Rate Controls on the Chemical Weathering of Natural Polymineralic Material

Applied Geochemistry

Applied Geochemistry 21 (2006) 377–403 www.elsevier.com/locate/apgeochem

Rate controls on the chemical weathering of natural polymineralic material. II. Rate-controlling mechanisms and mineral sources and sinks for element release from four UK mine sites, and implications for comparison of laboratory and ﬁeld scale weathering studies

K.A. Evans a,*, D.C. Watkins a, S.A. Banwart b a Groundwater Protection and Restoration Group, Department of Civil Engineering, University of Sheﬃeld, Mappin St, Sheﬃeld S1 3JD, UK b Camborne School of Mines, Tremough Campus, Treliever Road, Penryn, Cornwall TR10 9EZ, UK

Received 9 May 2005; accepted 24 October 2005 Editorial handling by R. Fuge

Abstract

Predictions of mine-related water pollution are often based on laboratory assays of mine-site material. However, many of the factors that control the rate of element release from a site, such as pH, water–rock ratio, the presence of secondary minerals, particle size, and the relative roles of surface-kinetic and mineral equilibria processes can exhibit considerable variation between small-scale laboratory experiments and large-scale field sites. Monthly monitoring of mine effluent and analysis of natural geological material from four very different mine sites have been used to determine the factors that control the rate of element release and mineral sources and sinks for major elements and for the contaminant metals Zn, Pb, and Cu. The sites are: a coal spoil tip; a limestone-hosted Pb mine, abandoned for the last 200 a; a coal mine; and a slate-hosted Cu mine that was abandoned 150 a ago. Hydrogeological analysis of these sites has been performed to allow field fluxes of elements suitable for comparison with laboratory results to be calculated. Hydrogeological and mineral equilibrium control of element fluxes are common at the field sites, far more so than in laboratory studies. This is attributed to long residence times and low water–rock ratios at the field sites. The high water storativity at many mine sites, and the formation of soluble secondary minerals that can efficiently adsorb metals onto their surfaces provides a large potential source of pollution. This can be released rapidly if conditions change significantly, as in, for example, the case of flooding or disturbance. 2006 Elsevier Ltd. All rights reserved.

1. Introduction

* Corresponding author. Present address: Australian National University, RSES, Building 61, ANU, Mills Road, Canberra Mining activities expose large volumes of fresh ACT 0200, Australia. Fax: +61 2 61250738. rock to atmospheric conditions. Subsequent weath- E-mail address: [email protected] (K.A. Evans). ering and release of rock constituents, especially via

0883-2927/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.apgeochem.2005.10.002 378 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 sulphide oxidation, can lead to acidic solutions and oratory results to field situations. Laboratory- elevated concentrations of environmentally undesir- derived mineral dissolution rates are often 2–4 able elements in surrounding waterways (Younger, orders of magnitude faster than those measured 2000; Banks et al., 1997). This global problem is for minerals in the field (e.g. Schnoor, 1990; White exacerbated by the lack of an effective framework et al., 1996). for determination of legal liability, which means A number of factors have been proposed to that resources for remediation of contaminated account for this discrepancy. These include differ- abandoned sites are limited (Younger, 1997). Fortu- ing pH, grain size, temperature, hydrology, cation nately, pre-mining environmental assessments are exchange characteristics, availability of reactants now required in the majority of countries, and for such as O2, degree of physical and chemical heter- this, as well as for cost-effective remediation, it is ogeneity, secondary mineral behaviour, mineral necessary to develop the ability to predict the sever- surface characteristics and proximity to chemical ity and longevity of mining-related contamination saturation with respect to the dissolving minerals, on a site-specific basis. An understanding of the pro- between weathering environments (e.g. Sverdrup cesses affecting weathering is also relevant to cli- and Warfvinge, 1995; Malmstrom et al., 2000; matic modelling, where a better knowledge of White and Brantley, 2003). Algorithms that quan- field-scale mineral dissolution rates is necessary to titatively account for some of these factors have facilitate assessment of relationships between tem- been devised (e.g. Sverdrup and Warfvinge, 1995; perature, precipitation, elevation and weathering Malmstrom et al., 2000). These are based on the rate (e.g. White and Blum, 1995). premise that each of the different parameters that Laboratory assessments are the most conve- affect dissolution rates can be considered sepa- nient method of measuring the contamination rately, and a scaling factor calculated for each. potential of spoil. Detailed studies of the dissolu- The combination of the scaling factors then allows tion of individual mineral phases have determined extrapolation of laboratory-based weathering rates dissolution mechanisms and relationships between to the field. Calculations are based on fundamental dissolution rates and pH, temperature, surface relationships between reaction rate and the physi- and crystallographic characteristics, and rate- cal parameter of interest, and so site-specific cali- determining concentrations of reactants that influ- bration is not required for most parameters. ence kinetic mass action (e.g. .Wieland et al., Thus, the algorithms should prove to be robust 1988; Xie and Walther, 1992; Martello et al., and generally applicable. The algorithm of Malm- 1994; Peiffer and Stubert, 1999; Holmes and Crun- strom et al. (2000) has been demonstrated to be dwell, 2000), but it is difficult to use such work to successful in a study of the weathering of granitoid account for interactions between the components waste rock from the Aitik Mine, Sweden (Malm- of the phases in a polymineralic assemblage. Rudi- strom et al., 2000). mentary acid–base measurements on polymineralic However, application of this type of algorithm materials are popular (e.g. Adam et al., 1997), involves the implicit assumption that the same because of the relative ease and rapidity with mechanism determines the rate of element release which they can be undertaken and interpreted. in the laboratory and in the field, and that mineral Such methods measure the total potential of a sources and sinks play the same role in the two sample for contamination, but do not provide environments. This is not necessarily the case. The information on the rate at which contamination present study investigates rate-determining mecha- is produced, and are thus of limited use (e.g. Jam- nisms and mineral sources and sinks at four UK bor, 2000). Batch and column experiments have mine sites via a year-long mine-water monitoring also been used (van Grinsven and van Riemsdijk, program and bulk and mineralogical analysis of 1992; Stromberg and Banwart, 1999; Banwart samples from the sites. Results are related to those et al., 2002), to measure rates and to distinguish produced by laboratory experiments using results of rate-controlling mechanisms of mineral dissolu- hydrogeological analyses of the sites. Results for tion. Column experiments can be more useful than one site are then compared to column and batch batch experiments because they involve water:rock experiments on material from the site (Evans and ratios and hydrological solute transport processes Banwart, 2006). The implications of results for similar to those found in the field. Regardless of prediction and treatment of mine-water related method, however, it is difficult to extrapolate lab- pollution are discussed. K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 379

2. Field sites (Gandy, 2002) with associated publications (Gandy and Evans, 2002; Gandy and Younger, 2003), which Locations of all sites are shown in Fig. 1. The have developed groundwater flow and chemical criteria for field sites were that they should: be models for the site. hydrologically well defined with point discharges;have a well-documented mining history; 2.2. Grattendale exhibit commonly observed contaminant characteristics; be representative of the selected type of Limestone hosted Pb/Zn deposits in the valley of mining environment; have available data for rain- Grattendale in the Peak District, Derbyshire, Eng- fall, discharge, topology, and geology, and have a land (Fig. 1b), were mined up to the end of the reasonably simple geological structure. 18th Century. Grattendale is in the White Peak area, which comprises an inlier of Carboniferous 2.1. Quaking Houses limestone surrounded, except to the extreme SW, by younger fluvio-deltaic sandstones (Stevenson The spoil tip at Quaking Houses, near Newcastle- et al., 1982). Mississipi Valley Type (MVT) Pb upon-Tyne, County Durham, England (Fig. 1a), and Zn ores formed when cooling metal-rich fluids comprises spoil excavated during working of the passed sub-horizontally through the limestones Morrison Busty pit between 1922 and 1973. The (Ewbank et al., 1995). Ore minerals include galena, tip overlies sand and clay drift deposits, which them- sphalerite, baryte, and fluorite. Mineralisation was selves overlie Carboniferous interbedded mudstones, controlled by lithology and by pre-existing struc- sandstones and coal seams (Pritchard, 1997). The ture, and resulted in a geometry of near vertical or area of the tip is approximately 35 ha and the height horizontal sheets, known as rakes and flats respec- varies from 4.25 to 11 m. The spoil material is a het- tively, and occasional linear pipes (Ford and Rieuw- erogeneous mix of shale, ash, coal, and coal dust, erts, 1970). Mineralised zones rarely penetrate the with scattered cobbles, sandstone boulders, timber, volcanic horizons, known locally as toadstones, and traces of red burnt shale. The vast majority of and are thus found mainly within higher levels of particles are smaller than 5 cm diameter. Recently the formation (Ford and Rieuwerts, 1970). Gratten- formed secondary minerals are common in samples dale is a steep sided valley with walls of clean, from the upper 5 m of the tip. Hydrology of the tip fossiliferous (rugose, fragmented corals, and bra- is simplified by the underlying clay-rich drift deposits chiopods) limestone. The Matlock Lower Lava, which act as an aquitard, isolating the tip from known locally as the Grattendale Lava, outcrops groundwater flow (Pritchard, 1997). The discharge partway up the valley; where it is around 9 m thick at Stanley Burn (output G, OS grid ref: 41770 and dips to the NE (Oakham, 1979). Surface expres- 55095) is thought to collect water from approxi- sion of mineralisation is rare because most outcrop- mately 10% of the tip, plus relatively uncontami- ping deposits were extensively worked to a depth of nated water that flows from a surface drain 10–15 m and then filled with earth. (Pritchard, 1997). Mining in the White Peak since Roman times has Significant tip-related pollution was not recorded produced a poorly mapped network of largely inter- between closure of the mine in 1974, and 1986, connected mines and drainage channels, or soughs when construction of the A693 Annfield Plain (Rieuwerts, 1987). The resultant hydrology is com- bypass cut through almost the full depth and plex; however, Grattendale is effectively hydrologi- breadth of the tip. During construction, drains that cally bounded by topography, geology and soughs serviced the tip were incorporated into the road and so presents a relatively simple system. The dis- drainage system, the bulk of which discharges into charge (OS grid reference: 42083 36078) is continu- the nearby Stanley Burn. A subsequent decrease in ous year round, with a flow rate between 0.1 and 2 1 pH and increases in Fe, Al, and SO4 contents in 2Ls . Mining exposed large areas of fresh sul- the Stanley Burn have been attributed to drainage phide surfaces to relatively oxidising conditions. from the tip. There have been a number of previous Regional S oxidation rates higher than those that studies of water quality at the site. Local authority pertained before mining began are indicated by S reports (e.g, Markey-Amey, 1995; Newbegin, isotopes (Bottrell et al., 1999), and analysis of 1997) have been complemented by two M.Sc. theses sediments in cave systems shows fine grained sulp- (Pritchard, 1997; Srour, 1998) and a Ph.D. thesis hides with a relatively high potential for dissolution 380 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

(a) (b)

(c)

(d)

