American Mineralogist, Volume 95, pages 1642–1649, 2010

0003-004X/10/1112–1642$05.00/DOI: 10.2138/am.2010.3505 1642

Metal retention, mineralogy, and design considerations of a mature permeable reactive barrier (PRB) for acidic mine water drainage in Northumberland, U.K.

Manuel a. Caraballo,1,* esther santofiMia,2 and adaM P. Jarvis3

1Geology Department, University of Huelva, Campus “El Carmen”, E-21071 Huelva, Spain2Dirección de Recursos Minerales y Geoambiente, Área de Investigación sobre Impacto Minero y Uso Sostenible de los Recursos,

Instituto Geológico y Minero de España (IGME), Ríos Rosas, 23, 28003 Madrid, Spain3HERO Group, School of Civil Engineering and Geosciences, Newcastle University, Newcastle upon Tyne, NE1 7RU, U.K.

abstraCt

Mineralogical characterization of the precipitates developed in passive systems treating mine-polluted waters is an essential tool to fully understand and control the removal processes taking place in these systems. In 2008, after five years of operation, a section of the permeable reactive barrier (PRB) at Shilbottle, Northumberland, was subjected to a low intrusive/non-destructive solid sampling. These solid samples were mineralogically characterized by XRD, ESEM-EDS, and sequential extractions. In addition to the solid sampling, 44 water samples obtained in the PRB from January 2004 to August 2009 were used to study the mineral stability of some selected phases in these waters. It was observed that the main iron phases in the PRB were those associated with mineral phases typically developed in non-reducing environments (schwertmannite and goethite), while the presence of a significant amount of pyrite was also observed. The low residence time of the water within the PRB (from 10 to 40 h) appears to be the reason for the absence of a more reducing and less acidic environment in the reactive substrate. An increase of residence time in the PRB, by increasing reactive mixture porosity and resiz-ing the PRB, changes in the reactive material employed (smaller limestone grain size and inclusion of zerovalent iron) and changes in the PRB design (isolating top layer and forced homogeneous flow upward through all the reactive material) are proposed for future reconditioning of the system.

Keywords: Permeable reactive barrier, acid mine drainage, water treatment, metal removal

introduCtion

Inorganic water pollution caused by water-sulfide interaction in mining districts is a world-wide environmental problem. These waters, also referred to as acid mine drainage (AMD), are charac-terized by high metal concentrations (Fe, Al, Cu, Zn, and Mn in the order of mg/L as typical main constituents and a wide set of minor elements such as As, Pb, Ni, Cd, Cr, and Co among others, in the order of µg/L), high sulfate concentrations, and low pH.

AMD affected water remediation can be addressed by two generic approaches: active or passive treatment. The latter is generally preferred for abandoned mine sites due to lower costs, greater sustainability, and because active treatment may be impracticable/inappropriate at these sites, which are often in remote and/or scenic upland areas. Many different passive treat-ment options have been developed to remediate surface AMD, including: (1) anoxic limestone drainages (ALD) (Cravotta and Trahan 1999; Robbins et al. 1999); (2) reducing and alkalinity producing systems (RAPS) (Jage et al. 2001; Mayes et al. 2009); (3) compost wetlands (Jarvis and Younger 1999); and (4) dis-persed alkaline substrates (DAS) (Rötting et al. 2008a, 2008b; Caraballo et al. 2008, 2009b). However, where discharges do not emerge at the surface (e.g., contaminated groundwater plumes)

passive treatment is typically restricted to the use of permeable reactive barriers (PRBs).

One definition of a PRB is “an engineered treatment zone of reactive material(s) that is placed in the subsurface to remedi-ate contaminated fluids as they flow through it” (Environment Agency 1998).This technology was first deployed at full scale to remediate chlorinated solvents in February 1995, at Sunny-vale, California (Warner et al. 1998). On the basis of a recent estimate by the Interstate Technology and Regulatory Council (ITRC 2005), the number of PRBs currently operating in the world could be around 200, with the great majority in North America. About three quarters of the current full-scale PRBs use zerovalent iron (ZVI) as reactive material (Jambor et al. 2005) and are designed to treat waters polluted with organic constituents (Kenneke and McCutcheon 2003) or metals such as U (Morrison 2003; Morrison et al. 2000, 2006), Cr (Puls et al. 1999a, 1999b; Wilkin et al. 2005), or As in industrial sites (Wilkin et al. 2009). There are fewer examples of PRBs designed to remediate AMD. Typically, such PRBs use organic matter as one of their main reactive materials, commonly in combination with limestone or other alkaline material (Blowes et al. 2000; Jambor et al. 2005; Jarvis et al. 2006; Ludwig et al. 2009). Or-ganic carbon is used to favor and enhance dissimilatory bacterial sulfate reduction, increase alkalinity, and remove metals as metal sulfides. Limestone is used to raise water pH to the range of 6 * E-mail: manuel.caraballo@dgeo.uhu.es

CARABALLO ET AL.: SHILBOTTLE PRB MINERALOGY 1643

to 7 to optimize conditions of microbial growth during the first stages of treatment, and to help maintain these neutral condi-tions through the treatment system. Ludwig et al. (2009) also included 20% (v/v) ZVI filing in the reactive mixture to ensure arsenic removal and to provide H2 (g) to help sustain sulfate reduction. Mineralogical studies performed on two of these PRBs treating AMD revealed that precipitation of iron within them can occur in various forms, the most common of which are sulfides and oxyhydroxides in a ring type form (Jambor et al. 2005). It was also observed that sulfides were less abundant than the oxyhydroxides as sinks of metals.

