the stream during high-flow conditions. The September 2004 sampling occurred during typical low-flow conditions. Correlation analyses were conducted to determine strong relationships (r ≥ ±0.70) between each of the metals and their forms, and between metals and water chemistry parameters. Of particular interest were relationships between dissolved and particulate Mn and Zn; between dissolved and particulate Fe and Zn; and between dissolved and particulate Cu and Fe. These relationships and their correlation coefficients are provided in Table 1, along with relationships between TOC and Cu concentrations and between TSS and particulate metals concentrations. 3.2. Visual-MINTEQ modeling Fig. 7 shows model results compared to observed results for the percentage of particulate Cu and Zn over distance for each of the sampling dates. Dissolved Cu was below detection at NCC-SW-26 and 19 in June 2004, and there was no observed particulate Cu at NCC-SW28; thus, these sites are not included on the Cu graph in Fig. 7. Similarly, Zn was not modeled at those sites for the same date due to their being none in the particulate phase. Particulate Mn was not present at NCC-SW-16 in May 2002 or at NCC-SW-19 in September 2004; thus, modeling results presented for these sites and dates include surface complexation only with HFO. Additionally, TOC was not measured at NCC-SW-14 in May, and therefore was not included in modeling input. Differences between observed and modeled results at each sampling site and for all sampling dates are shown in Table 2 for Cu and in Table 3 for Zn, where negative values indicate overprediction by the model and positive values represent underprediction. 4. Discussion 4.1. Observations SO4 and metal loads each increased on all dates downstream from the first AMD input (NCC-SW-27). Sulfate was not lost from the system over distance, consistent with other studies in similar systems (e.g., Bencala et al., 1987; 1990). Dissolved Fe, Cu, and Zn began to partition to the particulate phase at NCC-SW-26. For the 2004 sampling dates, Mn did not begin to partition significantly to the particulate phase until downstream from NCC-SW-15. During the warmer months of the year, there was often a visible HMO coating on rocks downstream from NCC-SW-12, which corresponds to the region where both dissolved and particulate Mn began to be lost from the water column. It is possible that input of microorganisms from the
WWTP (NCC-SW-15A) influenced the fate of Mn below this location through the formation of biogenic HMO (Tebo et al., 2004), but this mechanism was not confirmed. Over distance, Mn was transported predominantly in the dissolved phase, while Cu and Fe were transported predominantly in the particulate phase on all sampling dates. This is believed to be due to the rapid oxidation of Fe2+ to Fe3+in the turbulent water with subsequent precipitation of HFO and sorption and/or coprecipitation of the Cu. Unpublished studies of this stream have found the dissolved oxygen to be close to saturation, although it was not measured as part of this work. The behavior of Zn was intermediate to that of the Cu and Mn, with transport of both dissolved and particulate fractions being significant on all dates. Particulate Zn generally was correlated strongly to both particulate Fe and particulate Mn, which suggests coprecipitation and/or sorption with HFO and HMO. There were very few significant (r ≥ ± 0.70) correlations observed in the June 2005 data, which may be a result of the metals not being at equilibrium at the much higher flows. A pseudo-equilibrium test of the stream water at SW3 was conducted under low-flow conditions in February 2004 that indicated both Cu and Zn were at pseudoequilibrium (Butler, 2005). A later study of suspended and bed sediments conducted in 2006 suggested that Cu was not at equilibrium in the water column at SW4 under either high- or low-flow conditions, but that it was at equilibrium under low-flow conditions at SW3; Zn appeared to be at equilibrium at both sites, regardless of flow (unpublished data, CSM). Increases in dissolved SO4, Mn, Zn, and Cu loads at NCC-SW-15 were observed on all sampling dates, although sometimes only minimal. These loads were not balanced by the input from the WWTP, but instead are believed to originate from small seeps that have been observed in the region between NCC-SW-16 and NCC-SW-15. Other seeps have been observed along the stream, which may explain observed increases in metal loadings not attributed to known AMD sources, including those observed at NCC-SW-3 during some dates for some metals. A complete study of all seeps in the watershed has not been conducted; however, studies by others (e.g., Kimball et al., 2002) have found seeps to play a significant role in overall metal loading in similarly mining-influenced mountain streams. Dissolved Fe2+ was observed to be input from the AMD sources and gradually decreased over distance until no ferrous ion was present below NCC-SW-15. There was concurrent change of total dissolved Fe to particulate Fe; this is indicative of oxidation and subsequent precipitation of HFO. The presence of dissolved Fe3+ (difference between ICP dissolved Fe and measured Fe2+) has two possible explanations. First, it could be that the dissolved Fe in the absence of colorimetrically/spectrophotometrically measured Fe2+ is actually colloidal iron that passes through the 0.45-μm filters, thus appearing to be dissolved. This is supported further with the June 2004 and 2005 sampling events each having had lower ionic strength and higher DOC concentrations, each of which leads to increased colloidal stability (Buffle et al., 1998). Second, it is possible that the conversion of dissolved Fe3+ to particulate HFO was kinetically limited during the shorter travel time in the stream. The strong correlation between particulate Fe and Cu seen on all sampling dates, but weaker in June 2005 (see Table 1), is consistent with the known high affinity of copper for the HFO surface (e.g., Kinniburgh et al., 1976). Similarly, there generally was a strong correlation between particulate Zn and Mn and between Zn and Fe indicating sorption of the Zn with both HFO and HMO. Metal loads never decreased to levels observed above the first of the AMD inputs, indicating that remediation of metals in the stream will require removal/treatment of at least the point-source inputs. The percentages of total loadings of each metal lost to the streambed were greater during the lower flows in May 2002 and in September 2004 and smaller during the higher flows in June of both 2004 and 2005.
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
6227
Fig. 3. Cu, Fe, Mn, Zn, and SO4 loads; pH, TOC, and discharge data for each site sampled May 21, 2002. The table presents these data for the point and non-point sources of AMD and the WWTP.
6228
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Fig. 4. Cu, Fe, Mn, Zn, and SO4; pH, TOC, and discharge data for each site sampled June 3, 2004. The table presents these data for the point and non-point sources of AMD and the WWTP. Error bars represent instrument and replicate standard deviations, propagated through mathematical calculations.
The pH spanned 1 to 2 units throughout the stream on all sampling dates. The pH of the AMD sources (see inset tables in Figs. 3–6) is higher at the point of input to the stream than at the source adits
(generally 3 to 5). This may be due to interaction of the water with the pipes and culverts that convey it from the adit sources to the stream, although the exact mechanism is not certain.
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
6229
Fig. 5. Cu, Fe, Mn, Zn, and SO4 loads; pH, TOC, and discharge data for each site sampled September 14, 2004. The table presents these data for the point and non-point sources of AMD and the WWTP. Error bars represent instrument and replicate standard deviations, propagated through mathematical calculations.
6230
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Fig. 6. Cu, Fe, Mn, Zn, and SO4 loads; pH, TOC, and discharge data for each site sampled June 3, 2005. The table presents these data for the point and non-point sources of AMD and the WWTP. Error bars represent instrument and replicate standard deviations, propagated through mathematical calculations.
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234 Table 1 Correlation coefficients (r) between selected relationships. Relationship
May 2002
June 2004
Sept. 2004
June 2005
Particulate Fe/particulate Cu Particulate Fe/particulate Zn Particulate Mn/particulate Cu Particulate Mn/particulate Zn Dissolved Mn/dissolved Zn TOC/dissolved Cu TSS/particulate Cu TSS/particulate Fe TSS/particulate Mn TSS/particulate Zn
0.97 0.75 0.70 0.83 0.88 0.73 n/aa n/aa n/aa n/aa
0.94 0.85 < 0.70 0.88 0.88 0.80 < 0.70 0.87 < 0.70 < 0.70
0.91 0.84 0.86 0.92 <0.70 <0.70 0.98 0.90 0.91 0.90
<0.70 <0.70 <0.70 <0.70 <0.70 <0.70 0.80 0.76 <0.70 <0.70
a
TSS was not collected on this sampling date.
TOC remained essentially constant over the course of the stream within each sampling date, but varied between higher and lower flows from between approximately 0.6 and 3.3 mg C/l. TOC was highest throughout the stream reach in the June 2004 and June 2005 samplings, which is likely due to flushing of DOC that had accumulated in the soils during the winter months, as has been observed in several other studies in mountainous watersheds (e.g., Hood et al., 2005; Hornberger et al., 1994; Lewis & Grant, 1979; McKnight & Bencala, 1990). This phenomenon was not as prominent in the May 2002 data, which is most likely because there was minimal snowmelt input to the stream with the drought. There was a substantial input of TOC originating from the WWTP on all dates (see concentrations provided in inset tables in Figs. 3–6), but this was diluted by the stream water. The construction activities occurring during the September 2004 sampling resulted in input of a large load of all metals and TSS, as shown in Fig. 5 at NCC-SW-14, the site downstream from the activities. Interestingly, downstream from NCC-SW-14, all metals settled much more quickly from the water column than on other sampling dates. This might be due to enhanced aggregation and/or physical removal by the heavy load of larger particles from the hillside. It was expected that TSS concentrations would be higher during the higher flows in June 2004 and 2005, due to the increased water volume being able to carry a heavier load, but the opposite was observed. This was most likely due to dilution effects combined with the fact that the TSS in this stream generally comprises metal precipitates formed in-stream. Additionally, a larger proportion of the metal load was transported in the dissolved phase with the higher flows versus the lower flows. These observations might be an indication that the metals had not reached equilibrium under the higher flows due to having shorter residence times in the stream. Ionic strength (data not shown) ranged from ~0.5 to 1 mM prior to AMD inputs. Following AMD inputs, it increased to between 1 and 5 mM depending on season (May 2002: 4 mM; June 2004: 2 mM; Sept. 2004: 5 mM; and June 2005: 1 mM) and remained essentially the same throughout the stream reach. The increase following input of AMD is because the AMD contains high concentrations of calcium, magnesium, and sulfate. Ionic strength was decreased over the course of the stream during the higher-flow sampling events versus when flows were lower, which was most likely due to dilution from snowmelt, which drives the higher flows of spring-runoff. 4.2. Visual-MINTEQ modeling 4.2.1. Copper Modeled percentage particulate Cu results matched observed percentage particulate results most closely, and at the majority of sites, under lower flow conditions (May 2002 and September 2004) versus under higher-flow conditions (June 2004 and June 2005).
6231
Aside from a few of the sites over all the sampling times, differences are due to over-prediction by the model (negative values in Table 2). For the larger deviations from the observed particulate, it may be that Cu was not at equilibrium in these locations along the stream reach, or that processes in addition to DOC complexation were keeping more Cu in solution in the stream. Additionally, the assumption that all particulate Fe and Mn were available for sorption sites as ferrihydrite and birnessite (see Materials and methods) might have resulted in predicted percentages being higher than actual. X-ray diffraction and FT-IR (Butler 2005) showed a mix of iron (ferrihydrite, goethite, schwertmannite, and chlorite) and manganese (birnessite and todokorite) minerals in the loose floc material and black coatings on the rocks. Interestingly, the best comparison between the modeled and observed results occurred with the September 2004 sampling, which had the lowest flows of all sampling dates (see Figs. 3–6) and did not occur during the typical spring-runoff season. Although May 2002 flows were low, the sampling date did correspond with springrunoff. Thus, different processes might be more important depending on the season, as well as the actual rate of flow. 4.2.2. Zinc For all sampling dates except June 2004, modeled percentage particulate Zn matched observed percentage particulate Zn very well in locations upstream of the general region of NCC-SW-15 to 12. Downstream from NCC-SW-12, there were much larger deviations (>10 to 30%, observed minus modeled) between the data, with the model generally under-predicting percentage particulate (positive values in Table 3). It is probable that there was another sorbent not considered in the modeling that contributed to the control of Zn in the stream. Russell Gulch (NCC-SW-7) has been observed to contribute metal loadings to NFCC, including particulate aluminum, but it only flows during spring-runoff; thus, sorption of Zn to these solids may have been important during the higher flows of June 2004 and 2005 at sites downstream from NCC-SW-7. It also is possible that the particulate Zn observed in the stream originated from a nonin-stream process, such as being washed in from the watershed in particulate form. 5. Conclusions 1. Manganese and zinc were present predominantly in the dissolved fraction throughout the stream reach on all dates, but there was significant partitioning of the Zn to the particulate phase in the presence of both HFO and HMO. Copper and iron were predominantly in the particulate phase, with rapid precipitation of the iron to HFO and subsequent sorption/coprecipitation of the Cu. Loss of metals from the water column, presumably to the streambed as the water traveled downstream, was most obvious under lower flow conditions. 2. Comparisons between modeled and observed percentage particulate Cu appeared to be best when flows were lower; for Zn the differences were similar between seasons, with the exception of June 2005 having the best overall comparison. The average difference between observed and modeled values for Cu over all sites at higher flow was 24%, 11% at lower flow, and it ranged over all sites and seasons from 1 to 54%. For Zn, there was an overall mean difference of 12% at higher lows and 16% at lower flows, with a range of 1 to 34% over time and space. Cu was predominantly in the particulate form; thus, it is possible that under higher flow conditions it had not reached equilibrium, which is assumed true for modeling. 3. It appears that the model generally over-estimated particulate Cu, while it under-estimated particulate Zn. The cause for the overprediction of Cu may be due to an over-estimation in the amount of HFO and HMO used in modeling. The cause for under-prediction of Zn is most likely that another solid was sorbing the Zn in the stream
6232
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Fig. 7. Comparison between modeled and observed percentage particulate Cu and Zn over distance for each sampling date. Error bars represent instrument and replicate standard deviations, propagated through mathematical calculations.
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234 Table 2 Differences between observed and modeled percentages of particulate copper (% observed − % predicted). Sampling Date site May 2002
June 2004
September 2004 June 2005
31 30 28 26 19 18 16 15C 15 14 12 10 9 6 5 4 3
Not sampled Not sampled No particulate Not quantifiablea Not quantifiablea 6 − 25 Not sampled − 54 Not sampled − 21 Not sampled − 15 − 37 Not sampled Not sampled − 20
Not sampled Not sampled % CV > 30 4 3 − 11 −8 Not sampled − 16 −2 −3 Not sampled −9 −5 −5 Not sampled −8
a
No particulate 2 − 39 − 49 − 23 − 14 − 24 Not sampled − 11 − 11 − 11 − 10 −2 −1 −6 − 10 −8
Not sampled Not sampled Not quantifiablea % CV > 30 −5 − 10 − 43 −4 − 31 Not sampled − 43 Not sampled − 40 −6 Not sampled 2 − 17
6233
Acknowledgements This research was conducted while the lead author was a Ph.D. candidate at the Colorado School of Mines. This work was done in conjunction with the Center for the Study of Metals in the Environment, U.S. Environmental Protection Agency Grant #R82950001, designated as sub-award #522120. Partial support was provided under the National Science Foundation Computer Science, Engineering, and Mathematics Scholarships Program, Grant DUE9987037. This manuscript has been administratively reviewed and approved for publishing; however, because data were generated while the lead author was at CSM they were not subjected to NRMRL's quality assurance system requirements. The contents are solely the responsibility of the authors and do not necessarily represent the views of either the U.S. EPA or NSF. Citations of product, company, or trade names do not constitute endorsement by either the U.S. EPA or NSF and are provided only for the purpose of better describing information in this article. The authors thank numerous students from CSM for sampling assistance, Ron Abel from the CDPHE for providing the discharge data, and reviewers for very helpful comments.
Dissolved Cu was BDL.
References that was not considered in modeling. The degree of these underand over-estimations varied over the sites and dates sampled. Whether these under- and over-estimations are significant enough to preclude the use of the modeled data is dependent upon the specific use of the modeled results. 4. The assumption that a system is at equilibrium is required for using equilibrium-based geochemical speciation models. Results presented in this work suggest that this might not be a valid assumption over all time and/or space in MIW. Similarly, different processes may be important for different metals' behavior, as has been seen in this work, a temporal study of the NCC-SW-3 site (Butler et al., 2008a), Tonkin et al. (2002), and Balistrieri et al. (2003). Thus, it might be difficult for any one model to estimate accurately all metals of interest over all times and at all sampling sites. 5. Observed differences in the fraction of dissolved versus particulate metals between seasons suggest a need for seasonal evaluation of these types of streams for assessment of aquatic toxicological risk.
Table 3 Differences between observed and modeled percentages of particulate zinc (% observed − % predicted). Sampling site
Date May 2002
June 2004
September 2004
June 2005
31 30 28 26 19 18 16 15C 15 14 12 10 9 6 5 4 3
19 5 2 1 3 − 11 −5 Not sampled −3 1 21 − 20 − 15 25 30 34 30
Not sampled Not sampled No particulate No particulate No particulate 3 29 Not sampled 1 Not sampled 21 Not sampled 21 18 Not sampled Not sampled 14
Not sampled Not sampled 2 No particulate % CV > 30 9 5 Not sampled 20 22 17 Not sampled 26 23 17 Not sampled 29
Not Not 4 3 3 5 5 6 2 Not 3 Not 10 16 Not 15 21
sampled sampled
sampled sampled
sampled
Balistrieri LS, Box SE, Tonkin FW. Modeling precipitation and sorption of elements during mixing of river water and porewater in the Couer d'Alene River Basin. Environ Sci Technol 2003;37:4694–701. Balistrieri LS, Seal II RR, Piatak NM, Paul B. Assessing the concentration, speciation, and toxicity of dissolved metals during mixing of acid-mine drainage and ambient river water downstream of the Elizabeth Copper Mine, Vermont, USA. Appl Geochem 2007;22:930–52. Bencala KE, McKnight DM, Zellweger GW. Evaluation of natural tracers in an acidic and metal-rich mountain stream. Water Resour Res 1987;23:827–36. Bencala KE, McKnight DM, Zellweger GW. Characterization of transport in an acidic and metal-rich mountain stream based on a lithium tracer injection and simulations of transient storage. Water Resour Res 1990;26:989-1000. Bryan SE, Tipping E, Hamilton-Taylor J. Comparison of measured and modelled copper binding by natural organic matter in freshwaters. Comp Biochem Phys C 2002;133: 37–49. Buffle J, Wilkinson KJ, Stoll S, Filella M, Zhang J. A generalized description of aquatic colloidal interactions: the three-colloidal component approach. Environ Sci Technol 1998;32:2887–99. Butler BA. Assessing the fate and transport of metals in a high-gradient acid-mine drainage impacted mountain stream, North Fork Clear Creek, Colorado. Ph.D. Thesis T-6004, Colorado School of Mines, Golden, CO, 2005, 263 pp. Butler BA, Ranville JF, Ross PE. Observed and modeled seasonal trends in dissolved and particulate Cu, Fe, Mn, and Zn in a mining-impacted stream. Water Res 2008a;42:3135–45. Butler BA, Ranville JF, Ross PE. Direct versus indirect determination of suspended sediment associated metals in a mining-influenced watershed. Appl Geochem 2008b;23: 1218–31. Cunningham CG, Naeser CW, Marvin RF, Luedke PG, Wallace AR. Ages of selected intrusive rocks and associated ore deposits in the Colorado mineral belt. U.S. Geological Survey Bulletin, Report # 2109, 1994. Davis JA, Leckie JO. Surface ionization and complexation at the oxide/water interface: II. Surface properties of amorphous iron oxyhydroxide and adsorption of metal ions. J Colloid Interf Sci 1978;67:90-107. Dong D, Derry LA, Lion LW. Pb scavenging from a freshwater lake by Mn oxides in heterogeneous surface coating materials. Water Res 2003;37:1662–6. Dzombak DA, Morel FMM. Surface complexation modeling: hydrous ferric oxide. New York, New York: John Wiley & Sons; 1990. Gustafson JP. Modeling the acid–base properties and metal complexation of humic substances with the Stockholm humic model. J Colloid Interf Sci 2001;244:102–12. Gustafson JP. Visual-MINTEQ, Version 2.30: A windows version of MINTEQA2, version 4.0. [WWW document], accessed November 2004, URL http://www.lwr.kth.se/ english/OurSoftware/Vminteq. Hood E, Williams MW, McKnight DM. Sources of dissolved organic matter (DOM) in a Rocky Mountain stream using chemical fractionation and stable isotopes. Biogeochemistry 2005;74:231–55. Hornberger GM, Bencala KE, McKnight DM. Hydrological controls on dissolved organic carbon during snowmelt in the Snake River near Montezuma, Colorado. Biogeochemistry 1994;25:147–65. Jenne EA. Controls on Mn, Fe, Co, Ni, Cu, and Zn concentrations in soils and water: the significant role of hydrous Mn and Fe oxides. In: Gould RF, editor. Trace inorganics in water, advances in chemistry, vol. 73. Washington DC: American Chemical Society; 1968. p. 337–87. Kimball BA, Runkel RL, Walton-Day K, Bencala KE. Assessment of metal loads in watershed affected by acid mine drainage by using tracer injection and synoptic sampling: Cement Creek, Colorado, USA. Appl Geochem 2002;17:1183–207.
6234
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Kinniburgh DG, Jackson ML, Syers JK. Adsorption of alkaline earth, transition, and heavy metal cations by hydrous oxide gels of iron and aluminum. Soil Sci Soc Am J 1976;40:796–9. Lewis Jr WM, Grant MC. Relationshipsbetween flushing and yield of dissolved substances from a mountain watershed. Soil Sci 1979;128:353–63. Mahony JD, Di Toro DM, Gonzalez AM, Curto M, Dilg M, De Rosa LD, et al. Partitioning of metals to sediment organic carbon. Environ Toxicol Chem 1996;15:2187–97. McKnight DM, Bencala KE. The chemistry of iron, aluminum, and dissolved organic material in three acidic, metal-enriched, mountain streams, as controlled by watershed and in-stream processes. Water Resour Res 1990;26:3087–100. Paulson AJ, Balistrieri LS. Modeling removal of Cd, Cu, Pb, and Zn in acidic groundwater during neutralization by ambient surface waters and groundwaters. Environ Sci Technol 1999;33:3850–6. Shi B, Allen HE, Grassi MT, Ma H. Modeling copper partitioning in surface waters. Water Res 1998;32:3756–64. Smith KS, Ranville JF, Plumlee GS, Macalady DL. Predictive double-layer modeling of metal sorption in mine-drainage systems. In: Jenne EA, editor. Adsorption of metals by geomedia. San Diego: Academic Press; 1998. p. 521–47. Tebo BM, Bargar JR, Clement BG, Dick GJ, Murray KJ, Parker D, et al. Biogenic manganese oxides: properties and mechanisms of formation. Annu Rev Earth Pl Sc 2004;32: 287–328.
Tipping E, Rey-Castro C, Bryan SE, Hamilton-Taylor J. Al(III) and Fe(III) binding by humic substances in freshwaters, and implications for trace metal speciation. Geochim Cosmochim Ac 2002;66:3211–24. Tonkin JW, Balistrieri LS, Murray JW. Modeling metal removal onto natural particles formed during mixing of acid rock drainage with ambient surface water. Environ Sci Technol 2002;36:484–92. Tonkin JW, Balistrieri LS, Murray JW. Modeling sorption of divalent metal cations on hydrous manganese oxide using the diffuse double layer model. Appl Geochem 2004;19:29–53. Turner A, Millward GE, Le Roux SM. Significance of oxides and particulate organic matter in controlling trace metal partitioning in a contaminated estuary. Mar Chem 2004;88:179–92. Wildeman TR, Cain D, Ramiriz AJ. The relation between water chemistry and mineral zonation in the Central City Mining District, Colorado. Water resources problems related to mining, proc., vol. 18. Minneapolis, Mn: American Water Resources Association; 1974. p. 219–29.