Fig. 1. Locations and details of the four field sites. Grid references are to the Ordnance Survey grid. (Bottrell et al., 2000). However, solubility of mineral low under high pH conditions such as those found phases bearing Pb, Zn, and Ba, the principal poten- in limestone-rich areas because formation of spar- tially toxic contaminants in the area, are relatively ingly soluble Pb and Zn carbonates and hydroxides K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 381 effectively removes most of the metals from solution River Dulais, the bed of which passes over the Bla- (Benevenuti et al., 2000). enant workings. However, the coal barriers between the Blaenant and Ynysarwed mines (Fig. 1c) proved 2.3. Ynysarwed ineffective and a substantial discharge began at the Ynysarwed adit in 1993 (Younger and Adams, The Ynysarwed minewater discharge flows from 1997). Water discharges from the adit portal at an old adit connected to the lower Ynysarwed Ynysarwed are up to 36 L s1. Discharges were ini- workings in the Neath Valley, South Wales tially acidic with a pH as low as 3, but are currently (Fig. 1c). The regional geology consists of faulted circumneutral. The discharge is reducing, and rich 2 1 sandstones, mudstones, siltstones and coals of the in dissolved Fe and SO4 (200 and 1800 mg L Upper and Middle Coal Measures (Westphalian), respectively), although a gradual decline in Fe con- which dip at approximately 6 SW (Barclay centrations since 1993 has been observed (Ranson et al., 1988). The topography comprises steep sided and Edwards, 1997). Precipitation of ochre from narrow valleys, separated by larger areas of rela- discharging water has adversely affected fish and tively flat upland. Upper Coal Measures fluvial invertebrate populations in the Neath Canal, which Pennant sandstones, interspersed with a number receives the discharge, and the discharge was given of small, largely unworked coal seams, overlie the 11th priority in a 1993 water quality survey of Rhondda number two seam, which is the principal British minewater discharges. This led to a signifi- worked seam in the area. Below the coal lie Lynfi cant amount of work on both the hydrochemistry Middle Coal Measures mudstone beds, which show and hydrology (e.g. Younger, 1996; Younger and a marked marine influence (Barclay et al., 1988). It Adams, 1997). is this marine influence which gives the anthracitic Rhondda number 2 coal its high S content (2–4% 2.4. Church Coombe pyritic S). The hydrogeology of the area is dominated by The Church Coombe minewater discharge, Corn- the highly conductive mine workings. The Ynysar- wall, England flows into a small stream (Fig. 1d). wed discharge (OS grid reference: 28094 20186) is The surrounding area is a geometrically complex fed by recharge through the relatively permeable mix of Upper Devonian Mylor slates (killas), and Pennant Sandstone (approximately 20% of precipi- Late Carboniferous/Early Permian granites which tation) and groundwater flows from the Blaenant form part of the Cornubian Batholith. These are system to the west (Younger and Adams, 1997). intruded by NNE-WSW striking coarse granite por- Underlying Lynfi mudstones are relatively imperme- phyries (elvans) and dolerites (greenstones) (Dines, able and are assumed to prevent downward escape 1956). Mineralisation was associated with circula- of water from the system. Flows from the neigh- tion of basinal and meteoric fluids driven by heat bouring unsaturated Upper Ynysarwed, Crynant, from the intrusion of the Cornubian batholith and Lwynon workings, which bound the area under (Gleeson et al., 2001). Mines near Church Coombe consideration to the north, are thought to be minor are located in a trough of slate between the Carn- as all drain to adits well above the groundwater mellis granite to the south, and the Carn Brea gran- table (C. Rees, per. comm, 1999; Younger and ite to the north. The main ore-bearing body dips at Adams, 1997). Piezometer measurements show that around 30 S in this region and follows the granite- groundwater flow is from NW to SE, that is, from slate contact in the area shown in Fig. 1d. A number Blaenant (water table 75 m AOD) to Ynysarwed of ENE-trending sub-vertical off-shoots from the (water table 20 m AOD). Prior to the cessation of main body were also worked in the area (Dines, mining, the Blaenant works were pumped, while 1956). Principal ore minerals are chalcopyrite and the lower Ynysarwed workings drained through cassiterite, with a tourmaline/chlorite gangue the discharging adit. Lower Ynysarwed was closed (peach). The sub-vertical lodes were worked mainly in 1938, and although an occasional ferruginous dis- for Cu, although Sn and As were also produced charge was noted at the adit this was not of any vol- (Dines, 1956). Mining was well established in the umetric significance. The Blaenant mines closed in area by 1699, and by 1743 the adit was being used 1991, the pumps were turned off, and groundwater to produce water power (Hamilton-Jenkin, 1965). levels within the workings began to rise. Minewater Mining flourished until the 19th Century, but discharges were expected to begin at or near the underground activity had effectively ceased by 382 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

1919. The dumps were reworked for secondary and field, to quantify rarer elements such as Cd, U, accessory minerals in the 1930s (Dines, 1956). and Th. Significant concentrations of these elements Exploration and mapping of the adit (portal at were not found. Iron-rich samples from Ynysarwed OS grid reference: 16916 04076) suggest that it were not analysed by IC because Fe forms insoluble drains the Wheal Bassett system, which includes compounds inside the ion exchange columns, and, 2 North Bassett and Wheal Bassett (Fig. 1d). Flow in any case, it was known that SO4 is the dominant rates are up to 90 L s1. The region is classed as a anion (Younger, 1997). pH, redox potential, con- minor aquifer, with permeability controlled largely ductivity and temperature were measured with a by the depth of weathering, rather than by the Myron 6P Ultrameter. Alkalinity was measured by hydrological properties of the slate or the granite. titration with 0.035 N H2SO4 to pH 4.5 using bro- The water is circumneutral and contains 1 mg L1 mocresol green/methyl orange indicator. Flow rate of Cu all year round, and thus potentially poses was also measured at each sampling site. At Quak- some threat to receiving water courses. However, ing Houses and Grattendale this was accomplished this is mitigated by subsequent dilution. using a bucket and stopwatch. At Ynysarwed, flow rates were measured electronically with stream 3. Methods gauging by the Environment Agency. These measurements were affected by the construction of a 3.1. Element release mechanisms treatment plant for the minewater effluent and thus results from this site are subject to additional 3.1.1. Water sampling and analysis sources of uncertainty, discussed below. At Church Sites were monitored monthly (Table 1) for a per- Coombe, flow rates were calculated from the veloc- iod of at least 12 months between June 1999 and ity of flow in a channel of known cross-sectional June 2001. Quaking Houses samples were taken area. from outfall G (Fig. 1a), Grattendale samples from Wraithe Sough (Fig. 1b), Ynysarwed samples from 3.1.2. Interpretation the marked adit portal (Fig. 1c), and Church Coo- Relationships between element concentrations mbe samples from the adit portal marked in Fig. 1d. (moles L1), discharge flow rates (L s1) and ele- At Quaking Houses, unfiltered acidified (1% ment fluxes (mole s1; obtained by multiplying con- HNO3) and non-acidified samples were analysed centration and discharge flow rate) were used to by atomic absorption spectroscopy (AAS) and ion determine the rate-controlling mechanism for each chromotography (IC) for metals and common element. Rate-controlling mechanisms were split 2 anions (F ,Cl,NO2 ,Br,NO3 ,SO4 ) at the into those that involved principally (a) surface Department of Civil Engineering, University of kinetic-, (b) mineral equilibrium-, and (c) trans- Newcastle. pH, alkalinity, Eh and conductivity were port-related factors. determined in the field using portable meters. At Grattendale, Ynysarwed and Church Coombe, 3.1.2.1. Surface kinetic (S). If surface kinetic pro- water samples were filtered through 0.2 lm filters cesses control the rate of mineral dissolution then (Schleicher and Schuell) into new Nalgene bottles the flux of elements sourced from the dissolving which were rinsed in the discharge before use. Acid- mineral is constant, so long as other rate-determin- ified (1% HNO3) and unacidified samples were col- ing factors such as the quantity of source mineral, lected at each sampling visit. Acidified samples pH, temperature and the concentration of reactants were analysed by inductively coupled plasma- remain constant (e.g. Wieland et al., 1988). If pro- atomic emission spectroscopy (ICP-AES) for major cesses such as secondary mineral precipitation do and abundant trace elements at the Assay Office, not interfere then the flux of the element from the Sheffield. Unacidified samples were analysed, in mine should also be constant. Clearly, temperature, most cases, by Ion Chromatography (IC: Dionex) pH, and other rate-controlling factors are likely to for light alkali metals and alkaline earths (Na+, change in a mine over the course of a year and so K+,Ca2+,Mg2+), and common anions (F,Cl, some variation in surface kinetic-controlled element 3 2 NO3 ,NO2 ,Br ,PO4 ,SO4 ). One or two samples fluxes would be expected. An approximately con- from each site were screened using inductively cou- stant element flux was thus used to infer that the pled plasma-mass spectroscopy (ICP/MS) at the principal mineral source or sources for that element Centre for Analytical Services, University of Shef- were dissolving at a rate controlled by surface-kinetic Table 1a Quaking Houses discharge data Conc. (mg L1 ) 15/10/98 19/11/98 17/12/98 26/1/98 23/2/99 23/3/99 23/4/99 14/5/99 3/6/99 2/7/99 23/7/99 30/7/99 16/9/99 8/10/99 Al 3.23 5.36 5.48 5.13 8.49 7.734 3.854 2.85 4.091 7.334 8.5 10.18 5.79 2.24 Ca 193.2 171.4 362.2 381.9 182 263.7 168.1 144.8 196.5 257 314 Fe 6.61 3.73 5.24 4.92 11.46 2.486 2.851 2.706 4.39 7.5 9.41 9.89 16.95 7.14 K 100 140.6 294 297.2 54 106.6 164.2 130.2 188.3 327.3 36.5 Mg 52.1 54.4 126.1 120.9 46.4 82.3 51.6 38.8 65 100.6 98 Mn 3.16 3.56 3.72 3.18 2.37 3.786 0.948 1.301 2.552 4.127 5.6 6.18 4.95 4.15 Na 840 1074 2493 2572 566 1148 1263 792 1149 1922.1 2163

Zn 0.84 1.6 1.14 1.53 1.09 2.405 0.948 0.575 1.01 1.391 0.91 377–403 (2006) 21 Geochemistry Applied / al. et Evans K.A. Cl 2236 3132 2049 1691 2062 440 1851 3047 2621 2446 1581 3939 2 SO4 565 552 544 921 706 404 242 512 873 1227 1257 881 910 pH 6.46 6.37 6.58 6.48 5.7 6.09 6.14 7.1 6.45 5.45 5.22 4.2 6.2 5.73 Conductivity (lS) 5765 7672 7555 7130 7320 7138 8010 4787 6880 11,090 10,140 8760 5970 12,190 1 Alk. (mg L as CaCO3) 77 67 54 27 58 107 70 30 <5 0 31 <5 Flow (L min1 ) 55.2 107.7 71.2 76.3 39.7 89.3 110.5 168 76.36 131.2 20.5 23.8 28.6

Table 1b Wraithe Sough discharge data Conc. (mg L1 ) 14/10/99 19/11/99 02/12/99 19/01/00 02/02/00 16/02/00 01/03/00 30/03/00 04/05/01 07/06/00 18/08/00 19/09/00 28/10/00 Ca 92.40 92.60 80.20 78.89 98.23 95.23 91.23 70.25 67.74 76.18 91.38 96.65 82.29 K 0.80 1.00 0.74 0.47 1.25 1.08 1.19 0.67 0.65 0.26 0.56 0.63 0.99 Mg 34.10 31.60 26.04 24.64 43.67 41.53 39.39 27.32 26.03 28.84 29.24 31.89 32.09 Na 5.85 6.66 5.00 5.03 7.54 7.22 7.46 5.51 5.63 4.70 7.14 7.24 5.37 Si 2.58 2.32 2.41 2.24 3.02 4.24 3.73 2.37 2.29 2.00 2.51 2.73 2.52 Pb 0.01 0.02 0.06 0.01 0.04 0.02 0.01 0.00 0.00 0.00 0.08 0.00 0.00 Sr 0.03 0.03 0.04 0.04 0.04 0.03 0.03 0.03 0.03 0.04 0.03 Zn 0.20 0.16 0.06 0.18 0.15 0.11 0.11 0.15 0.15 0.23 0.15 0.18 0.20 Cl 14.42 13.38 14.05 15.19 15.71 14.87 15.65 11.04 11.16 11.04 11.21 11.14 11.64 F 1.87 1.55 1.65 1.77 1.81 1.65 1.91 1.67 1.68 1.58 1.60 1.58 1.65 NO3 23.83 25.58 26.90 35.41 28.66 27.20 29.23 30.14 25.73 25.26 27.37 27.22 2 SO4 24.41 22.79 25.23 24.96 24.80 24.78 25.04 19.46 19.35 19.87 22.01 23.65 22.45 pH 8.12 7.30 7.57 7.31 8.19 7.37 6.97 7.89 7.87 7.69 7.84 8.08 7.60 Conductivity (lS) 623.30 612.00 597.20 604.00 606.00 606.00 574.00 605.00 588.00 598.00 621.50 629.00 597.00 Alkalinity (mg L1 ) 270.00 168.00 192.00 192.00 161.00 161.00 176.00 184.00 200.00 200.00 330.00 250.00 Temp. (C) 9.30 7.80 8.90 8.80 8.80 8.40 9.40 8.90 9.20 9.40 10.00 9.30 9.50

ORP (mV) 134 205 178 205 16 246 153 91 109 n/a 97 521 101 383 Flow rate (ml min1 ) 142 205 142 563 889 800 1720 875 950 845 290 200 548 384 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

Table 1c Ynysarwed discharge data Conc. 17/08/99 20/09/99 13/10/99 15/11/99 10/12/99 26/01/00 22/02/00 28/03/00 26/04/00 16/05/00 28/06/00 (mg L1) Al <0.05 0.0 <0.01 0.2 0.2 0.2 0.8 0.7 0.8 0.6 0.6 Ca 285.0 314.0 313.0 300.0 239.8 241.4 289.4 238.7 226.9 288.8 241.0 Fe 173.0 170.0 169.0 163.0 150.6 147.6 154.8 138.1 131.8 159.5 139.8 K 23.1 24.9 25.2 25.5 23.3 24.3 29.1 26.8 25.8 30.2 26.9 Mg 156.0 170.0 171.0 154.0 119.6 120.5 176.5 121.8 115.0 144.6 119.2 Mn 3.9 3.9 0.9 3.7 3.6 3.5 4.0 3.7 3.5 4.1 3.4 Na 108.0 115.0 159.0 152.0 120.0 123.2 171.3 155.6 146.9 141.3 118.6 Si 7.5 7.1 7.2 7.8 7.1 7.3 8.9 8.3 28.9 7.1 6.2 As <0.01 0.1 0.1 0.1 0.1 0.1 0.2 <0.01 <0.01 <0.01 0.1 Ni 0.2 0.2 0.2 3.7 0.2 0.2 0.3 0.4 0.7 0.2 0.2 Pb 0.1 <0.1 <0.01 0.1 0.1 <0.1 0.2 0.1 0.1 <0.1 <0.1 Sr <0.01 2.2 <0.01 1.9 1.2 1.2 1.3 1.3 1.2 1.0 0.9 Zn <0.1 <0.1 0.1 <0.1 <0.1 <0.1 <0.1 <0.1 0.6 0.1 <0.1 2 SO4 1839.0 1950.0 1830.0 1980.0 1589.2 1595.9 2055.9 1911.6 1811.7 1520.3 1346.1 pH 6.44 6.47 6.38 6.20 6.08 6.37 6.46 6.35 5.98 6.28 6.36 Conductivity 2859 2763 2750 2800 2573 2709 2619 2642 2539 2583 2517 (lS) Alkalinity 100 160 180 98 112 112 112 96 112 128 140 (mg L1) Flow rate 26.31 24.07 25.97 26.74 29.09 28.43 29.05 28.25 26.52 28.37 27.21 (L min1) Temp. (C) 13 15 15 11 14 11 12 12 13 15 15 ORP (mV) 56 31 31 5 44 28 44 116 0 n.a. 54

processes. The definition of constant flux used here an element are likely to remain constant over the was that of a standard deviation of the monthly flux course of a year at any of the field sites studied. measurements less than 20% of the average value. Thus, an approximately constant concentration of The value of 20% is arbitrary, but choices of values an element was taken to infer equilibrium control. for the constant flux threshold between 15% and Constant concentration was diagnosed when the 40% give very similar results. Thresholds for the standard deviation of the monthly measurements other categories described below are also arbitrary, was less than 20% of the average value, and where and results show a similar weak sensitivity to the a suitable source/sink mineral was observed, and choice of value. calculated to be in equilibrium with the solution (see below). The second criterion is included because 3.1.2.2. Equilibrium (E). If the concentration of an a constant concentration in a mine discharge may element is controlled by chemical saturation with also result from constant concentration in some respect to a mineral then the concentration of that external source. element is fixed by the presence of the mineral, so long as factors such as pH, temperature, the chem- 3.1.2.3. Transport (T). Transport control of element ical activity of other aqueous species, and the activ- release occurs when changes in hydrogeological ity of the mineral phase remain constant. A solution parameters that affect flow rate, such as the rate of may be chemically saturated with respect to a source recharge, the percentage of water saturation in the mineral present in rock prior to weathering or to unsaturated zone, or water table height, determine some secondary mineral that incorporates the ele- the rate of mineral dissolution and element trans- ment of interest. Cation exchange surfaces on clay port. The transport control described here is similar minerals also exert some control on the concentra- to the transport-limited erosion described by Stal- tions of elements in solution, although the systemat- lard and Eddmond (1987). However, it is slightly ics of this are more complicated. It is unlikely that different to the process described by these authors factors that affect the equilibrium concentration of because it does not include situations in which Table 1d Church Coombe discharge data Conc. 15/08/00 13/09/00 26/10/00 29/11/00 18/12/00 30/01/01 27/02/01 29/03/01 25/04/01 31/05/01 26/06/01 25/07/01 377–403 (2006) 21 Geochemistry Applied / al. et Evans K.A. (mg L1 ) Al 0.41 0.73 0.45 0.41 0.38 0.39 0.44 0.39 0.39 0.38 0.38 0.38 Ca 8.13 11.94 11.47 9.97 10.96 10.39 10.05 11.01 10.43 8.69 9.38 9.49 Fe 0.32 0.04 0.01 0.01 0.02 0.01 0.02 0.03 0.70 3.78 1.06 0.72 K 2.76 4.34 3.96 0.56 3.31 3.28 3.14 3.59 3.26 13.88 11.76 5.96 Mg 4.92 5.84 5.57 4.05 4.93 5.02 4.76 4.39 4.17 4.65 5.03 5.16 Mn 0.05 0.08 0.06 0.05 0.05 0.05 0.06 0.05 0.06 0.04 0.04 0.05 Na 13.24 24.26 18.28 18.96 19.80 17.89 18.78 19.29 17.19 19.41 21.33 20.63 Si 3.23 4.83 4.06 3.39 3.60 3.66 4.28 3.64 3.63 4.19 4.42 4.46 Cu 1.47 2.09 1.28 1.11 1.08 1.20 1.36 1.14 1.27 1.57 1.67 1.74 Zn 0.23 0.30 0.22 0.18 0.17 0.19 0.33 0.19 0.22 0.23 0.26 0.26 Cl 27.57 28.13 28.58 27.00 27.27 26.52 24.61 24.30 23.69 25.79 24.47 25.18 F 0.86 0.86 0.74 0.59 0.64 0.59 0.61 0.60 0.69 0.79 0.83 0.81 NO3 12.06 11.76 13.28 13.55 13.01 13.91 14.89 13.94 12.44 13.08 n.a. 12.44 2 SO4 26.74 28.41 28.60 26.65 26.73 28.15 25.04 27.17 23.77 26.08 29.25 32.99 pH 5.6 6 5.7 5.4 6 5.6 5.6 6 6.1 5.3 6.2 5.44 Conductivity (lS) 200 190 210 210 240 230 250 240 270 220 210 220 Alkalinity (mg L1 ) 11.8 15 15 10 10 10 15 10 10 10 10 10 Flow rate (L s1 ) 14 2 40 60 45 90 60 50 40 14 16 12 Temp. (C) 12 11.7 12.3 12 12 12 12 12.1 12.2 12.2 12.5 12.6 ORP (mV) 186 220 215 235 253 261 244 247 235 215 198 200 385 386 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 mineral equilibrium is approached, although 3.2. Mineralogy transport-related processes are implicated in the attainment of mineral equilibrium. Mechanisms 3.2.1. Sample collection for transport-controlled element release include: Twenty kilograms of each of the main lithologies flushing, where an increase in the height of the water was collected from each site. Representative sam- table dissolves soluble secondary minerals precipi- pling of 20 kg of material from heterogeneous sys- tated in previously hydrologically-unsaturated parts tems containing many tons of material is not of a hydrogeological system; dilution, where remotely achievable. However, the samples can be increased water flow decreases element concentra- used to identify source and sink minerals present tions and may enhance the rate of mineral dissolu- at the sites. tion; and changes in the ratio of mobile to At Quaking Houses samples were taken from immobile water in the unsaturated zone (Evans material retrieved during borehole construction; and Banwart, 2006; Banwart et al., 2004), which and consisted of two types, black coal waste affect the efficiency of removal of dissolved mineral (QB), and orange clay-rich material from boulder constituents. Physical processes that accompany clay underlying the tip (QO). QB was made by transport control are complex (Clow and Drever, mixing approximately equal proportions of spoil 1996) and so the flux–flow characteristics of trans- taken at 1 m intervals down the tip. Weathering port control are variable; however, all are likely to of the top portion of the tip has caused significant result in a correlation between flow rate and element mineralogical variation with depth (Evans et al., release that could be positive (flushing) or negative 2003). Ore-bearing limestone from Grattendale (dilution). Transport control of element release was collected at a spoil tip 5 km from the adit was inferred when the absolute value of the correla- portal, and non-ore bearing limestone was col- tion coefficient between flow rate and element con- lected from Grattendale itself. All samples were centration was greater than 0.7. large (1–5 kg), partly weathered limestone boulders. Material from Ynysarwed was collected from 3.1.2.4. Depletion (D). It may be that the supply of a spoil tip that is known to contain material from a mineral may be depleted by dissolution, and that both Blaenant and Ynysarwed mines (C. Rees, this process is the main control on the flux of con- pers. comm.). Samples were almost unweathered stituent elements of that mineral from the system because recent landscaping of the area had of interest. If this is the case then the flux of this ele- exposed relatively fresh spoil at the surface. ment from the system will decrease with time. Twenty kilograms of each of pyrite-bearing mud- Depletion control of an element was inferred when stone (YM), clean, mature sandstone (YS) and the absolute value of the correlation coefficient pyrite-bearing coal (YC), in 1–5 kg lumps, were between element flux and time was less than 0.7. collected, to allow differences in mineral dissolution rates and composition between lithologies to 3.1.2.5. General transport (I). The categories be determined. At Church Coombe only one set defined above are end-members in a continuum of material (CC) was collected from a spoil tip of system behaviour that is created by the play- adjacent to the North Basset Mine. Here, 1–3 kg off between mineral reactivity, fluid flow rates, samples of a reddish-brown hornfelsed mudstone, and other hydrological parameters. There are a showing evidence of mineralisation, were taken number of situations for which none of the above from a freshly turned spoil tip. criteria would be met, which implies that the rate- controlling factor for element release varies over the course of a year. The rate-controlling element 3.2.2. Analysis release mechanism in these cases cannot be deter- Samples were jaw-crushed to a median grain size mined by a simple inspection of element flow and of 1 mm and homogenised in an industrial mixer flux characteristics and is classed as general trans- at the Department of Civil Engineering, Sheffield. port (I). This preparation was necessary for the laboratory Figs. 2–5 illustrate examples of the relationships experiments (Evans and Banwart, 2006). Major between element concentrations, fluxes, time, and and trace element compositions for each of the eight discharge flow rate. Inferred rate-controlling element lithologies were determined at the British Geologi- release mechanisms are summarised in Table 2. cal Survey, Keyworth, using X-ray fluorescence K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 387

(a) 9 (b) 1400 T 8 I 1200

) 7 ) 1000 -1 -1

6 800 5 600 4 400 3 Concentration (mg L Concentration Concentration (mg L Concentration 200 2 2- QuakingHouses: Al Quaking House: SO4 1 0 Oct Feb Jun Oct Oct Feb Jun Oct Date Date

) 0.20 ) -1 -1 24

23 0.15 22

21 0.10 Concentration (mg L Concentration Concentration (mg L Concentration 20 Grattendale: Zn Grattendale: SO42- 0.05 19 Oct Feb Jun Oct Oct Feb Jun Oct Date Date

(e) 4.5 (f) 180 E D 4.0 170 ) ) 3.5 -1 -1 160 3.0

2.5 150

2.0 140 1.5 Concentration (mg L Concentration

Concentration (mg L Concentration 130 1.0 Ynysarwedd: Mn Ynysarwedd: Fe 0.5 120 Aug Nov Feb May Aug Nov Feb May Date Date

(g) 2.2 (h) 15.5 Church Coombes: Cu T T 15.0 2.0

) ) 14.5 -1 -1 1.8 14.0

1.6 13.5

13.0 1.4 12.5 Concentration (mg L Concentration Concentration (mg L Concentration 1.2 12.0 - Church Coombes: NO3 1.0 11.5 Aug Nov Feb May Aug Aug Nov Feb May Aug Date Date

2 Fig. 2. Concentration versus time for representative elements from the four ﬁeld sites. (a) Quaking Houses: Al; (b) Quaking Houses: SO4 ; 2 (c) Grattendale: Zn; (d) Grattendale: SO4 ; (e) Ynysarwedd: Mn; (f) Ynysarwedd: Fe; (g) Church Coombes: Cu; (h) Church Coombe: NO3 . Shaded bands indicate the region of ±20% around the average value. 388 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

(a) 9 (b) 1400 IT 8 1200 ) ) -1 -1 7 1000 6 800 5 600 4

Concentration (mg L Concentration 400

Concentration (mg L Concentration 3

2 200 Quaking Houses: Al Quaking Houses: SO42- 1 0 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 Flow Rate (L min-1) Flow Rate (L min-1)

-1 0.20 -1 24

0.15 22 Concentration (mg L Concentration 0.10 (mg L Concentration 20

Grattendale: Zn 0.05 18 0 200 400 600 800 1000 1200 1400 1600 1800 0 200 400 600 800 1000 1200 1400 1600 1800 Flow Rate (ml min-1) Flow Rate (ml min-1)

(e) 4.5 (f) 180 E D 4.0 170 ) ) -1

3.5 -1 160 3.0

2.5 150

2.0 140

Concentration (mg L Concentration 1.5 Concentration (mg L Concentration 130 1.0 Ynysarwedd: Mn Ynysarwedd: Fe 0.5 120 23 24 25 26 27 28 29 30 23 24 25 26 27 28 29 30 Flow Rate (L min-1) Flow Rate (L min-1)

(g) 2.2 16 (h) T T Church Coombes: Cu 2.0 ) ) 15 -1 -1 1.8 14 1.6 13 1.4 Concentration (mg L Concentration Concentration (mg L Concentration 12 1.2 - Church Coombes: NO3 1.0 11 0 20406080100 0 20406080100 Flow Rate (L s-1) Flow Rate (L s-1)

Fig. 3. Concentration versus ﬂow rate for representative elements from the four ﬁeld sites. (a) Quaking Houses: Al; (b) Quaking Houses: 2 2 SO4 ; (c) Grattendale: Zn; (d) Grattendale: SO4 ; (e) Ynysarwedd: Mn; (f) Ynysarwedd: Fe; (g) Church Coombes: Cu; (h) Church Coombe: NO3 . K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 389

(a) 1200 (b) 1.2e+5 Quaking Houses: Al I Quaking Houses: SO42- T 1000 1.0e+5 r = 0.20 r = -0.11 ) )

800 -1

-1 8.0e+4

600 6.0e+4 Flux (mg min

Flux (mg min 400 4.0e+4

200 2.0e+4

0 0.0 Oct Feb Jun Oct Oct Feb Jun Oct Date Date

6 1500

4 1000 Flux (mg min Flux (mg min 2 r = 0.21 500 0 Grattendale: Zn 0 Oct Feb Jun Oct Oct Feb Jun Oct Date Date

(e) 8000 (f) 2.8e+5 E I 7000 2.7e+5 2.6e+5 6000 ) ) -1

-1 2.5e+5 5000 2.4e+5 4000 2.3e+5 Flux (mg min Flux (mg min 3000 r = 0.11 2.2e+5 r = --0.51 2000 2.1e+5 Ynysarwedd: Mn Ynysarwedd: Fe 1000 2.0e+5 Aug Nov Feb May Aug Nov Feb May Date Date

(g) 7000 (h) 80000

6000 T T 60000 5000 ) ) -1 4000 -1 40000 3000 20000

Flux (mg min 2000 Flux (mg min 1000 r = 0.05 r = 0.00 0 0 - Church Coombes: Cu Church Coombes: NO3

Aug Nov Feb May Aug Aug Nov Feb May Aug Date Date

2 Fig. 4. Flux versus time for representative elements from the four ﬁeld sites. (a) Quaking Houses: Al; (b) Quaking Houses: SO4 ; 2 (c) Grattendale: Zn; (d) Grattendale: SO4 ; (e) Ynysarwedd: Mn; (f) Ynysarwedd: Fe; (g) Church Coombes: Cu; (h) Church Coombe: NO3 . Shaded bands indicate the region of ±20% around the average value. 390 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

(a) 1200 (b) 1.2e+5 I T 1000 1.0e+5 ) )

-1 800 8.0e+4 -1

600 6.0e+4 Y Data

Flux (mg min 400 4.0e+4

200 2.0e+4 Quaking Houses: Al Quaking Houses: SO42- 0 0.0 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 Flow Rate (L min-1) Flow Rate (L min-1) (c) 14 (d) 3000

12 I E 2500 10 ) ) Flux (mg min -1 -1 2000 8

6 1500

4 Flux (mg min 1000 2 500 0 Grattendale: Zn Grattendale: SO42- 0 0 200 400 600 800 1000 1200 1400 1600 1800 0 200 400 600 800 1000 1200 1400 1600 1800 Flow Rate (ml min-1) Flow Rate (ml min-1) (e) 8000 (f) 2.8e+5 E D 7000 2.7e+5 2.6e+5 6000 ) Flux (mg min )

-1 -1 2.5e+5 5000 2.4e+5 4000 2.3e+5 Flux (mg min 3000 Flux (mg min 2.2e+5

2000 2.1e+5 Ynysarwedd: Mn Ynysarwedd: Fe 1000 2.0e+5 23 24 25 26 27 28 29 30 23 24 25 26 27 28 29 30 Flow Rate (L min-1) Flow Rate (L min-1)

(g) 7000 (h) 80000 T 6000 T 60000

) 5000 ) -1 -1 4000 40000 3000 Flux (mg min Flux (mg min 2000 20000 1000 - Church Coombes: Cu Church Coombes: NO3 0 0 0 20 40 60 80 100 0 20406080100 Flow Rate (L s-1) Flow Rate (L s-1)

2 Fig. 5. Flux versus ﬂow rate for representative elements from the four ﬁeld sites. (a) Quaking Houses: Al; (b) Quaking Houses: SO4 ; 2 (c) Grattendale: Zn; (d) Grattendale: SO4 ; (e) Ynysarwedd: Mn; (f) Ynysarwedd: Fe; (g) Church Coombe: Cu; (h) Church Coombe: NO3 . K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 391

Table 2 Element release mechanisms Field Si Al Fe Mg Ca Mn Na K Zn Cu Pb As S N Cl F QH I T I I T I I I T I DI EE EII I EIIE Y IIIEEEEEI IIE CC E I E I E I E I I T E D I T I: general transport; T: transport control; E: equilibrium control; D: depletion control.

Table 3 Composition of samples QO QB Da WYSYMYCb CC Major elements (wt.%)

SiO2 66.59 34.83 1.3 3.22 82.76 44.89 2.76 75.28 TiO2 0.88 0.72 0.01 <0.01 0.37 0.85 nd 0.62 Al2O3 16.04 20.3 0.14 0.13 7.65 22.17 1.83 11.85 Fe2O3 5.9 5.01 0.35 0.15 2.46 4.61 13.61 5.22 Mn3O4 0.05 <0.01 0.11 0.05 0.05 0.11 0.08 0.52 MgO 1.07 0.5 0.13 0.83 0.59 0.95 0.78 0.88 CaO 0.07 0.2 37.9 53.7 0.34 0.43 1.59 0.03

Na2O 0.89 0.19 0.13 <0.05 0.71 0.47 <0.05 0.08 K2O 2.72 2.04 0.03 <0.05 1.63 3.65 <0.05 2.50 P2O5 0.13 0.05 <0.01 0.02 0.08 0.29 nd 0.01 SO3 <0.1 <0.1 10 <0.1 <0.1 <0.1 nd nd Cr2O3 0.02 0.02 <0.01 <0.01 0.01 0.02 nd nd SrO 0.01 <0.01 0.21 0.04 <0.1 0.02 nd nd

ZrO2 0.03 <0.02 <0.02 <0.02 <0.02 0.02 nd nd BaO 0.07 0.07 14.35 0.09 0.03 0.08 nd nd NiO <0.01 <0.01 <0.01 <0.01 <0.01 0.01 <0.05 nd CuO <0.01 <0.01 0.01 <0.01 <0.01 0.01 <0.05 nd ZnO <0.01 <0.01 7.28 0.1 <0.01 <0.01 nd nd PbO <0.01 <0.01 0.18 <0.01 <0.01 <0.01 nd nd LOI 4.89 35.48 24.41 41.93 2.82 21.05 75.34 3.78 Total 99.36 99.41 96.54 100.26 99.5 99.63 95.99 101.97

Trace elements (ppm) Ni 38 42 31 4 24 52 nd 35 Cu 28 129 164 7 16 125 nd <150 Zn 108 114 >2000 819 73 62 nd 232 As 6 131 <1 2 10 13 nd 56 Sr nd nd >1500 348 nd nd nd 18 Ba nd nd >5000 367 nd nd nd 361 Pb 18 52 >1000 88 8 36 nd 70 WSS (% S) 0.04 1.28 0.00 0.00 0.00 0.04 0.21 0.00 ASS (% S) 0.03 0.24 0.22 0.00 0.04 0.21 0.01 Total S (wt.%) 1.90 2.86 5.36 0.06 0.31 0.18 8.4 0.03 BET (m2 g1) 8.66 2.86 0.47 0.8 0.99 1.05 0.52 6.7 nd: not analysed. a Low totals for D result from incomplete oxidation of sulphides during ashing process and/or the presence of ﬂuorine-bearing minerals. b YC was digested in aqua-regia before analysis of solution by ICP-AES. All other samples by XRF (bead and pellet).

(XRF) on glass beads and pellets respectively at the Department of Materials Engineering, Shef- (Table 3). Mineral assemblages and textural rela- field University, and optical microscope examina- tionships were identified using X-ray diffraction tion of thin sections. Total S was determined by (XRD) and scanning electron microscopy (SEM) LECO combustion analysis at the Assay Office, 392 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

2 Sheffield, and water- and acid-soluble SO4 were poorly constrained for minerals that exhibit signif- obtained using standard methods (Czerewko, icant compositional variation. This is because the 1997). Disulphide-S was calculated from the differ- fine-grained nature of the samples made representa- ence between total S and the sum of acid plus water tive mineral compositions difficult to quantify using 2 soluble SO4 . Organic S was assumed to be negligi- techniques such as electron microprobe on single ble (Taylor, 1989). The specific surface area of the grains. For this reason, important mineral compo- crushed samples was measured by BET analysis at sition variables, such as Tschermak’s substitution the Department of Engineering Materials, Univer- in illite, were accounted for by specifying the sity of Sheffield. exchange component to be an unknown in the set of equations: for example, the anorthite component 3.2.3. Mineral modes in albite was specified as CaAlNa1Si1 (notation Mineral modes (Table 4) were calculated for the of Thompson, 1982). The use of exchange compo- observed mineral assemblage by solving the set of nents is discussed further by Thompson (1982). simultaneous equations that relates bulk composi- Water-soluble, acid-soluble and disulphide-S were tion to mineral modes and compositions (c.f. included as compositional constraints. This allowed Ferry, 1988). This set consists of i equations of a maximum of 10 variables, including mineral the form: modes and composition variables, to be deter- X mined. The plausibility of results was checked by N ¼ n p ; ð1Þ i i;j j a comparison of the mass of the calculated assem- j blage to measured mass. Ideally, the assemblage where Ni is the number of moles of element i per calculated from the analysis of 100 g of material unit mass, ni,j is the number of moles of element i should weigh 100 g, and incorrect choices of assem- in mineral j and pj is the number of moles of min- blage or mineral formulae will result in deviation eral j per unit mass. Mineral compositions are from this value.

Table 4 Mineral modes (Weight %) QO QB D W YS YM YC CC Albite 6.2 1 Illite 40 31 0.07 20 51 0.8 30 Kaolinite 34 8 20 3.3 4 Montmorillonite 7.7 6.9 Pyrite 0.2 3.9 0.6 0.2 15 0.04 Quartz 41 11 1.2 6.2 65 15 1 58 Galena 2.9 Barytes 22 0.1 Sphalerite 7.0 Calcite 66 89 0.6 Dolomite 0.6 7 16 Siderite 7.4 4.2 Alunite 0.25 0.04

Fe(OH)3 5.0 10 6 Gibbsite 0.6 Gypsum 0.13 0.01 0.01 0.01 0.96 0.9 0 Jarosite 2.6 0.38 1.7 0

Pb(OH)2/CO3 3.9 Zn(OH)2/CO3 1.8 Total X K,ill (set) 0.6 0.6 0.6 0.7 0.7 0.7 0.7 X Mg,ill 0.23 0.08 0.8 0.2 0.2 X Na,mont 0.99 0.58 X an 0.004 0.1 0.15 Total (g) 97 77 101 104 98 85 42 100 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 393

3.2.4. Identification of source and sink minerals ion balance of ±10% were used to calculate mineral If a solution is saturated with respect to a mineral saturation indices. then the saturation index of that mineral in solution is 0. The saturation index is the base 10 logarithm of 3.3. Area-normalised element fluxes the observed Ion Activity Product of the stoichiometric mineral dissolution reaction of interest divided Comparison between laboratory and field ele- by the conditional thermodynamic constant for the ment release rates requires some methods of conver- corresponding stoichiometric reaction for mineral sion between the element fluxes from field sites solubility (e.g. Appelo and Postma, 1993). A posi- (M T1), and the element fluxes per unit area of tive saturation index indicates that the solution is reacting mineral surface obtained from laboratory supersaturated with respect to the mineral phase, a experiment (M L2 T1). Element fluxes per unit negative value, that the solution is undersaturated, area are related to element fluxes via and a value near zero suggests that saturation is Qi reached, and that equilibrium exists between the F i ¼ ; ð2Þ mineral and the aqueous solution. Mineral satura- V qSSA tion indices for minerals observed by XRD, SEM where Fi is the flux of element i in units of 2 1 or thin section observation (Table 5) were calculated moles m s , Qi is the flux of element i calculated using PhreeqC for Windows, v1.0 (Parkhurst and from field observations (moles s1), V is the volume Appelo, 1999) utilising the PhreeqC database (Ball of interacting rock in m3, q is the bulk density in and Nordstrom, 1991). This was used to test diag- kg m3, and SSA is the reacting surface area in m2 noses of equilibrium-controlled element release kg1. Uncertainties in the values of V and SSA rates (Table 2). Only solutions with a calculated are significant. This is because V is controlled partly by poorly known effects of hydrological focussing such as fracture flow, and because the SSA of samples in the field is less than that of the crushed Table 5 material for which BET analyses are available. Mineral saturation indices SSA, which is sensitive to both the freshness of QD Y CCthe mineral surface and to grain size, is difficult to Albite S U quantify, even for laboratory samples. Chlorite U U Here, three different approaches are considered. Illite S+ S A maximum value for the SSA, and thus a mini- Kaolinite S+ S+ mum value for the area-normalised flux, is given Montmorillonite S+ S Pyrite U U U U by the use of BET-derived SSA (SSABET). This flux Quartz S S S is described and tabulated (Tables 6 and 7)asFB. Chalcopyrite U Alternatively, SSABET can be scaled using assump- Galena U U tions of particle size variation between laboratory Sphalerite U U U and field. The SSA decreases by an order of magni- Alunite S S+ S Fe(OH)3 S S S tude for each order of magnitude increase in particle Gibbsite S S S radius. Flux estimates derived by this method are Gypsum S U S S referred to here as adjusted BET-derived fluxes Jarosite S U U (FABET: not tabulated). However, this method does Calcite U S S S not account for the increase in mineral surface reac- Dolomite U S S S Siderite U S S tivity that is caused by crushing of the samples Anglesite U S S S before BET measurement. A minimum value for Cerussite U S S the SSA, and thus a maximum value for the area- Cuprite S normalised flux, is given by a purely geometric Pb(OH) US S 2 approach to surface area calculation. The simplest Smithsonite U S S S method is to assume that the volume of material U: undersaturated, SI < 5. S: close to saturation 5 < SI < 1. that interacts with infiltrating fluids is divided into S: Saturated: 1 < SI < 1; S+: close to supersaturation, 1 < SI < 5. These relatively wide limits are used to compensate for cubic blocks with side length d m. This assumption errors in the calculated S.I. as a result of poorly known ther- gives a total reacting surface area of 6V/d, and a modynamic data for the phases of interest. surface area normalised flux of Qid/6V. The ratio 394 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

Table 6 Area-normalised field fluxes of elements 2 1 FB (moles m s ) QH flux QH Ln QH Ln D Flux D Ln DLn Y flux Y Ln YLn CC Flux CC Ln CC Ln flux r(flux) flux r(flux) flux r(flux) flux r(flux) Si 1.1E14 40 2.2 1.3E15 34 2.1 2.0E18 41 2.1 Al 4.4E16 35 2.1 6.7E17 37 2.4 2.2E19 43 2.1 Fe 2.1E16 36 2.1 1.0E14 32 2.0 7.0E20 44 2.7 Mg 7.9E15 32 2.1 1.5E13 37 2.2 2.3E14 31 2.0 2.9E18 40 2.1 Mn 1.2E16 37 2.1 2.4E16 36 2.1 1.4E20 46 2.1 Ca 1.5E14 32 2.0 4.9E13 36 2.1 2.6E14 31 2.0 3.9E18 40 2.1 Na 1.6E13 29 2.1 2.9E14 39 2.2 2.3E14 31 2.0 1.2E17 39 2.1 K 1.2E14 32 2.1 2.3E15 41 2.3 2.5E15 34 2.0 1.4E18 41 2.1 Zn 4.8E17 38 2.1 2.4E16 44 2.1 4.9E18 40 5.3 5.0E20 44 2.1 Pb 6.9E18 47 2.4 1.4E18 41 2.5 Cu 3.0E19 43 2.1 S 1.4E14 32 2.1 2.5E14 39 2.1 7.0E14 30 2.0 4.2E18 40 2.1 Cl 1.2E13 30 2.1 4.0E14 39 2.2 1.1E17 39 2.1 F 9.5E15 40 2.1 5.1E19 42 2.1 N 4.5E14 38 2.2 3.4E18 40 2.1

FG/FB:d = 0.01 8.6 E3 1.9 E3 3.0 E3 2.9 E4 FG/FB:d = 0.1 8.6 E4 1.9 E4 3.0 E4 2.9 E5 FG/FB:d = 1 8.6 E5 1.9 E5 3.0 E5 2.9 E6

Table 7 Comparison between laboratory and ﬁeld results for Ynysarwedd mudstone Mechanism Si Al Fe Mg Mn Ca Na K S Field I I I E I E E E E Column E E ID X, T X, T X, T X, T X T Batch S ID E S S S S S

2 1 Release rates (moles m s ) FG/FB Field 1.3E15 6.7E17 1.0E14 2.3E14 2.4E16 2.6E14 2.3E14 2.5E15 7.0E14 3 E4a Column 1.63E14 4.11E15 2.84E14 1.98E13 4.90E14 1.82E13 1.15E13 1.04E13 6.49E13 3 E2 Batch 1.36E12 n/a n/a 6.80E13 2.1E13 7.6E13 n/a 7.6E12 2.0E12 3 E2 I: general transport; ID: indeterminate; D: depletion; E: equilibrium; X: ion exchange; T: transport; S: surface kinetics. a Calculated for d value of 0.1 m.

