GEOCHEMICAL MODELLING OF ACID MINE DRAINAGE IN MILL TAILINGS: QUANTIFICATION OF KINETIC PROCESSES FROM LABORATORY TO FIELD SCALE
S. Ursula Salmon
November 2003
TRITA-LWR PHD 1009
ISSN 1650-8602 ISRN KTH/LWR/PHD 1009-SE ISBN 91-7283-607-5 S. Ursula Salmon TRITA-LWR PHD 1009
ii Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
ABSTRACT Assessment of the potentially acidic, heavy metal–laden leachates that leave deposits of sulfide ore mill tailings and evaluation of various possible options for mill tailing remediation are scientific problems of increasing practical importance. High costs may be associated with the mill tailing remediation, not least after recent changes in Swedish and European environmental legislation. This thesis presents a methodology for studying and quantifying geochemical processes that contribute to generation of so-called acid mine drainage (AMD). The methodology builds from first principles regarding geochemical processes, and is based on geochemical characterisation of the mill tailings combined with explicit model quantification of the effect of factors, such as temperature, pH, and mineral (BET) surface area, that influence mineral weathering rates. Application of the modelling methodology to a case study site, Impoundment 1, Kristineberg, northern Sweden, including quantification of slow processes through literature rate laws, successfully reproduced the pH and relative concentrations of major ions in the impoundment groundwater. Absolute concentrations of most major ions, with the exception of Zn, were 1-2 orders of magnitude higher in the model than in the field, which is consistent with the commonly observed scale dependence of mineral weathering rates; however, application of a -2 single calibration factor, Xr=10 , to all weathering rate expressions, sufficed to account for this apparent scale dependence. Subsequent laboratory determination of mineral weathering rates in Impoundment 1 tailings indicated that rates for the major minerals pyrite (FeS2) and aluminosilicates were in fact 1-2 orders of magnitude lower in the ~50-year-old tailings than rates reported in the literature.
Weathering rates of chalcopyrite (CuFeS2) and sphalerite (ZnS) were by contrast 1-3 orders of magnitude greater than predicted by the literature rate laws that were used in the modelling study. While the mechanism of Zn release requires further investigation for improved forward model prediction, the underestimation of Zn concentration in Impoundment 1 by the model was resolved. The laboratory study furthermore indicated that the weathering rates of most major minerals exhibited the same dependence on pH, temperature and surface area as reported in the literature, and thereby supported the use of literature rate laws for model assessment of dominant geochemical processes in tailings deposits, once allowance is made for lower rates in older tailings material. Analysis of the dominant geochemical processes in the model of Impoundment 1 indicated that slow weathering of aluminosilicate minerals provided the bulk of proton attenuation and, as a result, considerably affected the rate of depletion of fast-reacting pH-buffering minerals. Inclusion of the kinetics of aluminosilicate dissolution and of the feedbacks between slow and fast processes is thus potentially crucial for prediction of pH and its long-term evolution. The sensitivity of modelled groundwater composition and pH to iron redox reactions, such as may be accelerated by acidophilic bacteria, indicated that, while iron redox cycling was low at the present case study site, quantification of microbial mediation of these reactions may be necessary for predicting AMD quality under other conditions. The laboratory studies also indicated that application of common sterilisation techniques, such as is necessary for study of relative contributions of abiotic and biotic weathering processes, had little effect on the long-term (>30 days) abiotic element release rates in the tailings. This study suggests that within certain limits, which appear narrower than currently recognised in industrial prediction practices, it is possible to predict the weathering behaviour of major minerals, and hence proton release and attenuation, in base metal tailings under field conditions.
iii S. Ursula Salmon TRITA-LWR PHD 1009
iv Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
LIST OF PAPERS APPENDED...... VII
LIST OF PAPERS NOT APPENDED...... IX
1 INTRODUCTION ...... 1
2 GENERATION OF ACID MINE DRAINAGE IN SULFIDIC MILL TAILINGS ...... 5 2.1 PROCESSES IN MILL TAILINGS...... 5 2.2 PREDICTION OF AMD ON FIELD SCALE ...... 8 3 MILL TAILINGS CASE STUDY: KRISTINEBERG IMPOUNDMENT 1...... 11
4 MODELLING OF PRE-REMEDIATION IMPOUNDMENT 1 ...... 15 4.1 CONCEPTUAL MODEL ...... 15 4.2 PROCESS QUANTIFICATION ...... 16 4.3 MODEL RESULTS ...... 17 5 MODEL SENSITIVITY AND REMEDIATION IMPLICATIONS ...... 21 5.1 SENSITIVITY ANALYSIS...... 21 5.2 IMPLICATIONS FOR IMPOUNDMENT 1 ...... 24 6 QUANTIFICATION OF ELEMENT RELEASE RATES FROM BATCH EXPERIMENTS ...... 27 6.1 MATERIALS AND METHODS...... 27 6.2 EFFECT OF PH, TEMPERATURE, AND SURFACE AREA...... 28 6.3 EFFECT OF STERILISATION...... 31 7 MODELLING WEATHERING RATES ON BATCH, COLUMN, AND FIELD SCALE .33 7.1 MODEL INTERPRETATION OF ELEMENT RELEASE RATES ...... 33 7.2 MINERAL WEATHERING RATES IN TAILINGS COMPARED TO MONOMINERALIC SAMPLES...... 33 7.3 COMPARISON: BATCH, COLUMN AND FIELD RATES ...... 37 7.4 PREDICTION OF FIELD WEATHERING RATES FROM LABORATORY EXPERIMENTS ...... 37 8 CONCLUSIONS AND IMPLICATIONS ...... 41
9 ACKNOWLEDGEMENTS ...... 45
10 REFERENCES ...... 47
v S. Ursula Salmon TRITA-LWR PHD 1009
vi Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
LIST OF PAPERS APPENDED
I. Salmon, S.U., Malmström, M.E., 2003. Geochemical processes in mill tailings deposits: Modelling of groundwater composition. (Applied Geochemistry; in print)
II. Salmon, U.J., Malmström, M., 2000. Assessing geochemical processes in mill tailings impoundments and the effect of remediation. In: Proc. ICAM 2000, 6th Int. Conf. on Applied Mineralogy, Göttingen, Germany, July 13-21, 2000, pp. 671-673.
III. Salmon, S.U., Malmström, M.E., 2001. Quantification of mineral weathering rates in tailings from Impoundment 1, Kristineberg, northern Sweden. In: Proc. Securing the Future, Int. Conf. on Mining and the Environment, Skellefteå, Sweden, June 25-July 1, 2001, pp. 747-756.
IV. Salmon, S.U., Malmström, M.E., 2003. Quantification of mineral weathering rates in mill tailings: Effect of sterilisation method. In: Proc. ICARD 2003, 6th Int. Conf. on Acid Rock Drainage, Cairns, Australia, July 14-17, 2003, p.1111-1116.
V: Salmon, S.U., Malmström, M.E., 2003. Mineral weathering rates in mill tailings from laboratory to field scale: Effect of pH, temperature, mineral content, and surface area (manuscript in preparation).
VI: Gleisner, M., Herbert, R., Salmon, S.U., Malmström, M.E., 2003. Comparison of sulfide oxidation in unweathered pyritic mine tailings. In: Proc. ICARD 2003, 6th Int. Conf. on Acid Rock Drainage, Cairns, Australia, July 14-17, 2003, p.1027-1030.
VII: Salmon, S.U., Destouni, G., 2001. National case studies: 3.Sweden. ERMITE Report: D1; the European Commission Fifth Framework Programme, Energy, Environment and Sustainable Development, Contract No EVK1-CT-2000-0078, University of Oviedo.
vii S. Ursula Salmon TRITA-LWR PHD 1009
viii Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
LIST OF PAPERS NOT APPENDED
Peer-reviewed publications: Holmström, H., Salmon, U.J., Carlsson, E., Paraskev, P., Öhlander, B., 2001. Geochemical investigations of sulfide-bearing tailings at Kristineberg, northern Sweden, a few years after remediation. Sci. Tot. Environ., 273, 111-133.
Herbert, R.B.Jr., Malmström, M., Ebenå, G., Salmon, S.U., Ferrow, E., Fuchs M., 2003. Sterilisation of mine tailings for the quantification of abiotic oxidation rates (in preparation).
MiMi-reports: Salmon, S., Malmström, M., 2002. Steady state, geochemical box model of a tailings impoundment: Application to Impoundment 1, Kristineberg, Sweden, and prediction of effect of remediation. MiMi 2002:2, Mitigation of the environmental impact from mining waste programme (MiMi), Stockholm, Sweden.
Malmström, M., Werner, K., Salmon, U., Berglund, S., 2001. Hydrogeology and geochemistry of mill tailings Impoundment 1, Kristineberg, Sweden: Compilation and interpretation of pre- remediation data. MiMi 2001:4, Mitigation of the environmental impact from mining waste programme (MiMi), Stockholm, Sweden.
Werner, K., Salmon, S., 2001. Mine waste deposits and mines in Kristineberg, Sweden: Summary of remediation programme and identification of potentially critical assumptions. MiMi 2001:3, Mitigation of the environmental impact from mining waste programme (MiMi), Stockholm, Sweden.
Salmon, S.U., 1999. Overview of models for biogeochemical modelling of acid mine drainage. MiMi 1999:4, Mitigation of the environmental impact from mining waste programme (MiMi), Stockholm, Sweden.
Herbert, R.B.Jr., Malmström, M., Ebenå, G., Ferrow, E., Salmon, S.U., Fuchs, M., 2003. Quantifying the effects of mine tailings sterilization (in review).
Licentiate thesis: Salmon S.U., 2000. Biogeochemical processes in mill tailings: Modelling and assessment of remediation effects. Licentiate thesis TRITA-AMI LIC 2053, Water Resources Engineering, Dept. Civil and Environmental Engineering, Royal Institute of Technology, Stockholm, Sweden.