To gain a better understanding of the specific role that removal processes like precipitation, sorption, or organic complexation have on the remediation of AMD within PRB treatment systems, it is necessary to improve and extend the studies performed to date to the few full-scale examples of such systems, and to those units that have been operating over longer periods than the ones typically feasible in lab- and pilot-scale experiments. This is because: (1) scale effects influence system performance, and therefore predictions of performance from lab- and pilot-scale systems are not necessarily reflected in full-scale systems, and (2) the longevity and sustainability of such passive systems can only be assessed with confidence by evaluating the performance of long-running systems e.g., after 5–10 years of operation.

The aim of this work is to study the mineralogy of the pre-cipitates developed within the Shilbottle PRB system and to gain a first insight into how immobile these metals have become, via sequential extraction procedures.

Materials and exPeriMental Methods

Field site and PRB descriptionThe Shilbottle passive treatment system is located in Northumberland, 45 km to

the north of the city of Newcastle Upon Tyne, U.K. (Fig. 1). The system comprises one PRB intersecting the subsurface leachate coming from a colliery spoil heap, a series of three settlement lagoons below the PRB, and an aerobic wetland (Fig. 1). AMD arising from the spoil heap has been considered (in terms of water acidity and metal concentration) as one of the worst quality mining-related discharges in the U.K. (Amos and Younger 2003). The diffuse nature of the seepage, heterogeneity in terms of water quality, and seasonal variation of the flow rate and metal concen-tration coming from the spoil heap, makes it difficult to estimate a representative value for the main parameters. However, pH < 4, acidity > 1400 mg/L as CaCO3 equivalents, Fe > 300 mg/L, Mn > 165 mg/L, Al > 100 mg/L, SO4

2– > 6500 mg/L, and an estimated flow rate in the order of 10 L/s are typical mean values for the discharge from this spoil heap.

A schematic view of the PRB design and its spatial situation with respect to the other parts of the treatment system can be seen in Figure 2. The Shilbottle PRB is 180 m long, up to 3 m deep and 2 m wide. Assuming homogeneous flow into the PRB along its length and substrate porosity of about 30% (Amos and Younger 2003), a nominal residence time of 9 to 36 h can be assumed for flow rates of 2.5 to 10 L/s. These hydraulic parameters imply a hydraulic conductivity ranging from 1.54 × 10–3 to 6.17 × 10–3 cm/s depending on the flow rates considered. On the basis of previous studies by Amos and Younger (2003), it was decided to use a reactive material comprising 25% compost horse manure and straw, 25% green waste com-post, and 50% limestone gravel. The design of the PRB (Fig. 2) forces the AMD leached from the spoil heap to flow horizontally and slightly upward through the reactive media to the surface and subsequently to the settlement lagoons. Figure 2 also shows the distribution of the piezometer nests (B1 and B2) within the PRB, and the situation of the different drill cores taken for mineralogical analyses. Each piezometer nest comprises three individual piezometers, which enable sampling of water from the upper part of the PRB (0.3 m from the top of the PRB), the center of the PRB (1.0 m from the top of the PRB), and the lower portions of the PRB (1.9 m from the top of the PRB).

Water sampling and analysesFrom May 2003 to the present day, monthly sampling of representative points

across the Shilbottle passive treatment system has been carried out. For this study, only samples from piezometer nests B1 and B2 (into PRB) and borehole U1 (into spoil heap) were used (Figs. 1 and 2). Measurements of temperature, pH, Eh, dis-solved oxygen concentration, electrical conductivity and alkalinity, were made in the field, using a Myron 6P Ultrameter (temperature, pH, Eh, and conductivity), YSI 95 DO meter (dissolved oxygen concentration), and Hach digital titrator (alkalinity concentration). The Myron and YSI meters were calibrated prior to each site visit. Measurements were corrected to the Standard Hydrogen Electrode to calculate pe (Nordstrom and Wilde 2005). The Hach digital titrator was used with 1.6 N sulfuric acid and Bromcresol-Green Methyl-Red indicator powder, to give results in units of mg/L as CaCO3. Samples for laboratory analysis were collected in pre-washed polyethylene bottles. Bottles (125 mL) pre-acidified with concentrated hydrochloric acid were used for collection of samples for analysis of metals.

All metals were analyzed using a Varian Vista-MPX ICP-OES. Detection limits were 1 mg/L for K; 0.1 mg/L for Cl, Na, S, and Si; 0.05 mg/L for Al; 0.02 mg/L for Ca and Mg; and 0.01 mg/L for Fe, Mn, and Zn, standard deviations were <5%.

figure 1. Plan section of Shilbottle passive treatment system that includes a PRB, 3 settlement lagoons, and an aerobic wetland. Dashed area corresponds to the PRB section where the study was performed.

figure 2. Schematic section of the PRB design in Shilbottle (not to scale). B1 and B2 correspond to borehole 1 and 2 in the PRB, each one containing 3 piezometers at different depths. C1–C3 are the three cores made during the solid sampling with the position of the samples marked by small squares. U1 is a borehole in the spoil heap.

CARABALLO ET AL.: SHILBOTTLE PRB MINERALOGY1644

Table 1. Sequential extraction procedure employed on this studySequential extraction step Expected dissolved minerals(1) Water soluble fraction: 0.5 g of sample into 30 mL deionizied water, room temperature (RT) Secondary sulfates and other salts(2) Sorbed and exchangeable fraction: 20 mL of 1 M NH4-acetate (4.5 pH buffer), shake for 2 h at RT Cal, adsorbed and exchangeable ions(3) Poorly ordered Fe3+ oxyhydroxides and oxyhydroxysulfates: 20 mL of 0.2 M NH4-oxalate (3 pH buffer) Sch, two-line Fh, secondary Jt, MnO2

1 h shake in darkness and at RT(4) Highly ordered Fe3+ hydroxides and oxides: 20 mL of 0.2 M NH4-oxalate (3 pH buffer) 80 °C water bath for 2 h Gt, Jt, Na-Jt, higher ordered Fh’s(5) Organics and secondary sulfides: 5 mL of 35% H2O2 heated in water bath at 80 °C for 1 h Organic, Cv, Cc-Dg(6) Primary sulfides: combination of KClO3 and HCl, followed by 4 M HNO3 boiling Mainly Py(7) Residue digestion: 3 mL of HNO3 + 7.5 mL of HF + 2.5 mL of HClO4 Silicates, residualNotes: After Dold (2003b) and Caraballo et al. (2009a). Cal = calcite, Sch = schwertmannite, Fh = ferrihydrite, Jt = jarosite, Gt = goethite, Cv = covelline, Cc-Dg = calcocite-digenite, Py = pyrite.

Samples were not routinely filtered. However, periodic filtering of samples (almost every three months) is undertaken for quality assurance/quality control purposes, in accordance with APHA (1998). In the PRB, no statistical difference between the total and filtered concentration of elements were evident. On the light of this statistical resemblance, it was decided to use all the available data to perform the geochemical model although unfiltered samples must be cautiously interpreted.

PHREEQC Interactive 2.15.0 (Parkhurst and Appelo 1999) and the WATEQ4F database (Ball and Nordstrom 1991) were employed to calculate saturation indices (SI = logIAP – logK; IAP = ion activity product) of possible dissolved/precipi-tated minerals in the analyzed water of Borehole 1 and 2 from January 2004 to August 2009. Measured pH, pe, and alkalinity as well as the concentration of the different measured elements in the water samples were introduced into the geochemical model.

Solid sampling and analysesThe holes left after drilling into the PRB could potentially result in preferential

flow paths. To minimize this effect, it was decided to restrict the study to the 30 m PRB section between B1 and B2 (Figs. 1 and 2). This section has the additional benefit of being close to the borehole in the spoil heap (U1), making possible a comparison of water chemistry results from U1, B1, and B2 with the precipitates found in the solid sampling. Three sets of samples were taken at different distances in the line from B1 to B2 (Fig. 2) using an Eijkelkamp soil corer and named as C1 (5 m from B1), C2 (11 m from B1), and C3 (24 m from B1). One sample was recovered from C1 at 1 m depth; three samples from C2 at 0.5, 1, and 1.5 m depth; and two samples from C3 at 0.5 and 1 m depth. All the samples were collected in zip-top plastic bags and kept refrigerated and isolated from light between collection and analysis. Once at the laboratory and 3 h after collection, the inner section of the cores was sub-sampled to focus on the study of unexposed samples. Selected samples were dried in a continuously renovated nitrogen atmosphere to prevent possible oxidation of the reduced phases and were then ground in an agate mill.

In the first phase of investigation of the mineralogy of the samples, crystalline phases in the samples were studied by X-ray diffraction (XRD) using an XPERT-PRO Philips X-ray Diffractometer with CuKα radiation. Samples were also visually studied using an environmental scanning electron microscope (E-SEM) fitted with an energy dispersive spectrometer (EDS) that allowed not only observation of the detailed morphology of the precipitates, but also estimation of their chemical composition. A sequential extraction (SE) procedure (Table 1) was employed to study the distribution and retention of pollutants in the different constituents of the samples. In addition, mineral selective dissolvents were employed as another tool for mineral phase characterization. A more detailed discussion of the selectivity in mineral dissolution and other properties of the reagents employed in the different SE steps can be found in Dold (2003b) and Caraballo et al. (2009a).

results and disCussion

Hydrochemical approach evaluating possible mineral phases developed in the PRB

Monthly sampling of the PRB began in May 2003, but it was not until the first months of 2004 that complete sets of hydrochem-ical data were collected. For this reason, the hydrochemical study performed with PHREEQC to obtain the saturation index (SI) of the expected mineral phases in the PRB was restricted to the 44 water samples obtained from January 2004 to August 2009.

To gain a first insight into the different iron phases that could

be formed in the PRB, it was decided to plot water samples from the lower piezometer of B1 and B2 (the more reducing environ-ment in the PRB) into the pH-pe diagram for Fe-S-K-O-H system at 25 °C (Fig. 3). This diagram contains the stability fields for jarosite (Jt), schwertmannite (Sch), ferrihydrite (Fh), goethite (Gt), dissolved species, and pyrite (Py). Stability fields for the different mineral phases were plotted using available data from Bigham and Nordstrom (2000). Water data from borehole U1 were also plotted in the pH-pe diagram to observe the difference between the water before and after its interaction with the reactive material in the PRB. Inspection of these data (Fig. 3) illustrates that the PRB has performed as designed at least to some degree; pH increases from the borehole water in the spoil heap (U1) to the two piezometers in the PRB itself (B1L and B2L), and pe decreases, consistent with a more reducing environment. It is also evident that pH conditions at B1L are typically higher than at B2L, suggesting that there is heterogeneity in terms of treat-ment performance. Further examination of the data plotted in this diagram (Fig. 3) reveals that water in the spoil heap is distributed between the stability fields corresponding to schwertmannite and goethite, while the water in the PRB predominantly remains in the stability field of goethite with some samples hosted in the dissolved species field. The high-pe value commonly detected in the PRB water results in the plotted data being far from pyrite stability field, although some isolated samples have been plotted in the vicinity of this stability field.

To gain further insight into the mineral phases that could precipitate in the PRB, SI values for some typical minerals pre-cipitated in similar passive treatment systems were reviewed. On the basis of previous studies (Caraballo et al. 2009a; Jambor et

figure 3. pH-pe diagram for Fe-S-K-O-H system at 25 °C. Jt = jarosite, Sch = schwertmannite, Fh = ferrihydrite, Gt = goethite, and Py = pyrite. B1L and B2L correspond to the piezometers at the lower part of the PRB in B1 and B2, respectively, and U1 to a borehole in the spoil heap.

CARABALLO ET AL.: SHILBOTTLE PRB MINERALOGY 1645

al. 2005; Wilkin et al. 2009) and taking into account the major dissolved elements in these waters (Fe, Al, Mn, and SO4

2–), it was decided to select jarosite, schwertmannite, goethite, ferrihydrite, and pyrite (Fe-phases); basaluminite and gibbsite (Al-phases); and pyrolusite (Mn-phase) as candidate minerals, i.e., these are the phases typically found in other systems. As can be seen in Figure 4, results obtained for the lower part of B1 and B2 show basically the same tendency for all the mineral phases studied. A detailed examination of the different SI obtained reveals how goethite appears constantly oversatured, throughout the five year monitoring period, while ferrihydrite also displays a quite steady pattern, but in contrast is slightly undersaturated. Regard-ing both schwertmannite and jarosite, a more variable pattern is observed with respect to the SI. Although both minerals tend to be oversaturated throughout the whole sampling period, some sig-nificant stages of undersaturation are evident. SI for the selected Al-phases clearly show basaluminite and gibbsite consistently oversaturated for the entire monitoring period. Concerning the SI displayed by the two metal sulfide phases selected in this study Figure 4 clearly shows how both pyrite and pyrolusite constantly remain in the undersaturated zone (with the exception of two sampling occasions when pyrite appeared close to equilibrium and marginally oversaturated).

Mineralogical characterization of the precipitatesTo obtain an overview of the crystalline phases present in the

PRB solid samples, XRD analysis was performed. It is important to take into account that typical Fe and Al precipitates developed by AMD treatment tend to range from poor to moderate crystal-

figure 4. Saturation indices calculated with PHREEQC Interactive 2.15.0 for some selected mineral phases in the water of the lower PRB section in B1 and B2. Jt = jarosite, Sch = schwertmannite, Fh = ferrihydrite, Gt = goethite, Bas = basaluminite, Gib = gibbsite, Py = pyrite, and Prl = pyrolusite. Solid symbols represent unfiltered water samples while empty symbols are used for filtered water samples.

linity and therefore their detection by XRD is difficult if other more crystalline phases are present. This problem can sometimes be overcome when selective mineral dissolution of the more crystalline phases is possible (Caraballo et al. 2009a). The pres-ence of quartz and kaolinite in all the samples was confirmed by XRD (Table 2). Gypsum also appeared in all the samples except sample at 0.5 m in C2. Calcite was only detected in samples C1-1 and C3-1, perhaps implying reactive media heterogeneity. Due to the high quartz, gypsum, and calcite crystallinity the pres-ence of goethite could only be confirmed in one sample (C2, 1.5 m depth). After selective dissolution of gypsum, calcite, and goethite, the presence of pyrite was also confirmed in sample C2-1.5 (Sample C2-1.5b in Table 2). Illite was only detected in two samples of C2 (Table 2).

A selective mineral dissolvent (Table 2) was employed as an additional approach to investigating the mineralogy of the precipitates. Water extraction (Step 1, Fig. 5) of the precipitates releases a significant concentration of Ca and S. Calculations of

Table 2. Mineral phases identified by XRDSamples Mineral phases Qtz Kln Gyp Ill Cal Gt PyC1-1 x x x x C2-0.5 x x C2-1 x x x x C2-1.5 x x x x x C2-1.5b x x x xC3-0.5 x x x C3-1 x x x x Note: Qtz = quartz, Kln = kaolinite, Gyp = gypsum, Ill = illite, Cal = calcite, Gt = goethite, Py = pyrite

CARABALLO ET AL.: SHILBOTTLE PRB MINERALOGY1646

figure 5. Cumulative graphs for the concentration of the main constituent elements obtained after each step of the SE for the different solid samples studied. Ca/S, Fe/S, and S/Fe correspond to the molar ratio of these elements.

Ca/S molar ratio for this step reveal how the great majority of the samples exhibit a Ca/S molar ratio close to 1 (Fig. 5). Taking into account that Ca/S molar ratio for gypsum is 1, the presence of this mineral can be assumed in almost all the samples studied (exception of C2-1.5 and C3-05).

SE step 2 is designed to release into solution absorbed and exchangeable ions, as well as the dissolution products of calcite. Evidence of calcite dissolution can only be clearly observed in samples C1-1 and C3-1, in agreement with the XRD results pre-viously shown. According to the literature (Bigham et al. 1990; Caraballo et al. 2009a; Jönsson et al. 2005; Webster et al. 1998), the noticeable concentration of S released in this step could be

attributed to sulfate adsorption on schwertmannite. The origin of Mn released during this step is unclear, but it was probably hosted in the structure of calcite.

Significant iron recovery was observed using NH4-oxalate at the third extraction step. The use of this extractant in certain environmental conditions has shown quite robust results in terms of the selective dissolution of schwertmannite and other minerals like 2-line ferrihydrite and MnO2 (Caraballo et al. 2009a; Hall et al. 1996; Webster et al. 1998; Dold 2003a, 2003b). Although this extractant does not allow differentiation between schwertman-nite and 2-line ferrihydrite with certainty, the Fe/S molar ratio obtained in some samples (Fig. 5), as well as the hydrochemical

CARABALLO ET AL.: SHILBOTTLE PRB MINERALOGY 1647

conditions under which these precipitates were formed (Figs. 3 and 4), suggests the former as the most probable Fe-phase precipitated. At this step a high fraction of the total Mn was recovered, probably as MnO2 (Fig. 5). Notable Al recovery was also observed at this stage of the SE.

In previous studies where goethite was present (Caraballo et al. 2009a; Dold 2003a, 2003b; Hall et al. 1996), all the samples subjected to SE step 4 released a remarkable Fe concentration into solution. The results here also show significant Fe release during this step, showing agreement with the previous XRD detection of goethite, and suggesting the presence of this mineral in all the samples.

A combination of KClO3 and HCl followed by boiling 4 M HNO3 (Dold 2003b) was employed to achieve the selective dissolution of pyrite and other possible sulfide phases. A sig-nificant concentration of S was only unequivocally observed in sample C2-1.5 (the deepest one in the PRB), which was also observed to be an important sink for Fe and, to a lesser extent, Mn (Fig. 5). This observation clearly illustrates the presence of Fe and Mn sulfides in the lowest part of the PRB, while the S/Fe molar ratio obtained suggests pyrite as the most probable Fe sulfide phase (Fig. 5).

The SE final residue was dissolved by using HNO3, HF, and HClO4 (Caraballo et al. 2009a). The remarkable concentrations of Al, K, and Na obtained in this final step of the SE can be at-tributed to the presence of phyllosilicates in all the samples. The higher concentration of these elements in samples C2-1 and C2-1.5 is supported by the XRD observation of illite in these samples, in addition to the kaolinite detected in all the samples.

It was decided to additionally study the samples using ESEM-EDS to obtain visual confirmation of the existence of sulfides and to have an idea of the morphology and crystallinity of the different precipitates. In accordance with the XRD and SE stud-ies, some good examples of framboidal pyrite were found only in sample C2-1.5, one of which is shown in the left side of Figure 6. Another unexpected Ni-rich phase was also observed in this sample (Fig. 6, right side). The presence of native Ni has been reported in a previous study performed in an organic PRB treat-

ing AMD pollution (Jambor et al. 2005). A ubiquitous Fe-rich phase with a minor amount of S was found in all the samples. Morphology and semi-quantitative composition of this phase is in accordance with the appearance of schwertmannite, as previ-ously suggested.

Precipitates distribution, PRB performance, and future considerations

In light of the hydrochemical and mineralogical studies per-formed at the Shilbottle PRB, it can be concluded that although the design principle for the PRB was to act as a reducing and alkalinity producing system, over the full five years of monitoring data evaluated here, a reductive environment was only marginally achieved in some areas at the base of the PRB. The main sink for iron (Fig. 7) has been in the form of mineral phases typically developed in non-reducing environments (schwertmannite and goethite), while the presence of a significant amount of pyrite was restricted to sample C2 at 1.5 m depth. Both the small or negligible amount of S removed and the high Al removal (Fig. 7), indicate that the actual processes occurring in the PRB are closer to treatment systems based on the use of alkaline substrates (Caraballo et al. 2008, 2009b; Rötting et al. 2008b; Watten et al. 2005), rather than the systems using organic matter (Blowes et al. 2000; Jambor et al. 2005; Ludwig et al. 2009). However, previous investigations of the PRB at Shilbottle demonstrated significant removal of sulfate across the system, with concomi-tant increases in pH and removal of metals. Specifically, Jarvis et al. (2006) noted reductions in mean sulfate concentration of over 40% across the PRB for the period May 2003 to May 2005. Thus, it appears that the system initially operated effectively as a sulfate-reducing system, but in recent years has evolved to a point where the dominant sinks for iron are schwertmannite and goethite.

A possible explanation for the absence of a more reducing environment in the reactive substrate is the low residence time of the PRB, typical values for which are in the range of 10 to 40 h, coupled with the highly polluted nature of the AMD. Previous studies have shown that the nominal residence time

figure 6. ESEM images and EDS analysis for pyrite framboid (on the left side) and a Ni-rich phase, probably native Ni (right side of the figure).

CARABALLO ET AL.: SHILBOTTLE PRB MINERALOGY1648

for the achievement and maintenance of a bacterially mediated reducing environment is in the range of 7–10 days (Neculita et al. 2008). Another widely accepted requirement for effective bacterial sulfate reduction is elevated pH. It has been observed that sulfate reducing bacteria generally have an optimal growth in the pH range 6 to 7, although recent studies have noted that there is some potential for sulfidogentic processes to appears at pH values lower than 5 (Church et al. 2007; Kimura et al. 2006). During early years of operation higher pH was evident (Jarvis et al. 2006), suggesting higher potential for sulfate reduction and alkalinity generation within the PRB, but the pH achieved in the PRB in recent years (in the range 4 to 5) implied that alkalinity generation capacity and sulfidogenic processes were exhausted relatively quickly in this system, although the existence of the latter cannot be completely ruled out.

A consequence of the changes in the geochemical behavior of the PRB has been a gradual increase in the export of Fe, and Al in particular, to the receiving watercourse, principally as a result of the lower pH evident in the settlement lagoons and aerobic wetland. To achieve improved metal removal in the PRB, some moderate changes in its design are proposed for a future reconditioning of the system:

• The use of a greater proportion of limestone, with smaller grain size, in the reactive media. The use of calcite sand (d10 = 0.3 mm, d50 = 1.4 mm, dmax = 5 mm) has shown encouraging results in several passive treatment systems treating AMD with very high metal concentration in the Iberian Pyrite Belt (IPB), Spain (Caraballo et al. 2008, 2009b; Rötting et al. 2008b). ZVI could be another beneficial reagent to be included in the new PRB reactive

mixture. ZVI has been used to great effect, in combination with limestone and organic matter, in a PRB treating high polluted AMD in Charleston, U.S.A. (Ludwig et al. 2009), and also in some batch experiments were AMD metal removal was studied after adding different amounts of ZVI (Wilkin and McNeil 2003). The addition of ZVI would also help to maintain sulfate reduction via the generation of H2 (g) within the reactive media.

• Increase of the PRB residence time by increasing reac-tive substrate porosity and PRB size. Using a mixture of wood shavings and limestone sand some treatments in the IPB have achieved a porosity ranging from 50 to 70% (Caraballo et al. 2008; Rötting et al. 2008b, 2008c). Assuming a 50% porosity, 2160 m3 of PRB (180 m long × 4 m wide × 3 m deep) and a flow rate range 1 L/s to 10 L/s, the resulting residence time would increase to 1.25 to 12.5 days. Although such an increase in size might be feasible at this particular site, it is however worth not-ing that achieving residence time of 6–10 days in compost-based passive systems is in general very difficult in the U.K. and other parts of Europe, simply due to restrictions in land availability that limit the absolute system size.

• More complete capping of the PRB surface with a layer of clay or other impermeable material (Fig. 8) to enhance the reducing environment within the PRB.

• Addition of a layer of inert material of high porosity and hydraulic conductivity (e.g., brick rubble) in the base of the PRB (Fig. 8), which may help to resolve the flow heterogeneity problems that currently appear to be limiting system performance (Fig. 8).

aCknowledgMentsWe gratefully acknowledge Jane Davis, Patrick Orme, and Gerardo González

(Newcastle University) for assistance with fieldwork and chemical analysis, monthly water sampling and solid sampling. This study was funded by the European Union Coal mine Sites for Targeted Remediation Research (CoSTaR) project (2004–2008). CoSTaR was funded by the European Community under the “Structuring the European Research Area” specific program Research Infrastruc-tures Action (Ref: RITA-CT-2003-506069). The CoSTaR project was coordinated by Paul L. Younger (Newcastle University), to whom we express our gratitude. We also express our thanks to Northumberland County Council, U.K., which owns the Shilbottle PRB. M.A.C. was financially supported by the Spanish Government with a FPU Ph.D. fellowship. We also thank Richard Wilkin (Associate Editor) and two anonymous reviewers for their comments that significantly improved the quality of this paper.

referenCes CitedAmos, P.W. and Younger, P.L. (2003) Substrate characterization for a subsurface re-

active barrier to treat colliery spoil leachate. Water Research, 37, 108–120.APHA (American Public Health Association) (1998) Standard methods for the

examination of water and wastewater, 20th Edition. American Water Works

figure 7. Main constituent contribution in percentage to the composition of the precipitates developed in the PRB. Fe(ox) and S(ox) correspond to percentage of this element precipitated in the form of oxide, hydroxide, or oxyhydroxysulfates (Steps 2–4) while Fe(sulf) and S(sulf) correspond to the percentage in the form of sulfide or bonded to the organic matter (Steps 5 and 6). S in gypsum (Step 1) was omitted because due to its high water solubility retention of S in this mineral is quite ephemeral.

figure 8. New design proposes in this work for the PRB at Shilbottle.

CARABALLO ET AL.: SHILBOTTLE PRB MINERALOGY 1649

Association and the Water Environment Federation, Washington, D.C.Ball, J.W. and Nordstrom, D.K. (1991) User’s manual for WATEQ4F with revised

thermodynamic database and test cases for calculating speciation of major, trace and redox elements in natural waters, U.S. Geological Survey Water Resources Investigation, Report 91-183.

Bigham, J.M. and Nordstrom, D.K. (2000) Iron and aluminum hydroxysulfates from acid sulfate waters. In C.N. Alpers, J.L. Jambor, and D.K. Nordstrom, Eds., Sulfate Minerals: Crystallography, geochemistry, and environmental significance, 40, p. 351–403. Reviews in Mineralogy and Geochemistry, Mineralogical Society of America, Chantilly, Virginia.

Bigham, J.M., Schwertmann, U., Carlson, L., and Murad, E. (1990) A poorly crystal-lized oxyhydroxysulfate of iron formed by bacterial oxidation of Fe(II) in acid mine waters. Geochimica et Cosmochimica Acta, 54, 2743–2758.

Blowes, D.W., Ptacek, C.J., Benner, S.G., McRae, C.W.T., Bennett, T.A., and Puls, R.W. (2000) Treatment of inorganic contaminants using permeable reactive barriers. Journal of Contaminant Hydrology, 45, 123–137.

Caraballo, M.A., Macías, F., Rötting, T.S., Nieto, J.M., and Ayora, C. (2008) Disperse alkaline substrate passive remediation at Mina Esperanza (Iberian Pyrite Belt, SW Spain). Geochimica et Cosmochimica Acta, 72(12, Supple-ment 1), A135.

Caraballo, M.A., Rötting, T.S., Nieto, J.M., and Ayora, C. (2009a) Sequential extraction and DXRD applicability to poorly crystalline Fe- and Al-phase characterization from an acid mine water passive remediation system. American Mineralogist, 94, 1029–1038.

Caraballo, M.A., Rötting, T.S., Macías, F., Nieto, J.M., and Ayora, C. (2009b) Field multi-step limestone and MgO passive system to treat acid mine drainage with high metal concentrations. Applied Geochemistry, 24, 2301–2311.

Church, C., Wilkin, R., Alpers, C., and Rye, R. (2007) Microbial sulfate reduction and metal attenuation in pH 4 acid mine water. Geochemical Transactions, 8, 1–12.

Cravotta, C.A. and Trahan, M.K. (1999) Limestone drains to increase pH and remove dissolved metals from acidic mine drainage. Applied Geochemistry, 14, 581–606.

Dold, B. (2003a) Dissolution kinetics of schwertmannite and ferrihydrite in oxidized mine samples and their detection by differential X-ray diffraction (DXRD). Applied Geochemistry, 18, 1531–1540.

——— (2003b) Speciation of the most soluble phases in a sequential extraction procedure adapted for geochemical studies of copper sulfide mine waste. Journal of Geochemical Exploration, 80, 55–68.

Environment Agency (1998) The State of the Environment of England and Wales: Fresh Waters. The Stationery Office, London.

Hall, G.E.M., Vaive, J.E., Beer, R., and Hoashi, M. (1996) Selective leaches re-visited, with emphasis on the amorphous Fe oxyhydroxide phase extraction. Journal of Geochemical Exploration, 56, 59–78.

ITRC (Interstate Technology and Regulatory Council) (2005) Permeable reactive barriers: Lessons learned/new directions. PRB-4, Washington, D.C. (www.itrcweb.org).

Jage, C.R., Zipper, C.E., and Noble, R. (2001) Factors affecting alkalinity genera-tion by successive alkalinity-producing systems: Regression analysis. Journal of Environmental Quality, 30, 1015–1022.

Jambor, J.L., Raudsepp, M., and Mountjoy, K. (2005) Mineralogy of permeable reactive barriers for the attenuation of subsurface contaminants. Canadian Mineralogist, 43, 2117–2140.

Jarvis, A.P. and Younger, P.L. (1999) Design, construction and performance of a full-scale compost wetland for mine-spoil drainage treatment at quaking houses. Water and Environment Journal, 13, 313–318.

Jarvis, A.P., Moustafa, M., Orme, P.H.A., and Younger, P.L. (2006) Effective remediation of grossly polluted acidic, and metal-rich, spoil heap drainage using a novel, low-cost, permeable reactive barrier in Northumberland, U.K. Environmental Pollution, 143, 261–268.

Jönsson, J., Persson, P., Sjöberg, S., and Lövgren, L. (2005) Schwertmannite precipitated from acid mine drainage: phase transformation, sulphate release and surface properties. Applied Geochemistry, 20, 179–191.

Kenneke, J.F. and McCutcheon, S.C. (2003) Use of pretreatment zones and zero-valent iron for the remediation of chloroalkenes in an oxic aquifer. Environ-mental Science and Technology, 37, 2829–2835.

Kimura S., Hallberg K.B., and Johnson D.B. (2006) Sulfidogenesis in low pH (3.8–4.2) media by a mixed population of acidophilic bacteria. Biodegrada-tion, 17, 57–65.

Ludwig, R.D., Smyth, D.J.A., Blowes, D.W., Spink, L.E., Wilkin, R.T., Jewett, D.G., and Weisener, C.J. (2009) Treatment of arsenic, heavy metals, and acidity using a mixed ZVI-compost PRB. Environmental Science and Technology, 43, 1970–1976.

Mayes, W.M., Batty, L.C., Younger, P.L., Jarvis, A.P., Kõiv, M., Vohla, C., and

Mander, U. (2009) Wetland treatment at extremes of pH: A review. Science of the Total Environment, 407, 3944–3957.

Morrison, S. (2003) Performance evaluation of a permeable reactive barrier us-ing reaction products as tracers. Environmental Science and Technology, 37, 2302–2309.

Morrison, S.J., Metzler, D.R., and Carpenter, C.E. (2000) Uranium precipitation in a permeable reactive barrier by progressive irreversible dissolution of zerovalent iron. Environmental Science and Technology, 35, 385–390.

Morrison, S.J., Mushovic, P.S., and Niesen, P.L. (2006) Early breakthrough of molybdenum and uranium in a permeable reactive barrier. Environmental Science and Technology, 40, 2018–2024.

Neculita, C.-M., Zagury, G.J., and Bussière, B. (2008) Effectiveness of sulfate-reducing passive bioreactors for treating highly contaminated acid mine drainage: II. Metal removal mechanisms and potential mobility. Applied Geochemistry, 23, 3545–3560.

Nordstrom, D.K. and Wilde, F.D. (2005) Reduction-oxidation potential (electrode method). In F.D. Wilde, Ed., Techniques of Water-Resources Investigations Reports, Book 9; chapter 6.5. National Field Manual for the Collection of Water-Quality Data, U.S. Geological Survey, http://water.usgs.gov/owq/FieldManual/Chapter6/6.5_v_1.2.pdf.

Parkhurst, D.L. and Appelo, C.A.J. (1999) User’s guide to PHREEQC (Version 2)—A computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. U.S. Geological Survey Water-Resources Investigations Report 99-4259, Denver, Colorado.

Puls, R.W., Blowes, D.W., and Gillham, R.W. (1999a) Long-term performance monitoring for a permeable reactive barrier at the U.S. Coast Guard Support Center, Elizabeth City, North Carolina. Journal of Hazardous Materials, 68, 109–124.

Puls, R.W., Paul, C.J., and Powell, R.M. (1999b) The application of in situ permeable reactive (zero-valent iron) barrier technology for the remediation of chromate-contaminated groundwater: a field test. Applied Geochemistry, 14, 989–1000.

Robbins, E.I., Cravotta, C.A., Savela, C.E., and Nord, G.L. (1999) Hydrobiogeo-chemical interactions in anoxic limestone drains for neutralization of acidic mine drainage. Fuel, 78, 259–270.

Rötting, T.S., Ayora, C., and Carrera, J. (2008a) Improved passive treatment of high Zn and Mn concentrations using caustic magnesia (MgO): Particle size effects. Environmental Science and Technology, 42, 9370–9377.

Rötting, T.S., Caraballo, M.A., Serrano, J.A., Ayora, C., and Carrera, J. (2008b) Field application of calcite dispersed alkaline substrate (calcite-DAS) for pas-sive treatment of acid mine drainage with high Al and metal concentrations. Applied Geochemistry, 23, 1660–1674.

Rötting, T.S., Thomas, R.C., Ayora, C., and Carrera, J. (2008c) Passive treatment of acid mine drainage with high metal concentrations using dispersed alkaline substrate. Journal of Environmental Quality, 37, 1741–1751.

Watten, B.J., Sibrell, P.L., and Schwartz, M.F. (2005) Acid neutralization within limestone sand reactors receiving coal mine drainage. Environmental Pollu-tion, 137, 295–304.

Warner, S.D., Yamane, C.L., Bice, N.T., Szerdy, F.S., Vogan, J., Major, D.W., and Hankins, D.A. (1998) Technical update: The first commercial subsurface permeable reactive treatment zone composed of granular zero-valent iron. In G.B. Wickramanayake and R.E. Hinchee, Eds., Designing and Applying Treat-ment Technologies: Remediation of Chlorinated and Recalcitrant Compounds, 145–150. Battelle Press, Columbus, Ohio.

Webster, J.G., Swedlund, P.J., and Webster, K.S. (1998) Trace metal adsorption onto an acid mine drainage iron(III) oxy hydroxy sulfate. Environmental Science Technology, 32, 1361–1368.

Wilkin, R.T. and McNeil, M.S. (2003) Laboratory evaluation of zero-valent iron to treat water impacted by acid mine drainage. Chemosphere, 53, 715–725.

Wilkin, R.T., Su, C., Ford, R.G., and Paul, C.J. (2005) Chromium-removal processes during groundwater remediation by a zerovalent iron permeable reactive barrier. Environmental Science and Technology, 39, 4599–4605.

Wilkin, R.T., Acree, S.D., Ross, R.R., Beak, D.G., and Lee, T.R. (2009) Perfor-mance of a zerovalent iron reactive barrier for the treatment of arsenic in groundwater: Part 1. Hydrogeochemical studies. Journal of Contaminant Hydrology, 106, 1–14.

Manuscript received January 20, 2010Manuscript accepted June 30, 2010Manuscript handled by richard Wilkin