of the maximum flux (FG) to the minimum flux (FB) tainties on Qi at Quaking Houses, Grattendale and is then 6/(dqSSABET). This value is tabulated in Church Coombe were set to the standard deviation Table 6 for d values of 0.01, 0.1 and 1 m. of the flux measurements. Flux uncertainties at Yny- V values are calculated from geometric and sarwed were set to twice the standard deviation of hydrological considerations at each field site (see the measurements to account for the disturbance below). Uncertainties in FB were propagated assum- to flow rates caused by construction of the mine ing that the input parameters, Qi, V, q, and SSA, effluent treatment plant. Uncertainties in q and were uncorrelated. Calculations were made using SSA were set to 10% and 20% of their respective the standard expression for propagation of the measured values. Note that this strategy does not in- uncertainty on x where x = f(a,b,c...): clude systematic error that stem from the inappropriate choice of surface area estimation method; ox 2 ox 2 ox 2 r2 r r r 3 results that use the same method for surface area x ¼ o a þ o b þ o c þ ð Þ a b c estimation can be compared, but additional uncer- Error propagation used a natural log scale because tainty should be added if those results are to be com- of the large size of the uncertainties involved. Uncer- pared with results that use a different method of K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 395 surface area estimation. The bulk of the uncertainty ground surface. The estimated rock volume is was attributed to V via a standard deviation of ±2 1.4 · 108 m3. The average density of the material 3 3 natural log units. Uncertainties on FG are of a simi- was assumed to be 2.6 · 10 kg m , taken from lar order of magnitude for any given value of d. the lower range of estimates of density for crustal rocks (Verhoogen et al., 1970). 3.3.1. Masses of interacting rock (Vq) At Quaking Houses the volume of the tip, and 3.4. Comparison with laboratory work the proportion that interacts with the discharge at outfall G were calculated from plans of the site, Rate-controlling mechanisms and area-normalised and from borehole spoil logs (Pritchard, 1997). element fluxes from Ynysarwed are compared with The interacting proportion was estimated to be 0.1 rate-controlling mechanisms and area-normalised of the total volume of the tip (Pritchard, 1997): element fluxes for batch and column weathering 1.08 · 105 m3. The contribution of groundwater studies performed on the Ynysarwed mudstone from beneath the tip was assumed to be negligible (YM) material (Evans and Banwart, 2006). because the underlying clay has a very low permeability. The average density of the material was 4. Results assumed to be 1.8 · 103 kg m3 (Pritchard, 1997). At Grattendale, the recharge area was calculated 4.1. Water sampling and analysis from output flow rate, rainfall figures and evapotranspiration estimates. For an average output flow Representative element release patterns for con- rate of 800 ml s1, rainfall of 111 cm a1 (data for centration versus time (Fig. 2), concentration ver- 1999–2000 taken from rainmeter at nearby Birc- sus flow rate (Fig. 3), element flux versus time hover) and 50% evapotranspiration, the area of (Fig. 4) and element flux versus flow rate (Fig. 5) recharge was calculated to be 0.044 km2. The aver- illustrate the wide range of behaviour observed. age height of the ground above the level of the adit Rate-controlling mechanisms are summarised in portal was estimated, using contours from the geo- Table 2. Table 2 shows that, for just over half logical map (Stevenson et al., 1982), to be 177 m. the elements analysed, the rate-controlling mecha- The interacting rock volume (7.79 · 106 m3) is the nism is general transport. This is reflected by the product of these two parameters. The average scatter in many of the plots in Figs. 2–5. Equilib- density of the material was assumed to be 2.3 · rium-controlled element release is common at 103 kg m3. This was calculated assuming 83% car- Grattendale (Mg, Ca, Na, S: Figs. 2–5d, Cl and bonate (q = 2.8 · 103 kg m3) and 17% void, as F), Ynysarwed (Mg, Mn: Figs. 2–5e, Ca, Na, K, either porosity or mined area. S), and Church Coombe (Si, Fe, Ca, Na, S, Cl). At Ynysarwed, rock interacting with percolating Transport-controlled element release is rarer, and fluids was assumed to be collapsed mudstones and is shown at Quaking Houses (Fe, Mn, S: Figs. coal filling 80% of the previously mined volume of 2–5b) and Church Coombe (Cu, N: Figs. 2–5g, 1.8 · 106 m3 (Younger et al., 1996; Younger and and F: Figs. 2–5h). There is no evidence of sur- Adams, 1997). The average density of the material face-kinetic controlled release rates. Other note- was assumed to be 1.7 · 103 kg m3. The density is worthy features include: that of spoil (Pritchard, 1997), which has been reduced to reflect the presence of 17% (by weight) • The decline in Fe concentrations with time at of coal with a density of 1.2 · 103 kg m3. Ynysarwed (Fig. 2f), which is not accompanied The catchment area for the Church Coombe dis- by a significant decrease in Fe flux. This contrasts charge was identified using a digital elevation with the relatively constant Mn concentrations model (DEM) of the region with a 10 m horizontal (Fig. 2e). grid spacing. The DEM was used with a runoff • The contrast between patterns for Cu and N at algorithm to generate drainage networks (Jenson Church Coombe. Concentration versus time and Domingue, 1988), and catchments. The catch- plots (Fig. 2g and h) and concentration versus ment area for the adit at Church Coombe is flow rate (Fig. 3g and h) show opposite trends. 2.1 km2. The base of the interacting rock volume However, the two elements show identical trends was taken to be at the level of the adit portal in the element flux versus time (Fig. 4g and h) (123 m OD), and the top was taken to be the and element flux versus flow rate (Fig. 5g and h). 396 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

• The generally positive trend in the element flux– to enable small quantities of secondary minerals flow rate plots (Fig. 5). This reflects, to some such as gypsum to be shown. The total masses of extent, the fact that flux is a function of flow the calculated assemblage are close to 100 g in the rate, and that the axes are not independent. majority of cases (Table 4), validating the approach Nevertheless, elements whose release rates are taken here. However, the mass of the calculated inferred to be equilibrium- or transport-controlled assemblage is significantly less than 100 g for QB, exhibit stronger correlations than those whose YM, and YC. QB and YM are clay-rich and under- are not. estimation of the calculated mass in these cases is attributed to the presence of mixed-layer clays of unknown stoichiometry. The discrepancy for YC 4.2. Mineralogy is of less concern as it reflects the absence of C from the modal calculations. Bulk composition, mineral modes and mineral Most elements have multiple potential sources saturation indices are shown in Tables 3–5. Mineral and sinks at most sites (Fig. 6). For example, Si at saturation indices exhibit limited seasonal variation Quaking Houses could come from albite, illite, kaol- and so the results in Table 5 indicate the predomi- inite, montmorillonite or quartz. Lead at Gratten- nant values for saturation index over the year-long dale could come from either primary galena or study period. Fig. 6 combines a graphical represen- from secondary Pb hydroxide, sulphate or carbon- tation of mineral modes with a depiction of the ele- ate precipitates. Sulphur has multiple sources and ments present in each mineral. The log scale is used sinks at all of the sites except for Church Coombe.

(a) 100 (b) 100 S D S Cu W QO Pb Cu 10 QB 10 Zn Pb K Zn Na 1 K 1 Ca Na Mn Ca 0.1 Wt % Mineral Mg Mn Wt % Mineral 0.1 Fe Mg Fe Al 0.01 Si Al Si 0.01 Albite Illite Kaolinite Montmo Pyrite Qua Alunite Fe(OH) Gibbsit Gypsum Jarosit 0.001 Illite Qua Galena Barytes Sphalerit Calcite Dolomite Gypsum Zn(OH)2/CO3Pb(OH)2/CO3 rtz e r 3 e t rillonit z

e e Mineral Mineral

Mn Wt % Mineral Mn Wt % Mineral Mg Mg 0.1 Fe 0.1 Fe Al Al Si Si 0.01 0.01 Illite Kaolinite Py Qua Calcite Dolomite Side Alunite Gypsum Jarosit Albite Illite Kaolinite Pyrite Qua Fe(OH) r it e r rite tz r e tz 3 Mineral Mineral

Fig. 6. Mineral modes (left axis) and the relationship between mineral phases and dissolved elements in solution (right axis). (a) Quaking Houses. QO: Quaking Houses orange; QB: Quaking Houses black. (b) Grattendale. D: ore-bearing limestone; W: ore-free limestone. (c) Ynysarwedd. YS: sandstone; YM: mudstone; YC: coal. (d) Church Coombe. Spots indicate the presence of an element (RH y axis) in a mineral (x axis) K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 397

No mineral source of Cl or N was identified. Modes that the secondary minerals change composition of F-bearing minerals such as fluorite and fluor-apa- to remain in equilibrium with a seasonally chang- tite were not calculated because the atomic number ing fluid composition. If this is the case then the of F was too low for accurate analysis of this ele- assumption of relatively constant values for min- ment by XRF. However, it is likely that fluorite eral compositions and fluid parameters such as was present at Grattendale, which is close to the pH and O2 is invalid at a site such as Quaking famous Derbyshire Blue John fluorite mines, and Houses, and mineral equilibrium would be unable that fluor-apatite was present at Church Coombe to fix the solute concentrations of the constituent (apatite identified in thin section). elements of the secondary minerals. The diagnosis of equilibrium-controlled element release from concentration-flow rate characteristics 4.3. Area-normalised element fluxes is backed up, in most cases, by calculated chemical saturation of the solution with respect to at least BET surface area-normalised field fluxes of 13 20 one source or sink mineral identified as present elements (FB: Table 6) range from 10 to 10 within the rock samples (compare Tables 2 and 5). moles m2 s1. One sigma uncertainties are almost A notable exception is Cl at Grattendale and Yny- all close to 2 natural log units, which reflects the sarwed; this element is likely to have behaved con- predominance of the contribution of the uncer- servatively during fluid–rock interaction within the tainty in V to the total uncertainty. Larger uncer- mine system, and the concentration is probably tainties, such as Fe at Church Coombe, or Zn at determined externally to the adit systems. The Ynysarwed record scatter in the initial flux data. rate-controlled mechanism is therefore deemed to Fluxes of Na and Cl from Quaking Houses are be general transport (I). Other exceptions are pro- high (1013 moles m2 s1) and have very similar vided by Na at Gratttendale, Ynysarwed and values. This is likely to be the result of the dissolu- Church Coombe, Mg at Ynysarwed, and Ca at tion of a large uncovered pile of road salt that was Ynysarwed and Church Coombe. These elements stored unprotected at the tip at the time of the are commonly found as part of cation exchange monitoring program. Adjusted BET-derived fluxes assemblages in clays, and it is likely that cation (FABET: not tabulated) are 1, 2, or 3 orders of mag- exchanger surfaces on clay minerals (illite, kaolinite, nitude higher than the FBET values, depending on montmorillonite) buffered the aqueous concentra- the estimate of the field grain size. Geometric sur- tions of these elements. The equilibrium-controlled face area-normalised element fluxes (FG: Table 6) element release inferred for F at Grattendale is are 3–6 orders of magnitude higher, depending likely to result from saturation with respect to fluo- on the assumed value of d. rite. Solutions are predicted to be supersaturated with respect to minerals in some cases, for example, 4.4. Comparison with laboratory work chlorite at Ynysarwed. This results partly from slow kinetics of mineral precipitation for some minerals, A systematic variation in element release mecha- but may also reflect poorly known thermodynamic nism as a function of the type of system was data for the minerals, the effects of mineral solid observed (Table 7). Equilibrium-controlled element solution on mineral stability, and changes in solu- release was most common at field scale; transport- tion parameters such as pH and Eh when the dis- controlled element release was predominant in the charges are exposed to the atmosphere. column experiments, and surface-kinetic controlled A number of minerals that are observed to be element release was observed in the batch experi- present, and that are calculated to be saturated ments. There was also a systematic variation in the in the discharges from outflow G at Quaking rate of element release (Table 7; Fig. 7). Element Houses (alunite, jarosite, Fe oxy-hydroxides) do fluxes from the batch experiments were 1–2 orders not appear to exert an equilibrium control on the of magnitude faster than those from the column release of their constituent elements (Fe, K, S, experiments which were, in turn, 1–2 orders of Al). It is tempting to attribute this to dilution of magnitude faster than BET-normalised field fluxes the tip discharge by relatively clean rainwater from (FB). Adjusted BET-normalised fluxes for field grain the road drains. However, if this were the case then sizes that were two orders of magnitude higher than the saturation indices of the secondary minerals those in the columns (0.1 m) were similar to the would drop. This is not observed. It is possible BET-derived fluxes from the column experiments. 398 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

-10 -10 Si

) Si ) -1 Mg

s Al -12 -1 -12 s -2 Fe Mn -2 Mg Ca -14 -14 Mn K Ca S -16 -16 Na K -18 -18

Log Column Flux (mol m S Log Batch Flux (mol m a b -20 -20 -20 -18 -16 -14 -12 -10 -20 -18 -16 -14 -12 -10 Log Field Flux (mol m-2 s-1) Log Field Flux (mol m-2 s-1)

Fig. 7. Comparison between element ﬂuxes from ﬁeld (FB) and (a) column experiments and (b) batch experiments. Error bars indicate ±1 standard deviation.

GSA-normalised fluxes (FG) were higher in all cases, 1996). Consideration of surface-area-normalised but showed the same trend between batch and col- fluxes is not helpful in this case, because of the large umn results. Field FG values for d values of 0.1 m uncertainty on the surface-area appropriate for use were within error of the column FG values. The other in field-based calculations. However, the apparent rock types give similar results (work in progress). contradiction can be resolved if hydrological factors are considered. Long residence times and low 5. Discussion water:rock ratios are typical of field hydrological conditions, and these factors favour equilibrium- 5.1. Rate-controlling element release mechanisms controlled mineral dissolution. The scale of observation is also a controlling fac- The most important result of this study is the pre- tor. At grain scale, solubility equilibrium occurs dominance of equilibrium- and transport-controls when diffusion of solutes away from the surface is on element release rates (Table 2) which is observed slow compared to the rate of dissolution. At aquifer at four very different UK mine sites. This suggests scale, equilibrium results when hydrological flushing that direct scaling between laboratory-derived sur- is sufficiently slow that mineral dissolution proceeds face-kinetic controlled mineral dissolution rates to equilibrium over a time-scale shorter than the res- and equilibrium- or transport-controlled field rates idence time of water within the hydrological system. is unlikely to be appropriate. This conclusion is sup- A number of other scale-dependent factors that ported by the comparison between rate-controlling influence the rate-controlling mechanism are also mechanisms for the Ynysarwed material (Table 7). very different in laboratory and field (see below). Surface-kinetic processes dominate in the batch experiments, transport and cation-exchange pro- 5.2. Influences on rate-controlling mechanism cesses control element release in the column experiments, and equilibrium control is most important in An understanding of the parameters that control the field. the rate-controlling mechanism of element release The occurrence of equilibrium-controlled ele- requires consideration of the factors that favour ment release presents something of a paradox; min- each mechanism. Equilibrium control is favoured eral dissolution processes that result in equilibrium by factors or processes that lead to high element would normally be assumed to have proceeded concentrations. These can be split into hydrological more rapidly than those which did not, because rel- factors, such as low water:rock ratios, long resi- atively rapid dissolution is required to produce sol- dence times, and significant evaporation, and miner- ute concentrations that are high enough to reach alogical factors, such as high mineral reactivity and chemical saturation with respect to the minerals of high reactive surface area. Field environments tend interest. Yet field rates have been shown to be to have lower water:rock ratios and longer residence slower than laboratory-derived rates by 2–4 orders times than laboratory experiments, but lower min- of magnitude (e.g. Schnoor, 1990; White et al., eral reactivities, which result from large particle K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 399

1 Water:Rock Ratio

Equilibrium Control Equilibrium Control

Kinetic Control Residence Time

Equilibrium Control

Mineral Reactivity

Fig. 8. Schematic illustration of some of the factors that determine the mechanism of element release during weathering. sizes, and a general decrease in reactivity that results are present and are calculated to exert some control from prolonged weathering (e.g. White and Brant- on element concentrations at each of the four sites ley, 2003). The results of this study suggest that studied here (Table 5, Fig. 6). They have also been the hydrological factors exert a greater influence, shown to be important at a number of other types at least at the four field sites examined here. This of environment (e.g. Blowes, 1991; Benevenuti analysis suggests that a continuum exists between et al., 2000). In such situations it is inappropriate field sites with low fluid:rock ratios and high resi- to interpret field fluxes of polluting elements from dence times that have a high potential for mineral laboratory kinetic studies of primary minerals. It equilibrium and those with the opposite characteris- is instead necessary to identify dissolution/precipita- tics, that have a lower potential for mineral equilib- tion and adsorption reactions involving secondary rium (Fig. 8). Spoil tips would in general, be minerals, and quantify rates of element release with expected to fall at the low potential for equilibrium appropriate mathematical expressions. end of this scale, whereas large mined systems such as Ynysarwed would be expected to fall at the 5.4. Recognition of rate-controlling mechanism higher end. This conceptual model is supported by the results in Table 2. Recognition of rate-controlling mechanism for Transport-controlled element release is associ- element release requires both monitoring of dis- ated with sites that have relatively large variations charge compositions and investigation of the miner- in hydrological parameters such as flow rate and alogy of material at a mine site. For example, use of the height of the water table. Variations in these mineralogy and mineral equilibrium calculations parameters are more extreme at smaller, hydroge- alone at Quaking Houses would have concluded that ologically isolated sites with low residence times, the presence of alunite, Fe oxy-hydroxides, gibbsite, and where water storativity is not high enough to alunite and jarosite controlled the release of Fe, Al, buffer changes in recharge rates driven by temporal S, and K from the site, a result apparently inconsis- variation in rainfall and other precipitation. Quak- tent with those of the water-quality monitoring ing Houses is such a site, and exhibits transport- study reported here. Similarly, water analyses alone controlled element release for more elements than could have led to the conclusion that Cl and N were any of the other sites. fixed by the presence of some mineral at Church Coombe, but mineralogical analysis shows that this 5.3. Mineral sources and sinks is not the case and that some other mechanism is required to explain their constant concentrations. The results of this study emphasize the impor- Water-quality monitoring should also be long-term tance of secondary minerals in the control of ele- (at least a year). This allows recognition of mono- ment fluxes from mine sites. Secondary minerals tonic temporal trends in element concentrations 400 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 and/or fluxes, for example, the declining Fe concen- BET-derived surface area, and, in spite of exten- trations at Ynysarwed (Younger et al., 1996). More sive research (e.g. Fischer and Gaupp, 2004; detailed mineralogical analysis than that presented Metz et al., 2005), there is no commonly used here is necessary if the role of cation exchange is to method for measurement of this parameter. The be understood and quantified at field sites. strategy recommended here is that effective surface area conventions should be consistent within 5.5. Use of algorithms any study, and that the uncertainties introduced by scaling between different particle sizes and/ The use of scaling algorithms (Sverdrup and or mineral reactivities should be incorporated Warfvinge, 1995; Malmstrom et al., 2000) must be into error propagation calculations. undertaken with caution if there is reason to suspect 2. The presence of soluble secondary minerals such that element release rates are controlled by mineral as alunite and jarosite could provide a significant equilibrium. Application of the algorithms to sites source of contamination which can be accessed if such as the Aitik mine site are likely to be successful site hydrogeology changes significantly. This is (e.g. Malmstrom et al., 2000) because the mine spoil enhanced if the minerals incorporate toxic metals is porous and highly reactive, and water-residence (e.g. Winland et al., 1991). One possible example times are low (Eriksson et al., 1997; Stromberg of this is the change in water quality observed at and Banwart, 1999). However, these algorithms Quaking Houses after the construction of the would not have been suitable for the majority of Annfield Plain by-pass. Options such as dry-cap- the elements at the mine sites monitored for this ping by landscaping and soil-washing should study, and different methods should be used. therefore be considered carefully. Fortunately, the systematics of element concentrations controlled by mineral equilibrium are rela- 6. Conclusions tively simple; concentrations remain approximately constant so long as the mineral assemblage and Analysis of element concentration and element other factors such as temperature and pH remain fluxes suggest that equilibrium- and transport-con- unchanged. However, exhaustion of the controlling trolled element release mechanisms control the con- mineral phase will lead to changes in the concentra- centrations of major elements and the principal tions of elements in solution (e.g. Banwart and contaminant metals at the four UK mine sites stud- Malmstrom, 2001), and the potential for such ied. No evidence was found for surface-kinetic con- changes must be taken into account in any long trolled element release. Mineralogical analysis and term prediction exercise. mineral equilibrium calculations support this conclusion. However, diagnosis of rate-controlling 5.6. Practical implications mechanism from trends in water analyses can be incorrect if there is significant variation in the The practical implications of this work for pre- hydrological characteristics of the site, if there is sig- diction and remediation of contamination at mining nificant depletion of source minerals, or if the con- sites are twofold. centrations of conservative elements such as Cl and N are controlled by processes external to the 1. An integrated study of mineralogical and water- system under consideration. Secondary minerals quality parameters is necessary if the mechanisms such as goethite, gibbsite, alunite, jarosite and Pb that control element release at mine sites are to and Zn hydroxides, sulphates and carbonates play be understood. The most poorly known parame- an important role in the control of element concen- ters involved in the calculations presented here trations from the sites. Comparison of element are the volume of rock or spoil that interacts with fluxes from the mine sites with those from the labo- percolating water, and the effective surface area ratory requires that the field fluxes should be nor- of this material. More detailed hydrogeological malised to surface area. The biggest uncertainties analysis of sites (e.g. Rivett and Allen-King, in such calculations are the volume of rock that 2003) can be used to constrain the former param- interacts with percolating solutions and the specific eter. Issues with the effective surface area are surface area of reacting mineral grains in the field. more difficult to resolve. The effective surface A comparison of field- and laboratory-derived area is not equal to either the geometric or element fluxes for one of the sites shows that surface K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 401 kinetic mechanisms control the rate of element In: Younger, P.L., Robins, N.S. (Eds.), Mine Water Hydrog- release from the batch experiments, while transport eology and Geochemistry. The Geological Society of London, and equilibrium-control dominate in column exper- London, pp. 137–157. Banwart, S.A., Zhang, S., Evans, K.A., 2004. Resolving the scale iments and in the field. Element fluxes from batch dependency of laboratory and field weathering rates. In: experiments were 1–3 orders of magnitude faster Wanty, R.B., Seal, R.R. (Eds.), Proc. 11th Internat. Symp. than those from column experiments and field mea- WRI-11. Balkema, Leiden, The Netherlands, pp. 1443–1447. surements. This result presents a paradox, given Barclay, W.J., Taylor, K., Thomas, L.P., 1988. Geology of the that surface-kinetic mineral dissolution would be South Wales Coalfield. Part V. The Country around Merthyr Tydfil. British Geological Survey, Keyworth, UK. expected to proceed more slowly than that con- Benevenuti, M., Mascaro, I., Corsini, F., Ferrari, M., Lattanzi, trolled by mineral equilibrium. This apparent con- P., Parrini, P., Costagliola, P., Tanelli, G., 2000. Environ- tradiction can be resolved by the consideration of mental mineralogy and geochemistry of waste dumps at the hydrological parameters such as water:rock ratio Pb(Zn)–Ag Bottino mine, Apuane Alps, Italy. Eur. J. Min. and water residence time. The implications for prac- 12, 441–453. Blowes, D.W., 1991. The formation and potential importance of tical prediction and remediation of contamination cement layers in active sulphide mine tailings. Geochim. at mine sites are that water-quality monitoring Cosmochim. Acta 55, 965–978. should be accompanied by mineralogical analysis, Bottrell, S.H., Hardwick, P., Gunn, J., 1999. Sediment dynamics and that the role of secondary minerals should be in the Castleton karst, Derbyshire, UK. Earth. Surf. Proc. considered carefully. Landforms 24, 745–759. Bottrell, S.H., Webber, N., Gunn, J., Worthington, S.R.H., 2000. The geochemistry of sulphur in a mixed allogenic–autogenic Acknowledgments karst catchment, Castleton, Derbyshire, UK. Earth. Surf. Proc. Landforms 25, 155–165. Thanks to Chris Rees, of the Cefn Coed Mining Clow, D.W., Drever, J.I., 1996. Weathering rates as a function of Museum, Jim Rieuwerts from the Peak District flow through an alpine soil. Chem. Geol. 132, 131–141. Czerewko, M.A., 1997. Diagenesis of mudrocks, illite ‘crystallin- Mining Historical Society, Staff at Cornwall County ity’ and the effects on engineering properties. Unpubl. Ph.D. Records Office for help with local data sources. Also thesis. University of Sheffield. to Adam Jarvis, Patrick Orme and Lesley Batty for Dines, H.G., 1956. The Metalliferous Mining Region of South- Quaking Houses analysis, to staff at the Assay West England. Vol. 1. Memoirs of the Geological Survey of Office, Sheffield, to Bev Lane at the Centre for Great Britain. Her Majesty’s Stationary Office, London. Eriksson, N., Gupta, A., Destouni, G., 1997. Comparative Materials Engineering at the University of Sheffield analysis of laboratory and field tracer tests for investigating for BET analysis, and to Paul Younger for specific preferential flow and transport in mining waste rock. J. and general knowledge. This work was funded by Hydrol. 194, 143–163. NERC. Thanks also to two anonymous reviewers, Evans, K.A., Banwart, S.A., 2006. Rate controls on the chemical whose reviews have improved this work. weathering of natural polymineralic material. I. Dissolution behaviour of polymineralic assemblages determined using batch and unsaturated column experiments. Appl. Geochem. References 21, 352–376. Evans, K.A., Gandy, C.J., Banwart, S.A., 2003. Mineralogical, Adam, K., Kourtis, A., Gazea, B., Kontopoulos, A., 1997. numerical and analytical studies of the coupled oxidation of Evaluation of static tests used to predict the potential for acid pyrite and coal. Min. Mag. 67, 381–398. drainage generation at sulphide mine sites. Trans. Inst. Min. Ewbank, G., Manning, D.A.C.., Abbott, G.D., 1995. The relation- Metall. 106, A1–A8. ship between bitumens and mineralisation in the South Pennine Appelo, C.A.J., Postma, D., 1993. Geochemistry, Groundwater Orefield, Central England. J. Geol. Soc. Lond. 152, 751–765. and Pollution, vol. 1. A.A. Balkema, Rotterdam. Ferry, J.M., 1988. Infiltration-driven metamorphism in northern Ball, J.W., Nordstrom, D.K., 1991. WATEQ4F: users manual New England, USA. J. Pet. 29, 1121–1159. with revised thermodynamic data base and test cases for Fischer, C., Gaupp, R., 2004. Multi-scale rock surface area calculating speciation of major, trace and redox elements in quantification – a systematic method to evaluate the reactive natural waters. Open-File Report 90-129, US Geol. Surv., surface area of rocks. Chemie der erde-geochemistry 64, 241– Denver, Colorado. 256. Banks, D., Younger, P.L., Arnesen, P.L., Iverson, E.R., Banks, Ford, T.D., Rieuwerts, J.H., 1970. Lead Mining in the Peak S.B., 1997. Mine-water chemistry: the good, the bad and the District. Peak Park Planning Board, Bakewell, Derbyshire. ugly. Environ. Geol. 32, 157–174. Gandy, C.J., 2002. Reactive transport modelling of pyrite Banwart, S.A., Malmstrom, M.E., 2001. Hydrochemical model- oxidation at Quaking Houses, County Durham. Unpubl. ling for preliminary assessment of minewater pollution. J. Ph.D. thesis. University of Newcastle-upon-Tyne. Geochem. Explor. 74, 73–97. Gandy, C.J., Evans, K.A., 2002. Laboratory and numerical Banwart, S.A., Evans, K., Croxford, S., 2002. Predicting mineral modelling studies of iron release from a spoil heap in County weathering rates at field scale for mine water risk assessment. Durham. In: Robins, N.S., Younger, P. (Eds.), Minewater 402 K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403

Hydrogeology and Geochemistry. Geological Society, Lon- DNAPL: dissolved-plume retardation. J. Contam. Hydrol. don, pp. 205–214. 66, 117–146. Gandy, C.J., Younger, P.L., 2003. Effect of a clay cap on Schnoor, J.L., 1990. Kinetics of chemical weathering: a compar- oxidation of pyrite within mine spoil. Quart. J. Eng. Geol. 36, ison between laboratory and field weathering rates. In: 207–215. Stumm, W. (Ed.), Aquatic Chemical Kinetics. Wiley, pp. Gleeson, S.A., Wilkinson, J.J., Stuart, F.M., Banks, D.A., 2001. 475–504. The origin and evolution of base metal mineralising brines Srour, S., 1998. Modelling groundwater flow through a body of and hydrothermal fluids, South Cornwall, UK. Geochim. contaminated land: Morrison Busty Spoil Heap, County Cosmochim. Acta 65, 2067–2079. Durham. Unpubl. M.Sc. thesis. University of Newcastle upon Hamilton-Jenkin, A.K., 1965. Mines and Miners of Cornwall. X. Tyne. Camborne and Illogan. Truro Bookshop, Truro. Stallard, R.F., Eddmond, J.M., 1987. Geochemistry of the Holmes, P.R., Crundwell, F.K., 2000. The kinetics of the oxidation Amazon. 3. Weathering chemistry and limits to dissolved of pyrite by ferric ions and dissolved oxygen: an electrochem- inputs. J. Geophys. Res. 92, 8293–8302. ical study. Geochim. Cosmochim. Acta 64, 263–274. Stevenson, I.P., Price, D., Chisholm, J.I., Wedd, C.B., Smith, Jambor, J.L., 2000. Mineralogy and acid rock drainage. In: E.G., Eden, R.A., Rhys, G.H., 1982. Classical areas of British Cotter Howells, J.D., Campbell, L.S., Valsami-Jones, E., Geology: Bakewell, Sheet SK26, 1:25000. Geological Survey Batchelder, M. (Eds.), Environmental Mineralogy: Microbial of England and Wales. Interactions, Anthropogenic Influences, Contaminated Land Stromberg, B., Banwart, S., 1999. Weathering kinetics of waste and Waste Management. Mineralogical Society of Great rock from the Aitik copper mine, Sweden: scale dependent Britain and Ireland, London, pp. 141–159. rate factors and pH controls in large column experiments. J. Jenson, S., Domingue, J., 1988. Extracting topographic structure Contam. Hydrol. 39, 59–89. from digital elevation data for geographic information Sverdrup, H., Warfvinge, P., 1995. Estimating field weathering systems analysis. Photogramm. Eng. Rem. Sensing 54, rates using laboratory kinetics. In: White, A.F., Brantley, S.L. 1593–1600. (Eds.), Chemical Weathering Rates of Silicate Minerals. Malmstrom, M.E., Destouni, G., Banwart, S.A., Stromberg, Mineralogical Society of America, pp. 485–541. B.H.E., 2000. Resolving the scale-dependence of mineral Taylor, R.K., 1989. Report GPC/20410. Mining Department, weathering rates. Environ. Sci. Technol. 34, 1375–1378. National Coal Board. Markey-Amey, J., 1995. Stanley Burn Dissertation. University of Thompson, J.B.J., 1982. Composition space: an algebraic and Sunderland. geometric approach. In: Ferry, J.M. (Ed.), Characterisation Martello, D.V., Vecchio, K.S., Diehl, J.R., Graham, R.A., of Metamorphism through Mineral Equilibria, Reviews in Tamilia, J.P., Pollack, S.S., 1994. Do dislocations and Mineralogy. Mineralogical Society of America, Washington, stacking faults increase the oxidation rate of pyrite? Geochim. USA, pp. 1–31. Cosmochim. Acta 58, 4657–4665. van Grinsven, J.J.M., van Riemsdijk, W.H., 1992. Evaluation of Metz, V., Raanan, H., Pieper, H., Bosbach, D., Ganor, J., 2005. batch and column techniques to measure weathering rates in Towards the establishment of a reliable proxy for the reactive soils. Geoderma 52, 41–57. surface area of smectite. Geochim. Cosmochim. Acta 69, Verhoogen, J., Turner, F.J., Wieiss, L.E., Wahrhaftig, C., Fyfe, 2581–2591. W.S., 1970. The Earth an Introduction to Physical Geology. Newbegin, D., 1997. Stanley Burn – Monitoring of Pollution. Holt Rinehart and Winston, New York. Durham County Council, Durham. White, A.F., Blum, A.E., Schulz, M.S., Bullen, T.D., Harden, Oakham, C.D., 1979. The sough hydrology of the Wirksworth- J.W., Peterson, M.L., 1996. Chemical weathering rates of a Matlock-Youlgreave area, Derbyshire. Unpubl. Ph.D. thesis. soil chronosequence on granitic alluvium. 1. Quantification of University of Leicester. mineralogical and surface area changes and calculation of Parkhurst, D.L., Appelo, C.A.J., 1999. Users guide to PhreeqC primary silicate reaction rates. Geochim. Cosmochim. Acta (version 2): a computer program for speciation, batch 60, 2533–2550. reaction, one dimensional transport and inverse geochemical White, A.F., Blum, A.E., 1995. Effects of climate on chemical- calculations. Water-Resour. Investig. Rep. 99-4259, US Geol. weathering in watersheds. Geochim. Cosmochim. Acta 59, Surv., Denver, Colorado. 1729–1747. Peiffer, S., Stubert, I., 1999. The oxidation of pyrite at pH7 in the White, A.F., Brantley, S.L., 2003. The effect of time on the presence of reducing and non reducing Fe(III) chelators. weathering of silicate minerals: why do weathering rates differ Geochim. Cosmochim. Acta 63, 3171–3182. in the laboratory and field. Chem. Geol. 202, 479–506. Pritchard, J., 1997. Leachate release from a large body of Wieland, E., Wehrli, B., Stumm, W., 1988. The coordination contaminated land: Morrison Busty Spoil Heap, County chemistry of weathering. III. A generalization on the disso- Durham. UnPubl. M.Sc. thesis. University of Newcastle- lution rates of minerals. Geochim. Cosmochim. Acta 52, upon-Tyne. 1969–1981. Ranson, C.M., Edwards, P.J., 1997. The Ynysarwed experience: Winland, R.L., Traina, S.J., Bigham, J.M., 1991. Chemical- active intervention, passive treatment and wider aspects. In: composition of ochreous precipitates from Ohio coal-mine Younger, P.L. (Ed.), Minewater Treatment using Wetlands. drainage. J. Environ. Qual. 20, 452–460. University of Newcastle, Newcastle-upon-Tyne, pp. 151–154. Xie, Z.X., Walther, J.V., 1992. Incongruent dissolution and Rieuwerts, J.H., 1987. History and gazeteer of the lead mine surface-area of kaolinite. Geochim. Cosmochim. Acta 56, soughs of Derbyshire. Privately Published. 3357–3363. Rivett, M.O., Allen-King, R.M., 2003. A controlled field exper- Younger, P.L., 1996. Technical Audit of the Phase I Feasibility iment on groundwater contamination by a multicomponent Study on Long term remediation of the Neath Canal and K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403 403

River Dulais Minewater Discharges. University of Newcastle Younger, P.L., Adams, R., 1997. Improved modelling of aban- upon Tyne, Newcastle-upon-Tyne. doned coalfields. University of Newcastle-upon-Tyne, New- Younger, P.L., 1997. The longevity of minewater pollution: a castle-upon-Tyne. basis for decision-making. Sci. Total Environ. 194, 457–466. Younger, P.L., Pennell, R., Cowpertwait, P.S.P., 1996. Evalua- Younger, P.L., 2000. Predicting temporal changes in total iron tion of temporal changes in the quality of the lower concentrations in groundwaters flowing from abandoned Ynysarwed minewater discharge, Neath Valley, South Wales. deep mines: a first approximation. J. Contam. Hydrol. 44, Report from Nuwater Consulting Services Ltd., Dept. Civ. 47–69. Eng. University of Newcastle upon Tyne.