International conference abstracts: Ebenå, G., Ferrow, E., Fuchs, M., Herbert, R., Malmström, M.E., Salmon, S.U., 2003. A comparison of methods for mine tailings sterilisation. In: Proc. ICARD 2003, 6th Int. Conf. on Acid Rock Drainage, Cairns, Australia, July 14-17, 2003, p.1013.
Salmon, S.U., Malmström, M.E., 2001. Mineral weathering rates in mill tailings from Kristineberg, northern Sweden. In Eleventh Annual V.M. Goldschmidt conference, Abstract #3375. LPI Contribution No. 1088, Lunar and Planetary Institute, Houston (CD-ROM).
Salmon, S., Malmström, M., 1999. Importance of various sulfide oxidation processes in remediated mill tailings impoundments. EUG-10, Strasbourg, 1999, J. Conf. Abst. 4(1):513.
ix S. Ursula Salmon TRITA-LWR PHD 1009
x Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
sulfidic wastes after closure of the mines, 1 INTRODUCTION and in particular, discharge of Hg, Cd, Pb, Cu, Zn, As, Fe, Al, and acidity (SEPA, The history of mining in Sweden stretches 1995). In 1995, it was estimated that over back over 1000 years; Swedish mines were 60% of all Pb, Cd, Zn and Cu discharge to for many centuries one of the major Swedish waters came from mining and suppliers of steel, copper and silver to mining waste, with the majority originating central Europe (Gustafsson et al., 1999). from waste (SCB 2000b). The main Although the number of operating mines environmental impact of AMD is generally has decreased from approximately 500 in the considered to be associated with the heavy early 20th century to 16 in 1999, prospecting metal content, with effects such as acute and has also increased after changes in chronic toxicity and bioaccumulation, where ownership laws in the early 1990s, indicating the low pH of the mine waste leachate that there is both the resource potential and enhances metal mobility (SEPA, 1995). the will of the mining industry for new However, AMD is also often characterised mines to open (see also further discussion by high concentrations of Fe(II), aluminium, on mining conditions in Sweden in Paper and sulfate, where these components are VII and references therein). mobile even under non-acidic conditions Fine mill tailings and waste rock are waste (e.g., Smith and Huyck, 1999). Oxidation of products from mining and ore enrichment Fe(II) upon exposure to oxygen and processes. Mining at the current level in precipitation of characteristically yellow-red Sweden produces approximately half of all Fe(III) phases, as well as precipitation of waste in Sweden (SCB, 2000a); in 1996, over Al3+ phases, both release further acidity and two thirds of the mining waste produced, or may physically impair biological function 35Mt, was from mining of sulfide ores (e.g., Pentreath, 1994). Also dissolved Al3+ (SEPA, 1998), such as chalcopyrite (CuFeS2) may be harmful at high concentrations (e.g., and sphalerite (ZnS). As sulfide ore minerals Wachtmeister and Sundstsröm, 1986; are generally associated with large quantities Pentreath, 1994). In addition, in countries of economically undesirable iron sulfides, where water is in short supply, even high such as pyrite (FeS2), the iron sulfides often concentrations of sulfate may be critically comprise a large fraction of the waste, as do detrimental to water resources (Pulles, 2003). the so-called “gangue” minerals that host the The focus of regulation of water polluting ore, typically (alumino-)silicate and activities in Sweden and the E.U. has carbonate minerals. Oxidation of the iron recently changed, from placement of limits sulfide minerals upon exposure to dissolved on individual discharges, to regulation of the molecular oxygen, e.g., integrated impact of many different 7 FeS2 (s) + H 2O + O2 (aq) → pollution sources within a catchment on the 2 receiving water body (see also more detailed 2+ 2− + Fe + 2SO4 + 2H (1), legislation discussion in Paper VII). by abiotic or microbially mediated processes, Relatively new Swedish and EU legislation combined with reactions of the other has introduced concepts such as minerals in the waste, can lead to release of Environmental Quality Standards, which will acidic leachate with high concentrations in specify limits, for example, of concentrations metals and sulfate (referred to as acid mine of metals that may not be exceeded in drainage, AMD) over hundreds to thousands surface or ground water bodies. Inability to of years (SEPA, 1986). fulfil these criteria, or to demonstrate the capability to do so, may thus have serious The largest Swedish environmental concern consequences for the future of the mining associated with mining, according the industry; while the impact of the new Swedish Environmental Protection Agency legislation on AMD regulation is largely (here referred to as SEPA), is oxidation of untried, the Swedish mining industry is
1 S. Ursula Salmon TRITA-LWR PHD 1009
concerned for its viability (Paper VII). At geochemical control of AMD quality the same time, societies will for the (Salomons, 1995), and thereby largely foreseeable future continue to depend upon neglect the crucial interaction with oxygen mining to meet resource needs (Plumlee and availability. Logsdon, 1999). Sustainable regulatory A number of modelling studies of AMD decisions on mining require means to generation from mill tailings exist in the accurately assess the long-term literature (e.g., Scharer et al., 1994; environmental impact of mining, the Wunderley et al., 1996; see also review in associated waste material, and the outcome Alpers and Nordstrom, 1999) where models of any remedial measures. have been produced as tools to examine Remediation measures are commonly effects of dominant processes and compare applied with the intention of limiting different tailings remediation measures. generation of AMD through reduction of However, the predictive capability of many oxygen availability, for example through of these models is limited (MEND, 2000; application of water or soil covers. The large Parker and Robertson, 1999), due to, for quantities of generated mining waste leads to example, oversimplification or neglect of such remediation being a costly undertaking some important geochemical processes. (e.g., SEPA, 1998). Despite the considerable Recent model developments open for better financial burden of remediation, remediation consideration of kinetic processes (Lichtner, measures are often based on reproducing at 1996; Mayer et al., 2003), with application of any mine waste site similar practices as have such models to field data including model applied elsewhere, rather than linked to the calibration to field observations (Mayer et actual, site-specific downstream effect of the al., 2003). remediation measure (O’Kane and Wels, In parallel lines of investigation, focussed on 2003). In order to quantify and predict such weathering of aluminosilicate minerals, downstream effects, accurate source term independently determined weathering rate quantification and prediction is necessary. laws for monomineralic samples on Such quantification and prediction is also laboratory scale have been applied to field- necessary for optimisation of remediation scale prediction of, for instance, acidification measures with respect to individual mine and mineral weathering impact on global waste sites (O’Kane and Wels, 2003), or the climate (White and Brantley, 1995), or AMD total number of mine waste sites within a generation in waste rock (Strömberg and water catchment (Baresel et al., 2003); Banwart, 1994). It is then commonly optimisation is required for economic observed that weathering rates determined efficiency in environmental preservation and on the laboratory scale overestimate reduced costs to the mining industry and to apparent field rates by orders of magnitude society as a whole. (e.g., Schnoor, 1990; White and Peterson, Methods that have been used for prediction 1990; Strömberg and Banwart, 1994). The of AMD generation in tailings are both study by Malmström et al. (2000) indicates laboratory and model based, with that such discrepancies between apparent standardised methods of static and kinetic field and laboratory rates may be resolved laboratory testing being often applied (e.g., through relevant, explicit consideration of White et al., 1999; MEND, 2000). However, the influence of factors, such as temperature, static tests can at best only predict whether pH, and water flow conditions, which may leachate pH will become acidic at some be quite different on different scales. (unknown) point in time (Parker and This thesis presents a modelling Robertson, 1999), and typical application of methodology that is based on fundamental kinetic laboratory testing to prediction of geochemical principles and is adapted to field scale leachate quality is associated with field conditions of a mill tailings a high degree of uncertainty (Salomons, impoundment case study, Impoundment 1, 1995). Furthermore, these methods assume Kristineberg, northern Sweden, in analogy
2 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale with earlier modelling of mining waste rock deposits (Strömberg and Banwart, 1994). In particular, this thesis addresses the importance of aluminosilicate weathering and iron redox cycling for the groundwater composition and pH in mill tailings deposits. Furthermore, laboratory studies are performed to test model assumptions made in the application of literature rate laws to the modelling of field-scale AMD generation. These laboratory studies investigate the dependence of mineral weathering rates on pH, temperature, and BET surface area under conditions relevant for mill tailings. In addition, the laboratory studies include investigation of sterilisation techniques that are required to investigate abiotic vs biotic weathering processes. Specific objectives of this thesis can be summarised as: 1. To develop a modelling methodology for quantifying leachate composition and pH in pyritic base metal tailings impoundments, which include kinetic reactions and their feedbacks with equilibrium processes. 2. To apply this modelling methodology to a specific case study, Impoundment 1, Kristineberg, using independently reported mineral weathering rate laws, and particularly investigate the site-specific importance of slow aluminosilicate dissolution and iron redox cycling coupled with fast geochemical processes and oxygen availability. 3. To test various model assumptions through systematic laboratory studies of site-specific tailings material from Impoundment 1 and separated sulfide phases, and through comparison of weathering rates determined on different experimental/observation scales.
3 S. Ursula Salmon TRITA-LWR PHD 1009
4 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
2 GENERATION OF ACID MINE DRAINAGE IN SULFIDIC MILL TAILINGS
Mill tailings are the finely ground waste product from ore enrichment by flotation. As mentioned in the introduction, common minerals in base metal pyritic mill tailings are iron sulfides and gangue minerals from the ore body host rock, such as aluminosilicate and carbonate minerals, as well as lesser amounts of ore minerals. The focus of this work is geochemical processes contributing to the major ion composition and pH in tailings impoundment groundwater. Below is an overview of major physical and geochemical processes (see also detailed reviews in Evangelou and Zhang, 1995; Nordstrom and Southam, 1997; Nordstrom and Alpers, 1999) and general means of quantification of kinetic geochemical processes, such as is applied in the modelling study in Chapter 3. In Chapter 2.2, earlier approaches to modelling the aqueous geochemistry of AMD are briefly reviewed.
the rate of oxygen consumption. If the rate 2.1 Processes in mill tailings of transport is relatively fast, the overall reaction rate is controlled by the surface
2.1.1 Oxygen availability and solute transport reaction, and the Po2 will be equal to the Fine tailings are often deposited as a slurry atmospheric Po2 (0.2 atm). If the rate of in a containing dam and allowed to transport is of a similar order of magnitude sediment. Tailings impoundment dams can as the rate of consumption, or slower, the contain vast volumes of material, for overall reaction rate will be mixed kinetic or example, the tailings dam at the Aitik open transport controlled, and Po2 will decrease. cut porphyry copper mine in northern Transport of reaction products from the Sweden, which has been in operation for reaction site to downstream environment approximately 35 years, covers up to 11 km2, depends upon inflow of precipitation, and is up to 40 m in depth (Lindvall and groundwater and/or surface water. The Eriksson, 2003). As diffusion of oxygen in aqueous composition and pH of the air is approximately 4 times greater than in groundwater in tailings deposits thus water (e.g., Nicholson et al., 1998), ingress of depends upon the relative rates of oxygen into tailings is relatively slow under geochemical processes occurring in the water-saturated conditions. After closure of tailings, as well as physical processes such as an impoundment, or even during use, gas and solute transport, and feedbacks decrease in the water level may lead to between these processes. formation of an unsaturated zone, thereby opening the way for oxygen diffusion in 2.1.2 Oxidation processes pore spaces, and hence faster oxidation The slow oxidation of sulfide minerals leads processes such as in Equation 1. to release of sulfate, Fe(II), protons and The dependence the overall reaction rate on heavy metals and metalloids, as exemplified the availability of oxygen implies that by Equations 1-6 in Table 1 (see also, e.g., oxidation, and hence proton production, will reviews Nordstrom and Southam, 1997, and predominantly occur in the unsaturated zone Nordstrom and Alpers, 1999). Under most of tailings deposits. This is consistent with conditions in the presence of dissolved the commonly reported field observations of molecular oxygen, Fe(III) is the depletion of sulfides above the water table in thermodynamically more stable specie of the tailings deposits (e.g., Holmström et al., iron redox couple; however, kinetic 2001). The partial pressure of oxygen, Po2, limitation on the oxidation of Fe(II) results in the impoundment is a result of the in a relatively slow abiotic rate of oxidation balance between the rate of transport into to Fe(III) (Equation 10 in Table 1), pore spaces, by diffusion and advection, and particularly at low pH. The produced Fe(III)
5
Table 1. Typical slow geochemical processes in base metal ore/py unsaturated zone of Impoundment 1 in this study, and rate expressions. Process and reaction stoichiometry Rate lawa 1 Pyrite oxidation (oxygen path): −0.11 7 r = k [O (aq)]0.5 [H + ] FeS (s) + H O + O (aq) → Fe2+ + 2SO 2− + 2H + pyo pyo 2 2 2 2 2 4
2 Pyrite oxidation (ferric iron path): 3+ 0.62 3 2 2 r = k [Fe ] FeS +14Fe + + 8H O → 15Fe + + 2SO − +16H + pyf pyf 2(s) 2 4 ritic tailings, as exemplified by the weathering reactions used 3 Chalcopyrite oxidation (oxygen path): 2+ 2+ 2− - CuFeS 2 (s) + 4O2 (aq) → Fe + Cu + 2SO4
4 Chalcopyrite oxidation (ferric iron path): 3+ 0.43 3+ 2+ 2+ 2− + rcpf = kcpf [Fe ] CuFeS2 (s) +16Fe + 8H 2O → 17Fe + Cu + 2SO4 +16H 5 Sphalerite oxidation (oxygen path): 2+ 2− - ZnS(s) + 2O2 (aq) → Zn + SO4
6 Sphalerite oxidation (ferric iron path): 3+ 0.58 in the modelling of the 3+ 2+ 2− 2+ + rspf = k spf [Fe ] ZnS(s) + 8Fe + 4H 2O → Zn + SO4 + 8Fe + 8H 7 Chlorite (chlinochlore)b weathering: (Mg Fe II Fe III Al )AlSi O (OH ) (s) + 16H + → + 0.50 4.5 0.2 0.2 3 10 8 rch = k ch1 [H ] + k ch2 2+ 2+ 3+ 3+ 4.5Mg + 0.2Fe + 0.2Fe + 2Al + 3SiO2 (s) +12H 2O 8 Muscoviteb weathering: K Na ()Fe Al AlSi O (OH ) (s) + 10H + → + 0.40 0.8 0.2 0.1 1.9 3 10 2 rmu = kmu1 [H ] + kmu2 + + 3+ 3+ 0.8K + 0.2Na + 0.1Fe + 2.9Al + 3SiO2 (s) + 6H 2O c 9 Plagioclase weathering: + 0.45 + + 2+ 3+ rpl = k pl1 [H ] + k pl 2 Na0.75Ca0.25 Al1.25Si2.75O8 (s) + 5H → 0.75Na + 0.25Ca +1.25Al + 2.75SiO2 (s) + 2.5H 2O 10 Ferrous iron oxidation: 2+ rfe = [O2 (aq)](k fe1 [Fe ]+ 1 1 2+ + 3+ + Fe + O2 (aq) + H → Fe + H2O k [FeOH ]+ k [Fe(OH) ]) 4 2 fe2 fe3 2 a References given in Table 4 in Paper I. Rate laws were not found in the literature for oxidation of chalcopyrite and sphalerite by oxygen; in the modelling study, analogy was made with pyrite oxidation and rate constants were derived from reported rates (see Paper I). b Composition based on reported mineralogy of aluminosilicates in the Kristineberg mine (du Rietz, 1953). c Oligoclase; use of this plagioclase composition resulted in a solid phase chemical composition that most closely resembled that in Impoundment 1.
6 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale is also a powerful oxidant for sulfides O (Equations 2, 4, and 6 in Table 1) and is 2 itself reduced to Fe(II), leading to iron redox 1 cycling (Fig. 1). The availability of aqueous pyrite Fe(II) Fe(III) for reaction is potentially limited by relatively fast precipitation of secondary Fe(III) phases, such as amorphous ferric 2 O2 hydroxide: 3 3+ + Fe(III) Fe + 3H 2O ↔ Fe(OH )3 (s) + 3H (2) where the final position of equilibrium is a 4 function of, and may also affect, solution Fe(OH) (s) pH. Availability of ferrous and ferric iron for 3 slow reactions and equilibria, such as Equation 2, will also depend upon aqueous Figure 1. Iron redox cycling. Pyrite oxidation speciation of these components and by 1) dissolved molecular oxygen and 2) amounts of complexing agents present. aqueous ferric iron; 3) Oxidation of aqueous Table 3 in Paper I gives examples of ferrous iron by oxygen. Reaction aqueous speciation reactions for stoichiometry and abiotic rate laws are given components released by reactions such as in Table 1. 4) Solubility equilibrium between aqueous solution and Fe(OH)3(s), as those in Table 1. Modelling of pH and represented in Equation 2. aqueous component concentrations thus clearly requires consideration of coupling Nordstrom and Southam, 1997; and Nemati between slow kinetic and fast equilibrium et al., 1998), particularly at low pH where the processes. abiotic oxidation rate is relatively low. It has The slow oxidation of sulfide minerals is even been suggested that the overall rate of accelerated by mediation by acidophilic pyrite oxidation is limited by the rate of bacteria. Optimal conditions for microbial generation of ferric iron (e.g., Singer and mediation have been reported to be Stumm, 1970); particularly in microbial ~30 °C, Po2=0.2 atm, acidic pH, and with studies, this is still a topic of debate and nutrients such as phosphate, nitrate, and investigation (Fowler et al., 2001, Yu et al., carbon readily available (e.g., Ehrlich, 1996). 2001; Sand et al., 2001). Oxidation of In laboratory studies under such favourable aqueous Fe(II) and sulfide minerals may also conditions, pyrite oxidation by oxygen has be increased by abiotic factors; for example, been reported to be up to a factor 35 greater acceleration of aqueous Fe(II) oxidation has in microbial experiments than in abiotic been reported subsequent to adsorption of controls (Olson, 1991; Fowler et al., 2001, Fe(II) on to surfaces of iron oxy-hydroxides Yu et al., 2001). The acceleration of sulfide (Tamura et al., 1976; Wehrli, 1990), and oxidation by microorganisms is even utilised contact between two sulfide minerals can in industrial hydrometallurgical metal lead to accelerated dissolution of the sulfide extraction (e.g., Rossi, 1990). Observations with lower electrochemical rest potential made on the field (Elberling et al., 2000) and (e.g., Kwong, 1993, 2001). near-field (large columns; Strömberg and Rate laws quantifying the dependence of Banwart, 1999) scales at low temperature reaction rates, r, upon various key factors are also indicate higher rates of oxidation, by a given in Table 1 for slow, abiotic reactions factor 1.5-5, prior to addition of bactericides. of common minerals in mill tailings and The rate of oxidation of Fe(II) is also aqueous Fe(II). The rate laws are selected, increased by microbial mediation, by up to 4 where available, from studies reported in the orders of magnitude or more (Lacey and literature, where they have been determined Lawson, 1970; Singer and Stumm, 1970; from systematic laboratory experiments. As Nemati and Webb, 1997; see also reviews by can be seen, reaction rates [mol s-1] depend
7 S. Ursula Salmon TRITA-LWR PHD 1009
on factors such as the value of the with aqueous solution and buffer pH at temperature-dependent rate constant, the around neutral, e.g.: availability of oxidant and protons, and CaCO + H+ ↔ Ca2+ + HCO - (3). mineral surface area. 3(s) 3 Maintenance of pH at high levels, combined Although acceleration of oxidation reactions with release of components such as iron and in Table 1 by various mechanisms as aluminium, leads to precipitation of described above is widely reported in the secondary carbonates and hydroxides such literature, these alternative reaction pathways as amorphous Fe(OH)3(s) and Al(OH)3(s). are seldom quantified in a way that is useful After depletion of calcite, pH is buffered at for modelling (e.g., Edwards et al., 2000); successively lower levels by the dissolution additionally, there is no consensus as to sequence of the secondary minerals (Blowes which of the few rates laws that are available and Ptacek, 1994). Acid neutralisation by is best (Kirby et al., 1999). In the modelling dissolution of aluminosilicate minerals (e.g., study presented in Papers I and II (see also Equations 7-9 in Table 1) is relatively slow, Chapters 4-5), abiotic rate laws are applied but as these minerals are often present in and the sensitivity of model results to large quantities, the contribution to acid acceleration of rates is tested. Furthermore, attenuation may be significant, as has been systematic study of the relative contribution suggested in a number of field studies (e.g., of abiotic and biotic processes in tailings, Nesbitt and Jambor, 1998; Johnson et al., such as is required for determination of rate 2000) and in a recent water saturated column laws, requires sterilisation. However, the study with tailings (Jurjovec et al., 2002). effect on mineral weathering rates of Decrease of proton production with time, exposure to extreme chemical or physical due to, for example, eventual depletion of conditions usually associated with sulfide minerals in the unsaturated zone, will sterilisation has not been reported; this is lead to return of pH to higher levels (e.g., investigated in Paper IV. Banwart and Malmström, 2001). Selected rate laws for aluminosilicate minerals are given in Table 1. There is a 2.1.3 pH-buffering processes wealth of literature on the weathering rates As described above, processes such as of aluminosilicate minerals under laboratory oxidation of sulfides and precipitation of conditions, including the dependence of Fe(III) phases leads to release of protons in different mineral weathering rates on factors tailings impoundments. Natural attenuation such as pH and temperature (see e.g., review of the released protons and, thus, the final in White and Brantley, 1995). pH in the tailings impoundment groundwater, is dependent upon the reactions of minerals that dissolve and 2.2 Prediction of AMD on field scale consume protons, which in turn depend on the quantity of the mineral available for A number of previous modelling studies of reaction and the reaction rate. Acid- AMD from tailings deposits presented in the neutralising process rates are expected to be scientific literature focus on detailed higher in the unsaturated zone, where the modelling of oxygen diffusion, coupled to rate of protons production is higher, as a limited description of geochemical processes result of greater oxygen availability (e.g., (e.g., Jaynes et al., 1984; Elberling et al., Banwart and Malmström, 2001; Banwart et 1994; Scharer et al., 1994; Wunderly et al., al., 2002). 1996; Werner and Berglund, 1999; Bain et al., 2000; Romano et al., 2003). These Carbonate minerals, such as calcite, are models provide valuable insight into the generally considered to dissolve sufficiently coupling of, and feedback mechanisms quickly to maintain solubility equilibrium between, different physical processes and
8 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale oxidation of iron sulfide minerals. Other climate change (see e.g., White and Brantley, geochemical reactions, such as pH-buffering 1995, and references therein). Comparison processes, have often been simplified or of rates obtained from laboratory neglected (e.g., MEND, 2000; Parker and experiments with field observations typically Robertson, 1999), thus limiting the yields rates in the laboratory that are up to 2 possibility to investigate geochemical orders of magnitude greater than in the field processes that contribute to the master (e.g., Schnoor, 1990; White and Peterson, geochemical variable pH. 1990; Strömberg and Banwart, 1994; Detailed representations of kinetic Malmström et al., 2000). For example, limitations on slow geochemical processes application of rate laws obtained from the are presented in Lichtner (1996) and Mayer literature for weathering of aluminosilicates et al. (2003). Both studies use AMD and sulfides at the Aitik waste rock site in generation as one case study and, by fitting northern Sweden required a scaling down of to field observations, Mayer et al. (2003) ~2 orders of magnitude to reproduce determined diffusion constants for a absolute field rates (Strömberg and Banwart, shrinking core model of sulfide dissolution, 1994). Orders of magnitude differences were aluminosilicate mineral surface areas, and the also observed in weathering rates in Aitik reaction order of the rate of pyrite oxidation waste rock on laboratory, large column, and by Fe(III) with respect to ferric iron. The field scales; these differences were resolved thus calibrated model reproduced pH and by explicit account of independently aqueous concentration profiles and quantified differences in temperature, pH- furthermore simulated attenuation of Cu dependence, mineral content, and water flow below the groundwater level due to covellite conditions between different formation. As calcite was depleted, experimental/observation scales aluminosilicates were found to attenuate (Malmström et al., 2000). acidity. Geometric surface areas were Previous studies that have been briefly calculated for aluminosilicates, based on an outlined in this chapter have increased our average grain size of the tailings, and understanding of dominant geochemical compared with the surface areas obtained by processes that may occur in mill tailings and calibration; while some discrepancies were proposed different modelling tools for observed, the authors concluded that quantitative reproduction of these processes. laboratory-derived rate expressions for A number of qualitative and quantitative aluminosilicate minerals may be applicable in issues with regard to field scale prediction, tailings. however, have also been opened and remain In the work of Lichtner (1996) and Mayer et open, thus requiring further investigation in al. (2003), oxidation of aqueous Fe(II) was order to be resolved. The specific represented by equilibrium rather than a investigation objectives of this thesis, which kinetic processes; Lichtner (1996) explained have been summarised in Objectives 1-3 in this as an approximation to microbial the Introduction, include some of these mediation leading to a rates that approach open issues: equilibrium conditions. However, as - the applicability of literature discussed in the previous section, the degree monomineralic weathering rates laws of microbial activity in the field may be for reproduction of mineral limited by various environmental conditions, weathering rates in tailings; such as low temperature, high pH, and - the importance of aluminosilicate limited availability of nutrients. weathering and abiotic or microbially In parallel developments, a large number of mediated iron redox cycling for laboratory studies have been performed on tailings deposit groundwater aluminosilicate dissolution in order to composition and pH, and investigate processes such as acidification of soils and watersheds and long-term global
9 S. Ursula Salmon TRITA-LWR PHD 1009
- whether sterilisation of tailings samples, such as is required to investigate abiotic vs biotic weathering processes, affects mineral weathering rates.
10 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
of the case study used in the modelling 3 MILL TAILINGS CASE STUDY: application in Chapters 4 and 5 of this KRISTINEBERG IMPOUNDMENT 1 thesis. A summary of field data relevant for the modelling study is also given in Paper I. The case study used for model application is a mill tailings deposit called "Impoundment Impoundment 1 was remediated in 1997 by 1" in Kristineberg, which is also the special combined dry cover application and raised case study site of the Swedish multi- groundwater level. The remediation disciplinary research program "Mitigation of programme for the site is summarised in the environmental impact from mining Lindvall et al. (1999) and Werner and waste" (MiMi), funded by the Swedish Salmon (2000). Post-remediation Foundation for Strategic Environmental characterisation of the deposit to date is Research (MISTRA; see, e.g., MiMi, 2002). summarised in Holmström et al. (2001), Papers I-VI have been produced within the Werner et al. (2001), Corrége et al., (2001), framework of the MiMi research program. and Carlsson et al. (2002, 2003). The underground Kristineberg mine, owned As part of the present work, the BET surface area for tailings in Impoundment 1 and still operated by Boliden Mineral AB, is based on a massive sulfide Zn-Cu ore body (Papers I, III-V) and saturation indices for located in the Skellefte district 175 km both pre- (e.g., Paper I) and post- (Holmström et al., 2001) remediation south-west of Luleå, Sweden (Fig. 2). The average annual air temperature at the site is conditions were determined. These 1 °C with 5 months of average temperature quantities provide missing key parameter below 0 oC. The tailings deposit referred to values and contribute to the conceptual as “Impoundment 1” was in use from the model of geochemical processes occurring in 1940s to the 1950s for deposition of tailings Impoundment 1, which in turn is the basis from the enrichment plant at Kristineberg, for the desktop modelling study described in which processed ore from other mines in Chapters 4 and 5 (Papers I, II) and scaling of the area as well. The geology of the laboratory rates to field conditions in Kristineberg mine is described in du Rietz Chapter 7 (see also Paper V). Detailed (1953); field data for pre-remediation characterisation of the specific conditions in Impoundment 1 from Impoundment 1 tailings used in weathering sampling campaigns in the 1980s and early experiments was also performed; see Chapter 6 and Papers III-V. 1990s (e.g., Qvarfort, 1983; Axelsson et al., 1986, 1991; Ekstav and Qvarfort, 1989) have Impoundment 1 covers an area of 2 been compiled and interpreted by approximately 0.11 km and is on average Malmström et al. (2001) and form the basis 5 m deep. The extent of the unsaturated Kristineberg: Vormbäcken Sweden Impoundment 1
Rävlidmyr- bäcken Impoundment 4 Imp. 2 Imp. 3
Ditch
Waste Rock Heap Imp. 1B Tailings Water 1 km impoundment surface surface
Figure 2.. Location and overview of the Kristineberg site.
11 S. Ursula Salmon TRITA-LWR PHD 1009
infiltration recharge from adjacent slope 1m unsaturated zone 1.5m depth of groundwater Groundwater to sampling points 5m sampling plane depth of deposit
drainage to ditches drainage to etc fractures
Figure 3. Schematic illustration of the hydrological situation in our case study, “Impoundment 1”. The upper marked plane shows the average position of the water table (see Appendix in Paper I). The lower marked plane indicates the plane, 1.5 m below the groundwater surface, from which aqueous samples for the published groundwater quality data (Table 4) used in this study had been collected. zone, that is, the average depth to the The tailings within the deposit are pyritic groundwater, was 1 m prior to remediation. and carbonate-depleted; the main minerals The total water flow through the deposit, are chlorite, talc, muscovite, plagioclase, consisting of effective infiltration and quartz, and pyrite. Sphalerite and recharge from moraine slopes surrounding chalcopyrite are also present. More detailed the deposit, was ~47 000 m3 year–1. Figure 3 post-remediation characterisations of the summarises the main attributes of the water tailings confirmed this mineralogy and also balance over the deposit, along with the indicated the presence of feldspars and position of the groundwater table and the minor amounts of pyrrhotite, galena, and water sampling plane. arsenopyrite (e.g., Holmström et al., 2001). Neither concentrations of individual pore The tailings in the upper part of the deposit, gases, the influx of oxygen, the water on average down to the groundwater table, content of the unsaturated zone, nor the had been partly depleted in S, Fe, Zn, and oxygen diffusivity are reported for the pre- Cu, suggesting weathering of sulfides in the remediation conditions. water unsaturated zone. Volumetric The specific surface area of the tailings was fractions of major minerals in the determined using five-point, nitrogen gas unsaturated zone are given in Table 2. adsorption data evaluated through the BET The groundwater, which had been sampled equation, a method commonly used in 1.5 m below the groundwater table, had an kinetic mineral weathering studies. The average pH of 4.9; redox potentials are not specific surface area of eight samples from available in the literature. The groundwater seven locations within the deposit ranged contained high concentrations of dissolved 2 -1 2- 2+ 2+ 2+ 3+ from 0.2 to 10.1 m g , with an average SO4 , Fe, Mg , Zn , Ca , and Al , as well + 2+ value of 2.96 m2 g-1. The large variability in as lower concentrations of Na , Cu , and + specific surface area is consistent with the K (see Table 3). In order to investigate reported highly variable particle size whether aqueous concentrations were distribution (see Paper I and Malmström et controlled by fast dissolution or al., 2001). precipitation of secondary phases at solubility equilibrium with solution,
12 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
Table 2. Mineralogy of the unsaturated zone of Impoundment 1 as estimated for the modelling study (see Paper I).
Average volumetric fractions (γ) of minerals Sulfides Silicates Pyrite Chlorite } 0.04 } 0.45 Pyrrhotite Talc Chalcopyrite 0.002 Muscovite 0.15 Sphalerite 0.001 Plagioclase 0.10 Quartz 0.25
saturation index calculations were performed for the modelling study (see also Salmon, 2000, Salmon and Malmström, 2002) and on porewater composition sampled soon after remediation (Holmström et al., 2001) with the PHREEQC code (Parkhurst, 1995) in conjunction with the WATEQ4F database Table 3. Average concentrations (mol l-1) and (Ball and Nordstrom, 1991). The pH reported for Impoundment 1 (see Paper I calculations indicated that, of mineral phases and Malmström et al., 2001). likely to be controlling aqueous Average field Range of values concentrations in AMD environments values (Nordstrom and Alpers, 1999), the pH 4.87 4.05-6.15 groundwater was close to saturation with SO42- 0.10 0.07-0.14 respect to gypsum (CaSO4⋅2H2O(s)), Fe(tot) 0.080 0.06-0.09 amorphous silica (SiO2(am)), and an amorphous ferric hydroxide phase Mg2+ 0.011 0.007 - 0.017 2+ (Fe(OH)3(am)). Saturation indices also Zn 0.006 0.004 - 0.007 indicated that solution was close to solubility Ca2+ 0.005 0.002 - 0.012 equilibrium with gibbsite, as is often Al3+ 0.002 9 x 10-4 - 0.003 observed in AMD environment (e.g., Cu2+ 6 x 10-5 8 x 10-6- 2 x 10-4 Johnson et al., 2000; Jurjovec et al., 2002). K+ 6 x 10-5 3 x 10-6 -2 x 10-4 However, the presence of the precipitate in + -4 -4 -4 AMD environments has not been confirmed Na 4 x 10 2 x 10 - 5 x 10 (Jambor, 1994; Jurjovec, et al., 2002). Solution was also undersaturated with respect to amorphous aluminium hydroxide.
13 S. Ursula Salmon TRITA-LWR PHD 1009
14 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
4 MODELLING OF PRE-REMEDIATION IMPOUNDMENT 1
Based on analysis of the available data from pre-remediation conditions in Impoundment 1, BET surface determinations, and saturation index calculations for potential sources and sinks for the aqueous components in Impoundment 1, a conceptual model was formed, as is describe in this chapter. Solid phase data and the total water flow through the impoundment were used as model input data; process rates and equilibria were quantified using rate laws and thermodynamic data obtained from the literature. In Paper I (and this chapter), the resulting simulated aqueous component concentrations and pH were compared with those reported for Impoundment 1, and results were interpreted with respect to dominant geochemical processes.
concentrations in the groundwater suggested 4.1 Conceptual model dominance of Fe(II), as Fe(III) is highly In the modelling study of Impoundment 1, insoluble at the impoundment groundwater the unsaturated zone was considered to be pH of 4.9. This implied a kinetic limitation the major reactive zone in the tailings on the oxidation of Fe(II) to Fe(III). deposits and the dominant source of Aqueous ferrous iron oxidation by dissolved dissolved constituents in the groundwater. molecular oxygen was thus also For sulfide minerals, this was supported by conceptualised as a relatively slow, kinetic depletion above the groundwater level in process (cf. Strömberg and Banwart, 1994). Impoundment 1. Given the low spatial and Relatively fast, reversible geochemical temporal resolution of the available field processes included aqueous speciation, data, as a first approximation the solubility equilibrium between the aqueous groundwater composition was modelled by solution and secondary mineral phases assessing the unsaturated zone only; this (gypsum, ferrihydrite, and amorphous zone was conceptualised as a single, silicate), and Henry’s law equilibrium completely-mixed flow-through reactor between aqueous solution and oxygen and (box). The modelled leachate leaving the carbon dioxide in the gaseous phase in the unsaturated zone was compared with the unsaturated zone (see Table 3 in Paper I). available data for groundwater quality from The yearly average of the total water 1.5 m below the groundwater level (see Fig. flowrate through the deposit was used and 1). all physico-chemical properties, such as the The site-specific conceptual model for the water content, pore gas composition, and geochemical processes occurring in the mineral abundance, were assumed constant unsaturated zone of Impoundment 1 is over the modelled zone and in time. This shown in Figure 4. Based on the reported approach is consistent with the limited mineral abundance in Impoundment 1 spatial and temporal resolution of available (Table 2) and the composition of the site data for Impoundment 1 (Malmström et aqueous phase, the main primary sources of al., 2001) and for mine waste sites in general groundwater solutes were considered to be (cf. Banwart and Malmström, 2001). The the slow oxidation of sulfide minerals geochemistry was furthermore assumed to (pyrite, sphalerite, and chalcopyrite) by be at quasi-steady state (cf. Furrer et al., dissolved molecular oxygen and ferric iron 1989), as was later justified by the water and the slow weathering of aluminosilicates residence time in the saturated and chlorite, muscovite, and plagioclase unsaturated zones (0.5-1 years; Malmström (reactions presented in Table 1). Although et al., 2001) being much greater than the ferric iron is the thermodynamically mineral turnover time (see Chapter 4.3). favoured iron redox specie in the presence of dissolved molecular oxygen, the high Fe
15 S. Ursula Salmon TRITA-LWR PHD 1009
O (g) CO2(g) Impoundment surface 2
Fe2+ Fe(OH) (s) sulfides Fe3+ s
Po2 O2(aq) . zone CaSO4 2H2O(s) (alumino)- aqueous Unsaturated silicates speciation SiO2(s) Saturated zone Impoundment groundwater
Figure 4. Conceptual model for geochemical processes in the unsaturated zone of a mill tailings deposit. Kinetically controlled processes (single headed arrows) include weathering of aluminosilicate minerals (chlorite, muscovite, and plagioclase), oxidation of sulfide minerals (pyrite, sphalerite, and chalcopyrite) by both dissolved molecular oxygen and Fe(III), and oxidation of aqueous Fe(II). Fast, equilibrium controlled processes (double headed arrows) include aqueous speciation, Henry’s law equilibrium between the aqueous solution and pore gases, and dissolution/precipitation of secondary minerals.
Through consideration of the site specific 4.2 Process quantification properties of the unsaturated zone, the rate For each simulation, the partial pressure of laws were converted to rate expressions, R -2 -1 oxygen (Po2) was held constant, and the [mol dm s ], normalised to the surface area resulting flux (Fo2) determined by mass of the deposit; for heterogeneous reactions: balances. Simulations were performed over a Ri = hδAi ri (4) range of oxygen partial pressures; results are presented for i) surface reaction control where h and δ are the height and density of the unsaturated zone. For homogeneous (Po2=0.2 atm) and ii) mixed or transport reactions: controlled, Po2 <0.2 atm (see Chapter 2.1).
Fast processes in the conceptual model were Ri = hθri (5) quantified through mass action equations where θ is the water content. Table 4 in (see Table 3 in Paper I); Fe(II) and Fe(III) Paper 1 lists the final rate expressions, were specifically considered as separate together with associated rate constants that components, to allow investigation of slow have been adjusted for site temperature iron redox processes. The abiotic, kinetic using published activation energies. rate laws used were those given in Table 1, Additional parameter values for that is, the basis for quantification of all Impoundment 1 used in the rate expressions kinetic processes was independent of site are given in Table 2 in Paper I. observations. Rate laws were not found in the literature for oxidation of chalcopyrite The specific mineral surface area, Ai, to be and sphalerite by dissolved molecular used in conjunction with surface area oxygen; analogy was thus made with the rate normalised empirical rate laws in quantification of weathering rates (see Table law for oxidation of pyrite and rate th constants were calculated from reported 4 in Paper I), was for the i mineral (Scharer et al., 1994) abiotic rates of estimated by allocation of a fraction of the oxidation of these minerals by dissolved total specific surface area of the tailings molecular oxygen. (As,tot) in proportion to the volumetric fraction, γi, of the mineral:
16 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
Ai = γi As,tot (6). from laboratory to the field scale (e.g., White This modelling approach is based on the and Peterson, 1990; see also Malmström et al., 2000, and references therein). Release of assumptions that weathering rate laws for 2- 2+ 2+ monomineralic samples can be applied to Fe, SO4 , Zn , Cu , and pH were sensitive mixtures of minerals, such as in tailings, that to Po2. Decreased Po2 implied decreased minerals are evenly distributed over all the concentrations of these components, particle size fractions, and that the specific however, components from aluminosilicate surface area is constant between the minerals weathering still remained high, and pH for a given particle size. Volumetric fractions increased above the range of values reported for the field. (γi) for the case study are given in Table 2. In order to account for the scale dependence The mathematical model was implemented of mineral weathering rates from laboratory through STEADYQL (Furrer et al., 1989), values reported in the literature to field which allows a numerical, geochemical relevant rate values, a single calibration quasi-steady state box modelling approach factor, X =10-2, was applied to all weathering that previously has been successfully applied r rate expressions (i.e., multiplying the right to waste rock dumps and underground hand side of Equation 4). In the mines (Strömberg and Banwart, 1994; quantification of X , the concentration of Brown and Lowson, 1997; Brown et al., r Mg2+ was used as a tracer for the field 2000). Processes such as depletion and weathering rates, as Mg2+ originates from the accumulation of minerals that are much dominant (alumino)silicates, the weathering slower than the residence time of water can be rates of which do not depend (directly) upon considered through successive steady state the dissolved molecular oxygen simulations (see, e.g., Salmon, 2000, and concentration. A value of X =10-2 thus Chapter 5). r resulted in a Mg2+ concentration corresponding to that reported for the field. For all major components, the single 4.3 Model results -2 calibration factor Xr=10 sufficed to account for the scale-dependence of mineral 4.3.1 Comparison with field observations weathering rates, and resulted in modelled Simulated concentrations in the concentrations and pH that were close to impoundment groundwater were assessed in those reported for the field (Fig. 5b). The Papers I and II and are shown in Figure 5 vs possible physical reason for the difference concentrations observed in the field; the between absolute literature values of solid diagonal line indicates “perfect weathering rates, used in the model, and the prediction”. Uncalibrated results of the field weathering rates were further model described in the previous section investigated in the laboratory study in Papers (Po2=0.2 atm) are shown in Figure 5a, with III-VI (Chapters 6-7). With regards to the the exception that solubility equilibrium with field model, assumption of solubility gypsum is not assumed. The model equilibrium between solution and gypsum successfully reproduced the pH in the lead to improved reproduction of field Ca2+ 2- impoundment and the relative release rates, and SO4 concentrations (triangles in Fig. with the exception of Zn2+ and Ca2+. 5b), thereby supporting the inclusion of this Absolute rates were, however, one to two process in the conceptual model. Remaining orders of magnitude higher in the model discrepancies between model results and than in the field, despite accounting for field field observations are considered small, relevant temperature, pH, mineral content, given the uncertainty associated with, for and tailings surface area. This is consistent example, the limited availability of data from with the commonly observed scale- only a small number of observation points. dependence of mineral weathering rates
17 S. Ursula Salmon TRITA-LWR PHD 1009
-- 1 2- 2+ 3+ Mg SO4 0 Al a) Fe(tot) -1 + K+ Na -2 -3 Ca2+ Cu2+/10 -4 2+ Zn -5
-6 + free H -7 log model concentration [M] -6 -5 -4 -3 -2 -1 0 log average field concentration [M]
1 0 b) -1 2+ 3+ Mg 2- Al SO4 -2 Fe(tot) -3 + K 2+ -4 Cu /10 + 2+ Na Ca -5 2+ -6 Zn free H+ log model concentration [M] -7 -6 -5 -4 -3 -2 -1 0 log average field concentration [M]
Figure 5. Modelled concentrations of components (Paper I, Po2 = 0.2 atm) vs. reported (Ekstav and Qvarfort, 1989) average concentrations in Impoundment 1 groundwater (see Table 3). a) Initial model results; b) results after allowance for scale-dependent weathering rates through application of Xr (Xr = 0.02). The source of SO42- and Ca as indicated by diamonds is sulfide oxidation and plagioclase weathering, respectively; triangles in b indicate effect of assuming solubility equilibrium with gypsum, after application of Xr. The solid diagonal line indicates “perfect prediction”, where modelled and observed concentrations coincide. Vertical bars indicate the range of concentrations reported for the field.
The only major exception to the good and references therein; see also Chapter agreement between results of the calibrated 2.2.1 and Chapters 6 and 7). model and field observations was the Zn2+ concentration, where the modelled 4.3.2 Dominant processes and turnover times concentration was approximately three The calibrated model was analysed for orders of magnitude lower than the field dominant processes that contributed to the concentration (Fig. 5a and b). This suggested 2+ proton balance, oxygen consumption, redox that Zn release in the field was controlled cycling of iron, and composition of the by processes other than oxidation of groundwater in the deposit. For full oxygen sphalerite by oxygen or Fe(III), or that, for availability, the major processes contributing example, dissolution of sphalerite is to the proton balance were pyrite oxidation accelerated by galvanic interactions with by dissolved molecular oxygen (proton other sulfide minerals (e.g., Kwong, 2001, release) and chlorite dissolution (proton
18 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
consumption; Fig. 6a, filled bars). This K+, Na+, and Ca2+, as well as of Al3+ and Si. explains why the pH was not greatly affected Evolution of groundwater impoundment by application of Xr (compare Figures 5a geochemistry will depend on factor such as and b); the rates of the reactions dominating decrease in mineral content with time. proton production and consumption were Nonetheless, first-hand estimates of scaled by the same factor. As well as being turnover times for primary minerals in the the major source of protons, oxidation of unsaturated zone of the deposit can be made pyrite by dissolved molecular oxygen was by division of the amount of each mineral found to be the major source of Fe(II) and (Table 2) by their respective, modelled 2- SO4 and the major sink of dissolved dissolution rate (see e.g., Strömberg and molecular oxygen (Fig. 6). Pyrite oxidation Banwart, 1994; Banwart and Malmström, by ferric iron was relatively low (Fig. 6), due 2001) after application of Xr. For full oxygen to low solubility of Fe3+ at pH ~5. Ferrous availability (Po2 = 0.2 atm), pyrite had the iron oxidation was also low; Fe(II) was shortest turnover time of in the order of 100 found to be the dominant redox form of years. As pyrite is the major source of iron. protons, the characteristic turnover time of As well as being a major sink for protons pyrite also provides a first hand estimate of (Fig. 6), chlorite dissolution was also the the expected duration of AMD production source of Mg2+, the main source of Al3+ and (e.g., Banwart and Malmström, 2001). The Si, and a minor source of Fe(II) and Fe(III). characteristic turnover times for the other Dissolution of muscovite and plagioclase primary minerals in the model, that is, had only a minor effect on the proton sphalerite, chalcopyrite, and aluminosilicates, balance (not shown), but were sources of were in the order of 103-104 years. As Zn2+
i) Rfeo
ii) Rpyo
iii) Rpyf
iv) Rch
v) Outflow
vi) Fe(OH)3(s) a) b) c) d) e) vii) Gypsum
-0.8 0.0 0.8 -0.8 0.0 0.8 -0.8 0.0 0.8 -0.1 0.0 0.1 -1.0 0.0 + 2- 2 2 2 H SO4 Fe(II) Fe(III) O2(aq)
Fig. 6. Modelled fluxes of components from different processes normalised to the deposit surface area [µmol m-2s-1], of a) H+; b) SO42-; c) Fe(II); d) Fe(III); and e) O2(aq) for a fixed Po2 of 0.2 atm (filled bars) and 0.002 atm (open bars). Processes are i) ferrous iron oxidation, ii) pyrite oxidation by oxygen, ii) pyrite oxidation by Fe(III), iv) chlorite dissolution. Processes vi)-vii) show the effect of solubility equilibrium between aqueous solution and the given secondary phase.
19 S. Ursula Salmon TRITA-LWR PHD 1009
release in the model underestimated Zn2+ turnover time of sphalerite is potentially release in the impoundment by much lower. approximately 3 orders of magnitude, the
20 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
5 MODEL SENSITIVITY AND REMEDIATION IMPLICATIONS The sensitivity of model results to uncertainties such as may arise from natural variability in physico-chemical characteristics of the deposit and individual minerals, uncertainty in the experimentally determined kinetic parameters, and possible unquantified parallel reaction pathways, is presented in the following Chapter 5.1 (see also Paper I, as well as Salmon, 2000, and Salmon and Malmström, 2002). In Chapter 5.2, the model is used to approximate the evolution of the impoundment groundwater geochemistry with time under unremediated and remediated conditions (e.g., see Paper II); implications of this for Impoundment 1, and modelling of tailings AMD in general, are highlighted.
to an increase in Rpyf (compare Fig. 8a and b) 5.1 Sensitivity analysis and a decrease in the overall pH (Fig. 7b). 3 For β in the order of 10 , Rpyf, was the 5.1.1 Oxidation processes dominant source of protons (Fig. 8b). Abiotic rate expressions were applied in the The rate of pyrite oxidation by Fe(III) was model for pyrite oxidation by O2(aq) (Rpyo) and also increased with an increase in Rpyo. The Fe(III) (Rpyf), as well as Fe(II) oxidation (Rfeo; higher rate of proton production associated see Reactions 1, 2, and 10 in Table 1; see with a value of α as small as 3.5 was not O2 also Fig. 1). However, as discussed in compensated by an increase in the kinetically Chapter 2, these oxidation processes may be limited chlorite dissolution rate, but by accelerated by a variety of processes; in the increased Fe(OH)3(s) dissolution to maintain absence of rate laws explicitly defining the solubility equilibrium with solution (Fig. 8c). rates of such parallel reactions, the sensitivity 3+ Dissolution of Fe(OH)3(s) also released Fe , of model results to such mediation was and by this means lead to increased tested by application of factors α O , α Fe , and 2 importance of Rpyf,, which then contributed β to Rpyo, Rpyf,, and Rfeo, respectively. The up to ~ 40% and ~50% of the total proton values of α , α , and β were varied over a (Fig. 8c) and Fe(II) release, respectively. For Fe O2 a given value of α > 1, removal of the range of values that corresponds to the O2 effect of microbial mediation under condition of solubility equilibrium between favourable laboratory conditions, as reported Fe(OH)3(s) and solution led to lower pH (e.g., ° in Fig. 5b; α = 3.5), as then determined in the literature (see Chapter 2.1.1). Testing O2 in this manner also reveals sensitivity to by the balance between pyrite oxidation and uncertainty in rate expressions, for example aluminosilicate dissolution only. Under these due to choice of rate law (Salmon and conditions, the contribution by the ferric Malmström, 2002), mineral surface area iron pathway for pyrite oxidation was (mineral content and tailings specific surface insignificant, as there was no readily area, cf. Equation 6), rate constant, and available source of ferric iron. activation energy (Ea). All simulation results in Figure 7a-d are Model results indicated that the overall rate plotted in Figures 7e and f as a function of of pyrite oxidation is the result of complex Fo2. Irrespective of the values of α O , α Fe interactions between the processes depicted 2 and β and dominance of R or R , SO 2- in Figure 1, pH, and O availability, and also pyo pyf, 4 2 concentrations are determined by Fo (Fig. depends upon the presence/absence of 2 7e), as is consistent with conclusions of, e.g., Fe(OH) . For example, acceleration of 3(s) Elberling and Nicholson (1996). However at pyrite oxidation by Fe(III) (i.e., R with pyf, low Fo , SO 2- release is determined by α >1) alone had almost no effect (not 2 4 Fe gypsum dissolution. The pH, on the other shown). However, acceleration of Fe(II) hand, was sensitive to buffering processes, oxidation (β>1; Fig. 7a-b, Po2=0.2 atm) lead for example, whether solution was at
21 S. Ursula Salmon TRITA-LWR PHD 1009
8 1.00 7 6 5 0.10 pH 4
concentration [M] concentration 3 2- 4 2 a) b) SO 0.01 1 1 10 100 1000 10000 1 10 100 1000 10000 β β 8 1.00 7 6 5 0.10 pH 4
concentration [M] 3 2- 4 2 c) d)
SO 0.01 1 1 10 100 1000 10000 1 10 100 1000 10000 β β 8 1.00 7 6 5 0.10 pH 4
concentration [M] concentration 3 2- 4 2 e) f) SO 0.01 1 1E-7 1E-6 1E-5 1E-4 1E-7 1E-6 1E-5 1E-4 -2 -1 -2 -1 F [moles m s ] FO [moles m s ] O2 2 2- Figure 7. Modelled SO4 concentrations and pH as function of β (panel a-d) and Fo2 -2 -1 [mol m s ] (panel e and f) with α Fe = 1, where β and α Fe represent the magnitude of acceleration of ferrous iron oxidation and the ferric iron pathway for pyrite oxidation, respectively. Filled symbols indicate Po2 = 0.2 atm, open symbols indicate Po2 = 0.002 atm. Symbol shapes indicate the value of α : o =1, □ =3.5, ◊ =10, ∆ =35. Black crosses indicate O2 simulations without solubility equilibrium between aqueous solution and Fe(OH)3(s) (α = 3.5, O2 Po2 = 0.2 atm). solubility equilibrium with Fe(OH)3(s), and to for prediction of impoundment groundwater values of α and β, even at low Fo (Fig. composition and pH. O2 2 7f). 5.1.2 pH-buffering processes Model results, and in particular, the proton balance and pH, are thus sensitive to Given the importance of pH for the relatively small changes in the rate of pyrite environmental impact of AMD, model oxidation, irrespective of whether the results were tested for sensitivity to changes are due to, for example, microbial uncertainty in pH-buffering processes, mediation or uncertainty in mineral surface namely, chlorite dissolution and the area. This implies that more detailed condition of solubility equilibrium between quantification of microbial processes on the aqueous solution and Fe(OH)3(s). To start field scale, and the surface area of the sulfide with, simulation of pH-buffering by minerals available for reaction, is important aluminosilicate weathering alone, by removal of the forced solubility equilibrium between
22 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
α O2 1 1 3.5 3.5 β 1 1000 1 1000 pH 4.9 4.1 4.1 3.7
Rfeo
Rpyo
Rpyf,
Rch
Fe(OH)3(s) a) b) c) d)
-20 0 20 -20 0 20 -20 0 20 -50 -25 0 25 50
Figure 8. Effect of acceleration of pyrite oxidation by O2(aq) (Rpyo) and Fe(II) oxidation (Rfeo), through application of given values of α and β, respectively (Po2=0.2 atm), on rates of proton O2 release and consumption [µmol m-2s-1] by various processes and pH. Process notation as in Figure 6.
Fe(OH)3(s) and the aqueous phase, had little the pH at ~4.3, but also implied faster effect on pH and component depletion of Fe(OH)3(s) (compare Fig. 9a-c). concentrations, as Fe(OH)3(s) dissolution was Removal of both aluminosilicate weathering not a major process (cf. Fig. 6a). and Fe(OH)3(s) dissolution from the model, The sensitivity of the groundwater that is, simulation of the conditions after the composition and pH to uncertainty in the depletion of Fe(OH)3(s) without chlorite rate constant and/or chlorite consideration of aluminosilicate weathering, content was tested by increasing or resulted in a pH below 2. decreasing the rate expression for chlorite These simulations indicated that, while Fe 2- weathering, Rch, by a factor αch, the and SO4 concentrations and pH in magnitude of which was varied by up to an Impoundment 1 could be reproduced order of magnitude. The impact of this on without aluminosilicate dissolution as long the proton balance and pH is shown in as fast-dissolving minerals were present in
Figure 9. Increase in Rch by as little as a the model, accurate representation of a) the factor of two led to an increase in pH to proton balance, and thus the rate of around neutral (Fig. 9f). Further increase in dissolution of Fe(OH)3(s) and duration of
Rch lead, as expected as a result of the pH-buffering by this phase, and b) the pH associated increase in proton consumption, after depletion of Fe(OH)3(s), required to further increase in pH (Fig. 9e and f), and consideration of aluminosilicate dissolution. correspondingly higher concentrations of Furthermore, neglect of aluminosilicate Mg2+ and Al3+. weathering reactions excluded the possibility to quantify the primary sources of A decrease in Rch by a factor of two led to a 3+ 2+ + + drop in pH by about half a unit (Fig. 9f). components, such as Al , Mg , Na , K , and Ca2+, which are released by these With further decrease in Rch, the influence of the chlorite weathering rate on the proton processes, and which may participate in balance became insignificant other geochemical reactions subsequent to (Fig. 9a). As for increased α in Chapter release. O2 5.1.1, the expected decrease in pH was mitigated to an extent by increased dissolution of Fe(OH)3(s). This maintained
23 S. Ursula Salmon TRITA-LWR PHD 1009
Rfeo Rpyo Rpyf, Rch Outflow a) b) c) d) e) Fe(OH)3(s) -50 -40 -30 -20 -10 0 10 20 30 -10 0 10 -10 0 10 -10 0 10 -10 0 10
8
7
pH 6
5 f) 4 0.1 0.5 1 2 10 α ch
- Figure 9. Effect of application of αch on the balance of proton release and consumption [µmol m 2 -1 s ] for a) αch= 0.1, b) αch= 0.5, c) αch= 1, d) αch= 2, e) αch= 10, and on f) pH.
conditions of mineral content were 5.2 Implications for Impoundment 1 16 vol-% pyrite, which was the pyrite The box model presented in Chapter 4 was content of the saturated zone of used to provide a first hand approximation Impoundment 1, 10 vol-% calcite, as was of the temporal evolution of the reported for the Kristineberg mine impoundment geochemistry, in terms of (Qvarfort, 1983), and other minerals in the processes that have occurred since same proportions as given in Table 2. deposition and potential future evolution of Conditions were kept constant for each a tailings impoundment under continued steady-state period; if solution was unremediated conditions (Chapter 5.2.1 and oversaturated with respect to a phase likely Salmon, 2000), as well as potential effects of to be at solubility equilibrium with aqueous changed conditions such as may be caused solution in AMD environments (Nordstrom by remediation (Chapter 5.2.2 and Paper II). and Alpers, 1999), or if such a phase had precipitated at an earlier stage, solubility 5.2.1 Geochemical evolution of Impoundment 1 equilibrium between this phase and aqueous Indications from saturation index solution was assumed. The duration of each calculations on field data (Chapter 3) were stage was determined by the turnover time of the mineral that mass balances indicated that Fe(OH)3(s) and gypsum exerted solubility control on aqueous solution; the would be the first to be consumed. After modelling study presented in Chapter 4 each stage, remaining or accumulated indicated that these phases were dissolving amounts of minerals were calculated and to maintain solubility equilibrium with used as input conditions for the next steady- aqueous solution, raising questions with state period (for mathematical formulation, regards to where these phases could have see Salmon, 2000). originated from. It is not implied that accurate predictions of The temporal evolution of the the temporal evolution of the impoundment impoundment geochemistry was therefore geochemistry can be made with these simple approximated by considering consecutive calculations; however, this approach steady-state periods, or stages. Initial provides a qualitative insight into the time
24 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
1 8 iii) Al(OH)3 SO 2- Stage i) 4 0.9 Fe(OH)3 calcite Fe(II) 7 0.8 gypsum 3+ siderite Al 6 ) Al(OH)3 ii) siderite pH -1 0.7 iv) Fe(OH)3 Fe(OH) Al(OH) 0.6 3 3 gypsum 5 gypsum Fe(OH)3 gypsum
0.5 4 pH 0.4 v) gypsum 3 0.3 2 0.2 Concentration (mol l 0.1 1 0 0 0 1020304050 Time (years)
Figure 10. Evolution of modelled pH and concentrations of SO42-, Fe(II), and Al(III) with time. Mineral phases noted are those at solubility equilibrium with aqueous solution during the various different stages. Stage v) is estimated to continue for another ~120 years, after which gypsum will be totally depleted. Pyrite oxidation is estimated to continue for an additional ≈ 103 years. Parameter values: α =α =3.5; β =1: flowrate, q=7.7x10-8 dm s-1; Po2= 0.1 atm. O2 Fe evolution of the processes controlling the maintain solubility equilibrium with solution pH and the concentrations of major would be greater in the absence of proton components. attenuation by aluminosilicates. Modelling Simulations resulted in a sequence of without consideration of aluminosilicates precipitation and dissolution reactions and would in a scenario such as that depicted in associated consecutive drops in pH (Fig. 10) Figure 10 lead to underestimation of the similar to that which has previously been time to reach acid pH, and furthermore, the described in the literature (e.g., Blowes and simulated final low pH, as determined only Ptacek, 1994). Buffering of the pH was due by pyrite oxidation and iron redox processes, to the fast dissolution of, in order of would be lower. Implications for the geochemical modelling of the groundwater succession, calcite, siderite (FeCO3(s)) and aluminium and ferric iron hydroxides. After composition of Impoundment 1 include that depletion of ferric iron hydroxide, the pyrite it may be plausible that a sufficient amount content was 11 vol-%; the pH was of Fe(OH)3(am) and gypsum accumulated in controlled only by the relative rates of pyrite Impoundment 1 to be the dissolving at the oxidation and aluminosilicate dissolution, point in time represented by the case study, resulting in a pH below three. The lifetime ~30 years after deposition. of the pyrite in this simulation, and hence the duration of the acidic pH, was in the 5.2.2 Effect of remediation order of 1000 years. After depletion of Following a post-remediation investigation pyrite, the modelled pH returned to above of the performance of a composite soil neutral (not shown). cover applied on parts of Impoundment 1, As indicated in Chapter 5.1.2, the rate of Werner et al. (2001) reported an oxygen flux dissolution of fast-reacting minerals, such as of ~ 1 x 10-8 mol m2 s-1 and a water carbonates and hydroxides, required to infiltration rate of <1 x 10-9 m s-1. These values correspond, compared to the pre-
25 S. Ursula Salmon TRITA-LWR PHD 1009
remediation situation assessed in the case study, to a 1-2 order of magnitude decrease in both the oxygen and water fluxes. As shown in Figure 7e and f, such a dramatic decrease in the oxygen influx would be sufficient to slow down pyrite oxidation to a level where the resulting pH is above 7. Inclusion of the kinetic representation of Fe redox processes allowed testing of the hypothesis that, with decreasing oxygen availability, the oxidation of pyrite by ferric iron would dominate over the oxygen path; however, this hypothesis was not supported by the model results, even in simulations with the maximum degree of acceleration that is reportedly possible due to microbial mediation (α =α = 35, β = 104 - 106). Simulations O2 Fe with low Fo2 and in the absence of aluminosilicate dissolution resulted in lower predicted pH, of ~5, as controlled by solubility equilibrium with Fe(OH)3(s). Decrease in the water flux alone, with the fixed partial pressure of oxygen of 0.2 atm, had little effect on the pH, but did, however, result in higher concentrations of dissolved components (Paper II). As the impoundment has only relatively recently been remediated, the long-term efficiency of the applied techniques has not been assessed. However, comparison of release rates in saturated column experiments with Impoundment 1 tailings (Paper VI) with pre-remediation release rates (see Chapter 7) indicated that sulfide oxidation rates do decrease under saturated conditions. Holmström et al. (2001) and Corrège et al. (2001) also report positive trends in the impoundment water quality a few years after remediation.
26 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
6 QUANTIFICATION OF ELEMENT RELEASE RATES FROM BATCH EXPERIMENTS
The modelling study of Impoundment 1 indicated that mineral weathering rates from literature, based on experiments on freshly prepared, monomineralic samples, overestimated rates observed in the field by 1-2 orders of magnitude, despite consideration of dependence of rates on pH, temperature, mineral content, and (BET) surface area of tailings, and differences in these physico-chemical conditions between the laboratory and the field. This chapter presents results of weathering experiments that were then performed on different tailings samples and sulfide minerals in order to test whether literature rates and rate dependencies were applicable for weathering of minerals in tailings (Papers III-V). In addition, the effect of application of various sterilisation techniques, as is necessary to obtain abiotic weathering rates in tailings samples, on the weathering characteristics of tailings were tested (Paper IV). The experimental method is only briefly described below; Papers III-V are referred to for details.
sections). The particle size distribution of 6.1 Materials and methods the tailings was determined with a laser granulometer, and the specific surface area Tailings were obtained from Impoundment of all samples was determined by evaluation 1 in August 1999 from near the original of 5- and/or 1-pt N2 adsorption data. The tailings dam wall, far from the original results of the mineralogical determinations tailings slurry discharge point (“fine tailings” were consistent with other determinations sample) and in August 2000 from near the for the major minerals in Impoundment 1 as original discharge point (“coarse tailings” reported in Malmström et al. (2001), sample); this distribution of fine and coarse Holmström et al. (2001), and Carlsson et al., particles is typical for tailings impoundments 2002); based on the mineralogy and total (e.g., Robertson, 1994). A sulfide chemical composition, the mineral content concentrate was obtained from the coarse of the tailings samples was estimated (Table tailings by decanting of fine particles, 4; see Paper V). gravimetric separation of the heavy fraction, Table 4. Estimated mineral content and and sieving to 0.035-0.5 mm size. Pyrite BET surface area crystals were purchased (geographic origin unknown), crushed in a ceramic mortar and Fine Coarse tailings tailings pestle and sieved to obtain a 0.125-0.25 mm fraction. Both sulfide concentrate and pyrite Pyrite 18 18 crystals were treated briefly with ultrasound Chalcopyrite 1 0.2 in order to remove fine particles, which may Sphalerite 0.6 1 otherwise result in unrepresentative high Quartz 28 40 initial release rates (e.g., Nicholson, 1994). Chloritea 26 25 The mineralogy of the 2 tailings samples, the Muscoviteb 14 7 sulfide ore concentrate, and the pure pyrite Plagioclasec 12 6 was determined by X-ray diffraction (XRD), Surface area 10.0 1.2 2 -1 and samples were analysed for total chemical [m g ] composition. Tailings samples were further aChlorite and talc represented as chlinochlore, characterised by infrared spectroscopy (IR) (Mg5Al)Si3AlO10(OH)8 and optical microscopy (including point bKAl2(AlSi3)O10(OH)2 counting of opaque minerals in thin cNa0.55Ca0.45Al1.45Si2.55O8
27 S. Ursula Salmon TRITA-LWR PHD 1009
Table 5: Experimental conditionsa
B4b C-series G-seriesd E-series Reactor C1 C7 C8c C12 G1 G2 E1 E2 E3 Material Fine Fine Fine Fine Pyrite Coarse Coarse Sulfide Sulfide Sulfide tailings tailings tailings tailings tailings tailings conc. conc. conc. pH ~3 ~3 ~2 ~3 ~3 ~2 ~2 ~2 ~3 ~5
Acid medium HNO3 H2SO4 H2SO4 H2SO4 H2SO4 H2SO4 HNO3 HNO3 HNO3 NaHCO3 Start Dec‘99 Nov’00 Nov’00 Dec’00 Dec’00 Nov’01 Nov’01 Jun’02 Jun’02 Jun’02 Duration 89 480 480 343 455 100 100 35 35 35 [day] aSee also Table 4, Paper V bPreliminary results reported in Salmon and Malmström (2001; Paper III) c5°C; other experiments at ~23±1 °C. d Preliminary results reported in Salmon and Malmström (2003; Paper IV)
The higher pyrite content in the tailings and air-sparged. The pH and (in all but B4) samples (~18 vol-%; Table 4), which were redox potential were measured either in collected from the saturated zone, than was filtered samples or in reactors. Direct estimated for the unsaturated zone (~4 measurement of redox potential in fine
vol%; Table 2) is consistent with other tailings reactors with H2SO4 (C1, C7, C8), reported solid phase characterisation of the and on last 2 sampling occasions in the pure saturated zone (e.g., Malmström et al., 2001; pyrite reactor (C12) lead to influx of the
Holmström et al., 2001). The difference in redox electrode electrolyte (KNO3) to the mineral contents between the water reactor solution. Redox measurement over saturated and unsaturated zone is likely due an extended period of time on Day 74 in C7 to partial depletion of sulfides in the upper and Day 94 in C1 directly preceded an 2- part of the impoundment (see Chapter 3 increase in the SO4 release rate in these and 4.1). reactors, as is discussed in the results section The XRD analyses confirmed that the pyrite below (see also Paper V). For the C-series sample was pure within the detection limits experiments, the solid phase was collected of the XRD (~5 %). The sulfide concentrate after closure of the experiments for was found to consist predominantly of mineralogical analysis using XRD and IR. pyrite with some traces of feldspar, Samples were analysed for concentrations of 2- pyrrhotite, and possibly sphalerite. SO4 and major cations (see Papers III-V for analysis techniques). Calculation of saturation indices indicated that, for the 6.2 Effect of pH, temperature, and majority of components during the period of surface area time over which release rates are derived, aqueous concentrations were generally not 6.2.1 Weathering experiments controlled by solubility equilibrium with typical phases for AMD, with the major Weathering experiments were performed on exception of iron. fine and coarse tailings, sulfide concentrate, and pure pyrite over 89-480 days, at different conditions of pH, temperature, and acid medium; selected conditions are given in Table 5. Approximately two to six grams of samples were placed in 200 or 400 mL solution. Reactors were continually agitated
28 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
4.E-03 a) fine tailings
25 °C, pH ~3, HNO3
release [mol/g] 2.E-03 25 °C, pH ~3, H2SO4 2- 4 25 °C, pH ~2, H2SO4 5 °C, pH ~3, H2SO4 Acc. SO
0.E+00 0 100 200 300 400 time [day]
8.E-05 c) H SO medium b) 25 °C, H2SO4 medium 2 4 2.E-04
coarse tailings, pH ~2 4.E-05 1.E-04 fine tailings, pH ~2 fine tailings, pH ~3 25 °C, coarse tailings, pH ~2 25 °C, fine tailings, pH ~3 Acc Zn release[mol/g]
Acc Mg release [mol/g] 5 °C, fine tailings, pH ~3 0.E+00 0.E+00 0 100 200 300 400 500 0 100 200 300 400 500 time [day] time [day]
2- 2+ 2+ Figure 11. a) Accumulated release of SO4 in fine tailings experiments, and b) Mg and c) Zn in fine and coarse tailings experiments.
initially present in the solid phase in each 6.2.2 Results reactor had been released to solution, