Biogeochemical interactions in a compost wetland for spoil heap leachate remediation at Quaking Houses, County Durham, UK

The linked redox geochemistry of carbon, sulphur and iron is at the heart of remediation of acidic waters within passive, anaerobic wetland sediments. Sediment cores were extracted from the Quaking Houses constructed wetland in County Durham, England. Pore water profiles show concomitant reduction of both sulphate and iron oxide. Solid phase analyses show that iron sulphides are precipitated. The build up of dissolved iron in sediment pore waters indicates that the net rate of iron reduction (FeR) exceeds that of bacterial sulphate reduction (BSR). Incubation data also supports this, in that FeR (chemical plus microbial) proceeds at a faster rate than BSR. BSR was found to generate alkalinity at around 6 times the rate of microbial FeR. Despite high rates of BSR, only 10 – 15% of Fe is present as sulphide precipitates. Around 80 – 90% of solid phase iron is sequestered into the sediment as oxides/oxyhydroxides of unknown composition. INTRODUCTION The effective treatment of AMD discharges through passive techniques is not only a function of engineering design, but also biogeochemistry. Passive treatment systems for acidic drainage are designed to exploit the linked biogeochemical redox cycles of carbon, sulphur, and iron with the dual aims of raising pH and alkalinity, and removing metal pollutants such as iron, aluminium, manganese, and zinc. Both aerobic and anaerobic passive treatment systems have been deployed. At least for net acidic waters, aerobic systems are of limited use. Although iron can be successfully removed by oxidation and hydrolysis, and trace metals can be subsequently removed from solution by adsorption onto the surface of oxihydroxides (Johnson & Thornton, 1987) the oxidation and hydrolysis of both Fe and Al generates significant acidity: 4Fe + O2 + 6H2O → 4FeOOH + 8H (1) For this reason anaerobic systems provide a much higher potential for successful remediation. Successful remediation of AMD using anaerobic systems depends on the balance of reactions (chemical and abiotic) within the C-Fe-S cycles. There is an immense literature on the C-Fe-S geochemistry of marine and freshwater sediments (e.g.(Berner, 1985; Holmer & Storkholm, 2001; Kostka & Luther, 1994; Lin & Morse, 1991; Thamdrup, 2000) which reveals a diverse suite of potential reactions. Some of the key reactions include the microbially mediated dissimilatory reduction of sulphate and iron oxide, the formation of iron sulphides, and the reoxidation of sulphide by iron oxides, or ferric iron. For example: CH2O + 4FeOOH + 7H → 4Fe + HCO3 + 6H2O (2) 2CH2O + SO4 + → 2HCO3 + H2S (3) 2FeOOH + H2S → 2Fe + S + 4OH (4) 3H2S + 2 FeOOH → S + 2FeS + 4H2O (5) FeS2 + 14FeOOH + 6H2O → 15Fe + 2SO4 + 26OH (6) FeS2 + 14Fe + 8H2O → 2SO4 + 15Fe + 16H (7) Most of these reactions involve the generation or consumption of protons and/or the generation of alkalinity. It is clear therefore that the overall balance of the C-S-Fe cycles dictate the remediation potential of an anaerobic treatment system. Design of anaerobic constructed wetlands has focussed on microbial sulphate reduction as being the key process for alkalinity generation (Fortin et al., 2000; Hedin et al., 1988; Mcintyre et al., 1990; Younger et al., 2002) though literature regarding the geochemistry of constructed anaerobic wetlands is sparse. The general aim of this project was to study C-S-Fe cycling to test this assumption, using a range of geochemical methods. A temporal geochemical study and incubation experiments were employed to this end. Specific aims were to quantify the rates of key reactions within a passive treatment system i.e. bacterial sulphate reduction and iron reduction, and to use that information to estimate carbon turnover and alkalinity generation. This paper shows results of a biogeochemical study of the Quaking Houses constructed wetland in County Durham, England (Morrison, 2005). STUDY AREA Quaking Houses Wetland can be located on Ordnance Survey map, Landranger series 88, Tyneside & Co. Durham Area, at grid reference NZ 185 506. It lies alongside the Stanley Burn, a small tributary of the River Wear in Co. Durham, England. The artificial wetland was commissioned in November 1997 to treat acidic colliery spoil leachate, which was draining into the Stanley Burn. Prior to treatment the leachate was characterized by high 9 INTERNATIONAL MINE WATER CONGRESS 404 loadings of iron and aluminium (~ 0.4 mM for both), and low pH (~4) (Jarvis, 2000). In 1998 the spoil heap responsible for the discharge was capped, resulting in raised pH and lower metal concentrations, although still net acidic. The design of the wetland, constrained by hydrogeological factors, consists of two ponds (Figure 1). Both ponds contain a substrate composed of horse and cow manure, and composted municipal waste in a ratio of 30:40:30. Water from the influent pipe enters the first cell with a mean flow rate of 80 L min, then decants over a weir into the second cell where it is dispersed by vegetation and baffles. This is primarily a surface-flow system with nominal average water retention time of ~24 hours (Younger et al., 2002). Figure 1. A schematic of Quaking Houses Wetland. Over a period of 16 months, field data, surface waters and four sediment cores were sampled for analysis. Porewaters were extracted under anaerobic conditions from each core at depth intervals of 1 cm. Porewaters and sediments then underwent a series of extractions and analyses to determine concentrations, enabling concentration with depth profiles to be plotted, for redox sensitive Fe and S species. Following the geochemical characterization of the sediment and waters, a series of incubation experiments were carried out to determine the rates of sulphate and iron reduction. RESULTS Porewater and solid phase data for iron and sulfur species are shown in Figures 2-5, respectively. Superficially, the porewater data resemble trends which have been reported previously in natural, marine sediments (Canfield et al., 1993; Thamdrup et al., 2000; Wijsman et al., 2001). Reduction of iron oxides in the top few centimeters of the wetland sediments results in increasing abundances of dissolved iron, below which concentrations of dissolved Fe decline steadily. Microbial sulfate reduction occurs in the surficial sediments resulting in steadily declining concentrations of sulfate. Figure 2 illustrates this trend clearly, but also highlights a temporal effect where there is a net diffusion of sulphate out of the sediment over the top 2 cm in the November and January cores. Careful appraisal of the porewater profiles reveal that in fact they differ from those reported in natural sediments in several important ways. Most obviously, porewater Fe increases to the very high concentration of ~ 1.3 mM (Figure 3), despite the rapid generation of sulfide via sulfate reduction. In the top few centimeters of the sediment, rapid reduction of iron occurs concomitantly with rapid reduction of sulfate, with net production of reduced iron being greater than net production of reduced sulfur. Solid phase sulfide data (Figure 4) show that the reduced iron and sulfur precipitate as pyrite (FeS2) and acid volatile sulfide (FeS and similar compounds). Pathways of iron sulfide formation in these sediments are not yet defined but could include direct precipitation from the end products of bacterial sulphate reduction (BSR) and microbial iron reduction (MFeR), and the abiotic reaction of dissolved sulfide with solid phase iron oxides; the detection of significant concentrations of elemental sulphur (up to 225 μmol g) support abiotic reaction between Fe(III) and sulphide. 9 INTERNATIONAL MINE WATER CONGRESS 405 Figure 5 shows the exceptionally high concentration of amorphous iron oxide in the sediment (mean value for all cores is 500 μmol g over top 2cm). This was a surprising find given the very high organic carbon contents (10 – 35%) of these sediments, although ascorbate-extractable Fe(III) (Fe-asc) has been measured in natural sediments (Kostka & Luther, 1994; Luther et al., 2003; Rutten & De Lange, 2003).