Contents lists available at ScienceDirect
Science of the Total Environment j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s c i t o t e n v
Spatial variations in the fate and transport of metals in a mining-influenced stream, North Fork Clear Creek, Colorado Barbara A. Butler 1, James F. Ranville ⁎, Philippe E. Ross 2 Division of Environmental Science & Engineering, Colorado School of Mines, Golden, CO, 80401 USA
a r t i c l e
i n f o
Article history: Received 7 May 2009 Received in revised form 9 August 2009 Accepted 27 August 2009 Available online 3 October 2009 Keywords: Metal speciation Sorption Acid-mine drainage Iron oxyhydroxides Manganese oxyhydroxides Visual-MINTEQ
a b s t r a c t North Fork Clear Creek (NFCC) receives acid-mine drainage (AMD) from multiple abandoned mines in the Clear Creek Watershed. Point sources of AMD originate in the Black Hawk/Central City region of the stream. Water chemistry also is influenced by several non-point sources of AMD, and a wastewater treatment plant (WWTP). In-stream conditions immediately downstream from point-source inputs result in a visual and rapid precipitation of hydrous iron oxides (HFO). Hydrous manganese oxides (HMO) are seen to coat rocks further downstream during some seasons. Synoptic spatial sampling was used to assess the fate and transport of Cu, Fe, Mn, and Zn during different years and hydrological seasons. Visual-MINTEQ was used to compare observed and model-calculated percentage particulate Cu and Zn as influenced by sorption to both HFO and HMO and aqueous complexation with dissolved organic carbon (DOC). Over distance, Cu and Fe were transported predominantly in the particulate phase, Mn in the dissolved phase, and Zn was intermediate in its distribution, with generally about 50% being in each phase. Under higher flows, a larger fraction of the total metals was present in the dissolved phase, along with a lower total suspended sediment (TSS) concentration. This is consistent with the source of TSS being predominantly in-stream precipitation of metals, which might be kinetically limited under higher flows. Modeling results most closely represented observed percentage particulate Cu under lower flows; a strong seasonal trend was not evident for Zn. Model over-predictions of percentage particulate Cu suggest non-equilibrium with sorbent phases or that something in addition to DOC was keeping a portion of the Cu in solution; under-predictions for Zn suggest an additional sorbent. Differences between observed and modeled particulate varied significantly between sites and seasons; ranging from 1 to 54% for Cu and 1 to 34% for Zn overall. © 2009 Elsevier B.V. All rights reserved.
1. Introduction The transport of metals in mining-influenced waters (MIW) can be very complex, and may vary over both hydrologic season and space. In-stream mechanisms contributing to the fate and transport of metals include those shown in Fig. 1. Precipitated or sorbed metals can be transported downstream or aggregate and settle to the streambed. Desorption and dissolution may occur concurrently with sorption and precipitation in the water column. Metals can also be input from the watershed, as either dissolved or particulate. Scouring and settling are flow dependent, with higher flows capable of carrying larger sized particles in higher quantities and settling occurring
⁎ Corresponding author. Department of Chemistry & Geochemistry, Colorado School of Mines, 1500 Illinois Street, Golden, CO 80401 USA. Tel.: +1 303 273 3004; fax: +1 303 273 3629. E-mail addresses: [email protected] (B.A. Butler), [email protected] (J.F. Ranville). 1 Current address: U.S. EPA, National Risk Management Research Laboratory, Cincinnati, OH 45268 USA. 2 Posthumous. 0048-9697/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2009.08.040
mostly under lower flows. Synoptic spatial sampling affords a better understanding of the fate and transport processes occurring over the course of a stream reach, and may assist in remediation efforts. Geochemical modeling allows for confirmation and/or prediction of metal behavior and processes thought to be occurring in MIW. It is well known that metals form surface complexes with organic and inorganic particulates, such as particulate organic carbon (POC), hydrous iron oxides (HFO), aluminum oxyhydroxides (ALO), and hydrous manganese oxides (HMO) (Davis and Leckie, 1978; Dong et al., 2003; Jenne, 1968; Mahony et al., 1996; Turner et al., 2004). Geochemical modeling programs that include surface complexation reactions with these phases have been applied successfully for understanding metal partitioning behavior in MIW (e.g., Smith et al., 1998; Tonkin et al., 2002; Balistrieri et al., 2007). It also is well known that dissolved organic carbon (DOC) will complex with metals (Bryan et al., 2002; Shi et al., 1998; Tipping et al., 2002), maintaining them in solution. Models that include both aqueous DOC complexation and surface complexation are especially useful for understanding metals' fate and transport in these systems (e.g. Paulson and Balistrieri, 1999; Balistrieri et al., 2003). This research examined the fate and transport of dissolved (operationally defined as passing a 0.45-μm filter) and particulate
6224
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Fig. 1. Schematic representation of metal transport in a stream (Butler, 2005).
sulfate (SO4), manganese (Mn), zinc (Zn), iron (Fe), and copper (Cu) loads and water chemistry parameters in a stream influenced by acidmine drainage (AMD), specifically North Fork Clear Creek (NFCC), Colorado. Variations over distance for a given sampling date and between seasons were examined. Synoptic sampling assumes that the water being collected from a downstream site is essentially the same water as was sampled at the previously sampled upstream site and that any changes in the water quality between the two sites, collected within a short time of one another (less than 1 h), are due to in-stream processes. The purposes for the synoptic spatial studies were to 1) determine the extent to which metals are either transported or retained in the stream, 2) examine seasonal trends in spatial transport, and 3) use equilibrium geochemical speciation modeling to evaluate geochemical processes. It was hypothesized for this work that a simple geochemical speciation model would represent closely the percentages of particulate copper and zinc using the primary processes expected to be occurring, namely complexation with DOC and sorption to HFO and HMO solids. 2. Materials and methods 2.1. Study site NFCC is a high-gradient stream with discharge directly related to AMD input, snowmelt, and strong seasonal storms, with snow accumulation and melt in spring dominating the hydrology of the watershed. The watershed is mountainous with steep-walled canyons and narrow valley floors, ranging in elevation from approximately 2100 to 3380 m above mean sea level and is located in the Central City–Idaho Springs Mining District in the Colorado Mineral Belt (CMB). The geology includes Precambrian crystalline rocks interlaid with gneiss, granite, schist, and pegmatite. Tertiary intrusives are the sources of sulfide ores that contain deposits of precious metals (gold and silver) and base metals. The vein forming minerals are quartz (SiO2), pyrite (FeS2), chalcopyrite (CuFeS2), tennanite (Cu3AsS3), enargite (Cu3AsS4), sphalerite (ZnS), and galena (PbS) (Wildeman et al., 1974). The headwaters of NFCC are located approximately 48 km west of Denver, CO, it is approximately 29 km long, and it joins Clear Creek about 16 km west of Golden, CO. Extensive mining in the late 1800s (Cunningham et al., 1994) has resulted in several point sources of AMD entering the stream near the towns of Black Hawk and Central City, which are located near the middle of the watershed. Non-point sources, primarily seepage from waste and tailing piles along the stream reach and from Russell Gulch (Fig. 2) during spring runoff, also are important. The primary metals of concern in the stream include Cu, Fe, Mn, and Zn. Conditions in the stream immediately downstream from the main AMD inputs result in a visual and rapid precipitation of HFO that both travels downstream as particulate and forms loose coatings on rocks. HMO is seen to form a hard coating on rocks further downstream and aluminum oxyhydroxide (ALO) precipitation is observed below Russell Gulch during some seasons. The locations of water sample collection are indicated on Fig. 2, following the nomenclature of the Colorado Department of Public Health and Environment (CDPHE). AMD inputs are italicized; the
wastewater treatment plant (WWTP) is site NCC-SW-15A; sites NCCSW-27, NCC-SW-20, and NCC-SW-17 are the primary AMD inputs; and NCC-SW-13 and NCC-SW-7 are lesser AMD inputs. Since this study, the WWTP has been moved to approximately 8 km downstream from its original location. During the June 2005 sampling, the stream was piped below the WWTP to minimize impacts from construction activities, which included removal of a hillside during late summer into early fall of 2004. An additional site (NCC-SW-15C) was sampled on June 3, 2005, located just below the WWTP and just above where the water entered the pipe. Not all sites indicated on Fig. 2 were sampled on all dates, but specific sites sampled are specified on the graphs in the Results section. 2.2. Field methods and sample collection Synoptic water sampling was conducted at multiple sites along NFCC on four sampling dates: May 21, 2002, June 3, 2004, September 14, 2004, and June 3, 2005. Water sampling methods are discussed in Butler et al. (2008a), but also are briefly described here. All sample containers were either new or were acid-washed and triple rinsed with DI water before use. Additionally, the sampling containers were rinsed three times with stream water before the sample was obtained. Grab samples were obtained in high-density polyethylene (HDPE) bottles and samples for total organic carbon (TOC) were collected in borosilicate amber glass vials. Sub-samples were obtained for total (acid-soluble) and dissolved metals (filtered at 0.45 μm). Samples for total suspended sediment (TSS) were obtained by vigorously shaking the water sample and immediately filtering a known volume (50– 250 ml) through pre-weighed membrane filters (Butler et al., 2008b); TSS was not measured during the May 2002 sampling event. Dissolved ferrous iron (Fe2+), temperature, and pH were measured in the field. Fe2+ was measured using the Hach™ Method 8146 (DR/800 spectrometer) in 2005 and the Hach™ comparator method (model IR-18C) in 2004 (both available from www.hach.com); each method involves the measurement of color generated from complexation of 1,10 Phenanthroline with the ferrous ion. For the comparator method, a minimum of two individuals read the value independently and the average was recorded. Fe2+ was not measured in May 2002. Temperature and pH were measured in-stream with a 3-point calibrated field probe and meter. Alkalinity was measured via sulfuric acid titration using the Hach™ alkalinity test kit, Model AL-DT. All probes and containers used for water chemistry measurements were triple rinsed with DI water between samples. Discharge was provided by the CDPHE using the velocity/area method from flow measurements obtained using a Marsh-McBirney flow meter at each site (or estimated from upstream and input sites) and provided by the operator for the wastewater treatment plant (NCC-SW-15A). There may be a 10% error associated with the flow measurements using a Marsh-McBirney flow meter (Ron Abel, CDPHE, personal communication); thus, reported loads may differ from true values by ±10%. QA/QC included replication at NCC-SW-14 in May 2002; at NCCSW-28 and NCC-SW-9 in June 2004; at NCC-SW-19 and NCC-SW-3 in September 2004; and at NCC-SW-19 in June 2005, and a DI blank for each date, treated in the same manner as the site samples. QA/QC requirements were considered met if field replicates were within ±20% relative error and field blanks had concentrations of analytes of interest at least an order of magnitude below their lowest concentration in the samples. 2.3. Analytical methods Water samples for total (acid-soluble) and dissolved metals (<0.45 μm) were acidified with concentrated trace-metal grade HCl to pH < 2 and analyzed on a Perkin-Elmer Optima-3000 ICP-AES. QA/ QC for the ICP method included an internal standard (scandium), instrumental triplicates with a CV of less than ±10%, and a calibration
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
6225
Fig. 2. North Fork Clear Creek study area and sampling site locations following the Colorado Department of Public Health and Environment (CDPHE) designations [stream ID–surface water–number] (personal communication Ron Abel CDPHE). AMD inputs are shown in italic.
check standard and check standards run in 20-sample intervals to evaluate instrument drift with acceptability of less than ±10% and less than ±20% relative error from the known values for each analyte, respectively. Standards were prepared from High-Purity Standards (obtained from Fisher Scientific). Standard deviations of the total and dissolved metals were determined through three instrument replicates of each sample for the 2004 and 2005 sampling events. May 2002 data were obtained from EPA Region VIII and instrumental replication data were not available. Additionally, May 2002 sulfate data were obtained via ion chromatography (IC), while for all other dates, sulfate was obtained by multiplying the ICP sulfur concentration by three. TOC sample vials were shaken and then allowed to settle for approximately 20 min, so that any solids present would not be included in the sample to potentially clog the instrument. Preparation in this way also allowed for the assumption that measured TOC was approximately equivalent to DOC for use in modeling. After acidification to pH 3 with 6 N HCl, the samples were analyzed via a Shimadzu TOC 5000A analyzer. QA/QC for the TOC method included replicate samples, DI blanks, and standards of known concentration. Standards were required to be within ±10–20% relative error from the known value, and laboratory replicates were required to be within ±20% of one another. Filters for TSS were dried at room temperature for a minimum of 24 h and massed. The masses of the filters prior to use were subtracted to obtain the masses of suspended sediment per volume of water filtered. All QA criteria were met. Standard deviations of the total and dissolved metals were propagated through calculations for particulate metals and for field replicate means. Detection limits (DLs) for the Fe2+ methods were 0.5 mg/l and 0.05 mg/l for the comparator method and the spectrophotometer method, respectively. The DL for
alkalinity was 2 mg/l as CaCO3. DLs for Cu, Fe, Mn, and Zn were 6, 29, 7, and 13 μg/l, respectively. 2.4. Visual-MINTEQ modeling The Visual-MINTEQ (ver. 2.3.0) modeling program was obtained from Gustafson (2004); modeling was conducted as described in Butler et al. (2008a), but is summarized here. Data inputs included pH, temperature, alkalinity, Al3+, Ca2+, Cu2+, Fe2+, Fe3+, K+, Mg2+, 2+ Mn2+, Na+, Ni2+, SO2− . Complexation with DOC was 4 , and Zn modeled using the Stockholm-Humic sub-model and surface complexation was modeled using the double-layer surface complexation submodels for both HFO (ferrihydrite) and HMO (birnessite) (Dzombak and Morel, 1990; Gustafson, 2001, 2004; Tonkin et al., 2004). Particulate Fe concentrations were calculated from the difference between total acid-soluble Fe and dissolved Fe3+, and were assumed entirely HFO (Fe2O3·H2O, MW 177.7 g/mol) and converted to a mass of HFO for sorption modeling. Dissolved Fe3+ was assumed equal to the difference between the total dissolved (filterable) Fe from ICP measurements and the field Fe2+ measurements, where Fe2+ was present. Where Fe2+ was not present in the field, the ICP measured dissolved Fe was assumed Fe3+. Particulate Mn concentrations were calculated from the difference between total acid-soluble and dissolved Mn, and were assumed entirely HMO (MnO2·nH2O, 119 g/ mol) and converted to mass of HMO for sorption modeling (Tonkin et al., 2004). Measured TOC was assumed to represent DOC for modeling. Because the model uses the input DOC as being 100% active fulvic acid, the assumption that TOC is equivalent to DOC may cause over-estimates of DOC metal complexation. Sampling sites where observed percentage particulate Zn and/or Cu had a coefficient of variation (% CV) greater than 30% were not included in
6226
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
modeling. Additionally, some sites did not contain quantifiable particulate metals. 3. Results 3.1. Observations Total and dissolved loads of Cu, Fe, Mn, Zn, and SO4; discharge, pH, and TOC and TSS concentrations are shown in Fig. 3 for May 2002; Fig. 4 for June 2004; Fig. 5 for Sept. 2004; and in Fig. 6 for June 2005. Differences between the total and dissolved lines represent particulate metal. An inset table in each figure presents these data for the point and non-point sources of AMD and the WWTP. Instrument and replicate standard deviations, propagated through mathematical calculations, are shown on each of the graphs, but in many cases the error is less than the size of the symbol and not easily seen. Spring runoff occurred during the May 2002, June 2004, and June 2005 sampling events. The region experienced a severe drought in 2002 and a lesser drought in 2004; thus, data presented for June 2005 are more representative of the stream during high-flow conditions. The September 2004 sampling occurred during typical low-flow conditions. Correlation analyses were conducted to determine strong relationships (r ≥ ±0.70) between each of the metals and their forms, and between metals and water chemistry parameters. Of particular interest were relationships between dissolved and particulate Mn and Zn; between dissolved and particulate Fe and Zn; and between dissolved and particulate Cu and Fe. These relationships and their correlation coefficients are provided in Table 1, along with relationships between TOC and Cu concentrations and between TSS and particulate metals concentrations. 3.2. Visual-MINTEQ modeling Fig. 7 shows model results compared to observed results for the percentage of particulate Cu and Zn over distance for each of the sampling dates. Dissolved Cu was below detection at NCC-SW-26 and 19 in June 2004, and there was no observed particulate Cu at NCC-SW28; thus, these sites are not included on the Cu graph in Fig. 7. Similarly, Zn was not modeled at those sites for the same date due to their being none in the particulate phase. Particulate Mn was not present at NCC-SW-16 in May 2002 or at NCC-SW-19 in September 2004; thus, modeling results presented for these sites and dates include surface complexation only with HFO. Additionally, TOC was not measured at NCC-SW-14 in May, and therefore was not included in modeling input. Differences between observed and modeled results at each sampling site and for all sampling dates are shown in Table 2 for Cu and in Table 3 for Zn, where negative values indicate overprediction by the model and positive values represent underprediction. 4. Discussion 4.1. Observations SO4 and metal loads each increased on all dates downstream from the first AMD input (NCC-SW-27). Sulfate was not lost from the system over distance, consistent with other studies in similar systems (e.g., Bencala et al., 1987; 1990). Dissolved Fe, Cu, and Zn began to partition to the particulate phase at NCC-SW-26. For the 2004 sampling dates, Mn did not begin to partition significantly to the particulate phase until downstream from NCC-SW-15. During the warmer months of the year, there was often a visible HMO coating on rocks downstream from NCC-SW-12, which corresponds to the region where both dissolved and particulate Mn began to be lost from the water column. It is possible that input of microorganisms from the
WWTP (NCC-SW-15A) influenced the fate of Mn below this location through the formation of biogenic HMO (Tebo et al., 2004), but this mechanism was not confirmed. Over distance, Mn was transported predominantly in the dissolved phase, while Cu and Fe were transported predominantly in the particulate phase on all sampling dates. This is believed to be due to the rapid oxidation of Fe2+ to Fe3+in the turbulent water with subsequent precipitation of HFO and sorption and/or coprecipitation of the Cu. Unpublished studies of this stream have found the dissolved oxygen to be close to saturation, although it was not measured as part of this work. The behavior of Zn was intermediate to that of the Cu and Mn, with transport of both dissolved and particulate fractions being significant on all dates. Particulate Zn generally was correlated strongly to both particulate Fe and particulate Mn, which suggests coprecipitation and/or sorption with HFO and HMO. There were very few significant (r ≥ ± 0.70) correlations observed in the June 2005 data, which may be a result of the metals not being at equilibrium at the much higher flows. A pseudo-equilibrium test of the stream water at SW3 was conducted under low-flow conditions in February 2004 that indicated both Cu and Zn were at pseudoequilibrium (Butler, 2005). A later study of suspended and bed sediments conducted in 2006 suggested that Cu was not at equilibrium in the water column at SW4 under either high- or low-flow conditions, but that it was at equilibrium under low-flow conditions at SW3; Zn appeared to be at equilibrium at both sites, regardless of flow (unpublished data, CSM). Increases in dissolved SO4, Mn, Zn, and Cu loads at NCC-SW-15 were observed on all sampling dates, although sometimes only minimal. These loads were not balanced by the input from the WWTP, but instead are believed to originate from small seeps that have been observed in the region between NCC-SW-16 and NCC-SW-15. Other seeps have been observed along the stream, which may explain observed increases in metal loadings not attributed to known AMD sources, including those observed at NCC-SW-3 during some dates for some metals. A complete study of all seeps in the watershed has not been conducted; however, studies by others (e.g., Kimball et al., 2002) have found seeps to play a significant role in overall metal loading in similarly mining-influenced mountain streams. Dissolved Fe2+ was observed to be input from the AMD sources and gradually decreased over distance until no ferrous ion was present below NCC-SW-15. There was concurrent change of total dissolved Fe to particulate Fe; this is indicative of oxidation and subsequent precipitation of HFO. The presence of dissolved Fe3+ (difference between ICP dissolved Fe and measured Fe2+) has two possible explanations. First, it could be that the dissolved Fe in the absence of colorimetrically/spectrophotometrically measured Fe2+ is actually colloidal iron that passes through the 0.45-μm filters, thus appearing to be dissolved. This is supported further with the June 2004 and 2005 sampling events each having had lower ionic strength and higher DOC concentrations, each of which leads to increased colloidal stability (Buffle et al., 1998). Second, it is possible that the conversion of dissolved Fe3+ to particulate HFO was kinetically limited during the shorter travel time in the stream. The strong correlation between particulate Fe and Cu seen on all sampling dates, but weaker in June 2005 (see Table 1), is consistent with the known high affinity of copper for the HFO surface (e.g., Kinniburgh et al., 1976). Similarly, there generally was a strong correlation between particulate Zn and Mn and between Zn and Fe indicating sorption of the Zn with both HFO and HMO. Metal loads never decreased to levels observed above the first of the AMD inputs, indicating that remediation of metals in the stream will require removal/treatment of at least the point-source inputs. The percentages of total loadings of each metal lost to the streambed were greater during the lower flows in May 2002 and in September 2004 and smaller during the higher flows in June of both 2004 and 2005.
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
6227
Fig. 3. Cu, Fe, Mn, Zn, and SO4 loads; pH, TOC, and discharge data for each site sampled May 21, 2002. The table presents these data for the point and non-point sources of AMD and the WWTP.
6228
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Fig. 4. Cu, Fe, Mn, Zn, and SO4; pH, TOC, and discharge data for each site sampled June 3, 2004. The table presents these data for the point and non-point sources of AMD and the WWTP. Error bars represent instrument and replicate standard deviations, propagated through mathematical calculations.
The pH spanned 1 to 2 units throughout the stream on all sampling dates. The pH of the AMD sources (see inset tables in Figs. 3–6) is higher at the point of input to the stream than at the source adits
(generally 3 to 5). This may be due to interaction of the water with the pipes and culverts that convey it from the adit sources to the stream, although the exact mechanism is not certain.
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
6229
Fig. 5. Cu, Fe, Mn, Zn, and SO4 loads; pH, TOC, and discharge data for each site sampled September 14, 2004. The table presents these data for the point and non-point sources of AMD and the WWTP. Error bars represent instrument and replicate standard deviations, propagated through mathematical calculations.
6230
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Fig. 6. Cu, Fe, Mn, Zn, and SO4 loads; pH, TOC, and discharge data for each site sampled June 3, 2005. The table presents these data for the point and non-point sources of AMD and the WWTP. Error bars represent instrument and replicate standard deviations, propagated through mathematical calculations.
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234 Table 1 Correlation coefficients (r) between selected relationships. Relationship
May 2002
June 2004
Sept. 2004
June 2005
Particulate Fe/particulate Cu Particulate Fe/particulate Zn Particulate Mn/particulate Cu Particulate Mn/particulate Zn Dissolved Mn/dissolved Zn TOC/dissolved Cu TSS/particulate Cu TSS/particulate Fe TSS/particulate Mn TSS/particulate Zn
0.97 0.75 0.70 0.83 0.88 0.73 n/aa n/aa n/aa n/aa
0.94 0.85 < 0.70 0.88 0.88 0.80 < 0.70 0.87 < 0.70 < 0.70
0.91 0.84 0.86 0.92 <0.70 <0.70 0.98 0.90 0.91 0.90
<0.70 <0.70 <0.70 <0.70 <0.70 <0.70 0.80 0.76 <0.70 <0.70
a
TSS was not collected on this sampling date.
TOC remained essentially constant over the course of the stream within each sampling date, but varied between higher and lower flows from between approximately 0.6 and 3.3 mg C/l. TOC was highest throughout the stream reach in the June 2004 and June 2005 samplings, which is likely due to flushing of DOC that had accumulated in the soils during the winter months, as has been observed in several other studies in mountainous watersheds (e.g., Hood et al., 2005; Hornberger et al., 1994; Lewis & Grant, 1979; McKnight & Bencala, 1990). This phenomenon was not as prominent in the May 2002 data, which is most likely because there was minimal snowmelt input to the stream with the drought. There was a substantial input of TOC originating from the WWTP on all dates (see concentrations provided in inset tables in Figs. 3–6), but this was diluted by the stream water. The construction activities occurring during the September 2004 sampling resulted in input of a large load of all metals and TSS, as shown in Fig. 5 at NCC-SW-14, the site downstream from the activities. Interestingly, downstream from NCC-SW-14, all metals settled much more quickly from the water column than on other sampling dates. This might be due to enhanced aggregation and/or physical removal by the heavy load of larger particles from the hillside. It was expected that TSS concentrations would be higher during the higher flows in June 2004 and 2005, due to the increased water volume being able to carry a heavier load, but the opposite was observed. This was most likely due to dilution effects combined with the fact that the TSS in this stream generally comprises metal precipitates formed in-stream. Additionally, a larger proportion of the metal load was transported in the dissolved phase with the higher flows versus the lower flows. These observations might be an indication that the metals had not reached equilibrium under the higher flows due to having shorter residence times in the stream. Ionic strength (data not shown) ranged from ~0.5 to 1 mM prior to AMD inputs. Following AMD inputs, it increased to between 1 and 5 mM depending on season (May 2002: 4 mM; June 2004: 2 mM; Sept. 2004: 5 mM; and June 2005: 1 mM) and remained essentially the same throughout the stream reach. The increase following input of AMD is because the AMD contains high concentrations of calcium, magnesium, and sulfate. Ionic strength was decreased over the course of the stream during the higher-flow sampling events versus when flows were lower, which was most likely due to dilution from snowmelt, which drives the higher flows of spring-runoff. 4.2. Visual-MINTEQ modeling 4.2.1. Copper Modeled percentage particulate Cu results matched observed percentage particulate results most closely, and at the majority of sites, under lower flow conditions (May 2002 and September 2004) versus under higher-flow conditions (June 2004 and June 2005).
6231
Aside from a few of the sites over all the sampling times, differences are due to over-prediction by the model (negative values in Table 2). For the larger deviations from the observed particulate, it may be that Cu was not at equilibrium in these locations along the stream reach, or that processes in addition to DOC complexation were keeping more Cu in solution in the stream. Additionally, the assumption that all particulate Fe and Mn were available for sorption sites as ferrihydrite and birnessite (see Materials and methods) might have resulted in predicted percentages being higher than actual. X-ray diffraction and FT-IR (Butler 2005) showed a mix of iron (ferrihydrite, goethite, schwertmannite, and chlorite) and manganese (birnessite and todokorite) minerals in the loose floc material and black coatings on the rocks. Interestingly, the best comparison between the modeled and observed results occurred with the September 2004 sampling, which had the lowest flows of all sampling dates (see Figs. 3–6) and did not occur during the typical spring-runoff season. Although May 2002 flows were low, the sampling date did correspond with springrunoff. Thus, different processes might be more important depending on the season, as well as the actual rate of flow. 4.2.2. Zinc For all sampling dates except June 2004, modeled percentage particulate Zn matched observed percentage particulate Zn very well in locations upstream of the general region of NCC-SW-15 to 12. Downstream from NCC-SW-12, there were much larger deviations (>10 to 30%, observed minus modeled) between the data, with the model generally under-predicting percentage particulate (positive values in Table 3). It is probable that there was another sorbent not considered in the modeling that contributed to the control of Zn in the stream. Russell Gulch (NCC-SW-7) has been observed to contribute metal loadings to NFCC, including particulate aluminum, but it only flows during spring-runoff; thus, sorption of Zn to these solids may have been important during the higher flows of June 2004 and 2005 at sites downstream from NCC-SW-7. It also is possible that the particulate Zn observed in the stream originated from a nonin-stream process, such as being washed in from the watershed in particulate form. 5. Conclusions 1. Manganese and zinc were present predominantly in the dissolved fraction throughout the stream reach on all dates, but there was significant partitioning of the Zn to the particulate phase in the presence of both HFO and HMO. Copper and iron were predominantly in the particulate phase, with rapid precipitation of the iron to HFO and subsequent sorption/coprecipitation of the Cu. Loss of metals from the water column, presumably to the streambed as the water traveled downstream, was most obvious under lower flow conditions. 2. Comparisons between modeled and observed percentage particulate Cu appeared to be best when flows were lower; for Zn the differences were similar between seasons, with the exception of June 2005 having the best overall comparison. The average difference between observed and modeled values for Cu over all sites at higher flow was 24%, 11% at lower flow, and it ranged over all sites and seasons from 1 to 54%. For Zn, there was an overall mean difference of 12% at higher lows and 16% at lower flows, with a range of 1 to 34% over time and space. Cu was predominantly in the particulate form; thus, it is possible that under higher flow conditions it had not reached equilibrium, which is assumed true for modeling. 3. It appears that the model generally over-estimated particulate Cu, while it under-estimated particulate Zn. The cause for the overprediction of Cu may be due to an over-estimation in the amount of HFO and HMO used in modeling. The cause for under-prediction of Zn is most likely that another solid was sorbing the Zn in the stream
6232
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Fig. 7. Comparison between modeled and observed percentage particulate Cu and Zn over distance for each sampling date. Error bars represent instrument and replicate standard deviations, propagated through mathematical calculations.
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234 Table 2 Differences between observed and modeled percentages of particulate copper (% observed − % predicted). Sampling Date site May 2002
June 2004
September 2004 June 2005
31 30 28 26 19 18 16 15C 15 14 12 10 9 6 5 4 3
Not sampled Not sampled No particulate Not quantifiablea Not quantifiablea 6 − 25 Not sampled − 54 Not sampled − 21 Not sampled − 15 − 37 Not sampled Not sampled − 20
Not sampled Not sampled % CV > 30 4 3 − 11 −8 Not sampled − 16 −2 −3 Not sampled −9 −5 −5 Not sampled −8
a
No particulate 2 − 39 − 49 − 23 − 14 − 24 Not sampled − 11 − 11 − 11 − 10 −2 −1 −6 − 10 −8
Not sampled Not sampled Not quantifiablea % CV > 30 −5 − 10 − 43 −4 − 31 Not sampled − 43 Not sampled − 40 −6 Not sampled 2 − 17
6233
Acknowledgements This research was conducted while the lead author was a Ph.D. candidate at the Colorado School of Mines. This work was done in conjunction with the Center for the Study of Metals in the Environment, U.S. Environmental Protection Agency Grant #R82950001, designated as sub-award #522120. Partial support was provided under the National Science Foundation Computer Science, Engineering, and Mathematics Scholarships Program, Grant DUE9987037. This manuscript has been administratively reviewed and approved for publishing; however, because data were generated while the lead author was at CSM they were not subjected to NRMRL's quality assurance system requirements. The contents are solely the responsibility of the authors and do not necessarily represent the views of either the U.S. EPA or NSF. Citations of product, company, or trade names do not constitute endorsement by either the U.S. EPA or NSF and are provided only for the purpose of better describing information in this article. The authors thank numerous students from CSM for sampling assistance, Ron Abel from the CDPHE for providing the discharge data, and reviewers for very helpful comments.
Dissolved Cu was BDL.
References that was not considered in modeling. The degree of these underand over-estimations varied over the sites and dates sampled. Whether these under- and over-estimations are significant enough to preclude the use of the modeled data is dependent upon the specific use of the modeled results. 4. The assumption that a system is at equilibrium is required for using equilibrium-based geochemical speciation models. Results presented in this work suggest that this might not be a valid assumption over all time and/or space in MIW. Similarly, different processes may be important for different metals' behavior, as has been seen in this work, a temporal study of the NCC-SW-3 site (Butler et al., 2008a), Tonkin et al. (2002), and Balistrieri et al. (2003). Thus, it might be difficult for any one model to estimate accurately all metals of interest over all times and at all sampling sites. 5. Observed differences in the fraction of dissolved versus particulate metals between seasons suggest a need for seasonal evaluation of these types of streams for assessment of aquatic toxicological risk.
Table 3 Differences between observed and modeled percentages of particulate zinc (% observed − % predicted). Sampling site
Date May 2002
June 2004
September 2004
June 2005
31 30 28 26 19 18 16 15C 15 14 12 10 9 6 5 4 3
19 5 2 1 3 − 11 −5 Not sampled −3 1 21 − 20 − 15 25 30 34 30
Not sampled Not sampled No particulate No particulate No particulate 3 29 Not sampled 1 Not sampled 21 Not sampled 21 18 Not sampled Not sampled 14
Not sampled Not sampled 2 No particulate % CV > 30 9 5 Not sampled 20 22 17 Not sampled 26 23 17 Not sampled 29
Not Not 4 3 3 5 5 6 2 Not 3 Not 10 16 Not 15 21
sampled sampled
sampled sampled
sampled
Balistrieri LS, Box SE, Tonkin FW. Modeling precipitation and sorption of elements during mixing of river water and porewater in the Couer d'Alene River Basin. Environ Sci Technol 2003;37:4694–701. Balistrieri LS, Seal II RR, Piatak NM, Paul B. Assessing the concentration, speciation, and toxicity of dissolved metals during mixing of acid-mine drainage and ambient river water downstream of the Elizabeth Copper Mine, Vermont, USA. Appl Geochem 2007;22:930–52. Bencala KE, McKnight DM, Zellweger GW. Evaluation of natural tracers in an acidic and metal-rich mountain stream. Water Resour Res 1987;23:827–36. Bencala KE, McKnight DM, Zellweger GW. Characterization of transport in an acidic and metal-rich mountain stream based on a lithium tracer injection and simulations of transient storage. Water Resour Res 1990;26:989-1000. Bryan SE, Tipping E, Hamilton-Taylor J. Comparison of measured and modelled copper binding by natural organic matter in freshwaters. Comp Biochem Phys C 2002;133: 37–49. Buffle J, Wilkinson KJ, Stoll S, Filella M, Zhang J. A generalized description of aquatic colloidal interactions: the three-colloidal component approach. Environ Sci Technol 1998;32:2887–99. Butler BA. Assessing the fate and transport of metals in a high-gradient acid-mine drainage impacted mountain stream, North Fork Clear Creek, Colorado. Ph.D. Thesis T-6004, Colorado School of Mines, Golden, CO, 2005, 263 pp. Butler BA, Ranville JF, Ross PE. Observed and modeled seasonal trends in dissolved and particulate Cu, Fe, Mn, and Zn in a mining-impacted stream. Water Res 2008a;42:3135–45. Butler BA, Ranville JF, Ross PE. Direct versus indirect determination of suspended sediment associated metals in a mining-influenced watershed. Appl Geochem 2008b;23: 1218–31. Cunningham CG, Naeser CW, Marvin RF, Luedke PG, Wallace AR. Ages of selected intrusive rocks and associated ore deposits in the Colorado mineral belt. U.S. Geological Survey Bulletin, Report # 2109, 1994. Davis JA, Leckie JO. Surface ionization and complexation at the oxide/water interface: II. Surface properties of amorphous iron oxyhydroxide and adsorption of metal ions. J Colloid Interf Sci 1978;67:90-107. Dong D, Derry LA, Lion LW. Pb scavenging from a freshwater lake by Mn oxides in heterogeneous surface coating materials. Water Res 2003;37:1662–6. Dzombak DA, Morel FMM. Surface complexation modeling: hydrous ferric oxide. New York, New York: John Wiley & Sons; 1990. Gustafson JP. Modeling the acid–base properties and metal complexation of humic substances with the Stockholm humic model. J Colloid Interf Sci 2001;244:102–12. Gustafson JP. Visual-MINTEQ, Version 2.30: A windows version of MINTEQA2, version 4.0. [WWW document], accessed November 2004, URL http://www.lwr.kth.se/ english/OurSoftware/Vminteq. Hood E, Williams MW, McKnight DM. Sources of dissolved organic matter (DOM) in a Rocky Mountain stream using chemical fractionation and stable isotopes. Biogeochemistry 2005;74:231–55. Hornberger GM, Bencala KE, McKnight DM. Hydrological controls on dissolved organic carbon during snowmelt in the Snake River near Montezuma, Colorado. Biogeochemistry 1994;25:147–65. Jenne EA. Controls on Mn, Fe, Co, Ni, Cu, and Zn concentrations in soils and water: the significant role of hydrous Mn and Fe oxides. In: Gould RF, editor. Trace inorganics in water, advances in chemistry, vol. 73. Washington DC: American Chemical Society; 1968. p. 337–87. Kimball BA, Runkel RL, Walton-Day K, Bencala KE. Assessment of metal loads in watershed affected by acid mine drainage by using tracer injection and synoptic sampling: Cement Creek, Colorado, USA. Appl Geochem 2002;17:1183–207.
6234
B.A. Butler et al. / Science of the Total Environment 407 (2009) 6223–6234
Kinniburgh DG, Jackson ML, Syers JK. Adsorption of alkaline earth, transition, and heavy metal cations by hydrous oxide gels of iron and aluminum. Soil Sci Soc Am J 1976;40:796–9. Lewis Jr WM, Grant MC. Relationshipsbetween flushing and yield of dissolved substances from a mountain watershed. Soil Sci 1979;128:353–63. Mahony JD, Di Toro DM, Gonzalez AM, Curto M, Dilg M, De Rosa LD, et al. Partitioning of metals to sediment organic carbon. Environ Toxicol Chem 1996;15:2187–97. McKnight DM, Bencala KE. The chemistry of iron, aluminum, and dissolved organic material in three acidic, metal-enriched, mountain streams, as controlled by watershed and in-stream processes. Water Resour Res 1990;26:3087–100. Paulson AJ, Balistrieri LS. Modeling removal of Cd, Cu, Pb, and Zn in acidic groundwater during neutralization by ambient surface waters and groundwaters. Environ Sci Technol 1999;33:3850–6. Shi B, Allen HE, Grassi MT, Ma H. Modeling copper partitioning in surface waters. Water Res 1998;32:3756–64. Smith KS, Ranville JF, Plumlee GS, Macalady DL. Predictive double-layer modeling of metal sorption in mine-drainage systems. In: Jenne EA, editor. Adsorption of metals by geomedia. San Diego: Academic Press; 1998. p. 521–47. Tebo BM, Bargar JR, Clement BG, Dick GJ, Murray KJ, Parker D, et al. Biogenic manganese oxides: properties and mechanisms of formation. Annu Rev Earth Pl Sc 2004;32: 287–328.
Tipping E, Rey-Castro C, Bryan SE, Hamilton-Taylor J. Al(III) and Fe(III) binding by humic substances in freshwaters, and implications for trace metal speciation. Geochim Cosmochim Ac 2002;66:3211–24. Tonkin JW, Balistrieri LS, Murray JW. Modeling metal removal onto natural particles formed during mixing of acid rock drainage with ambient surface water. Environ Sci Technol 2002;36:484–92. Tonkin JW, Balistrieri LS, Murray JW. Modeling sorption of divalent metal cations on hydrous manganese oxide using the diffuse double layer model. Appl Geochem 2004;19:29–53. Turner A, Millward GE, Le Roux SM. Significance of oxides and particulate organic matter in controlling trace metal partitioning in a contaminated estuary. Mar Chem 2004;88:179–92. Wildeman TR, Cain D, Ramiriz AJ. The relation between water chemistry and mineral zonation in the Central City Mining District, Colorado. Water resources problems related to mining, proc., vol. 18. Minneapolis, Mn: American Water Resources Association; 1974. p. 219–29.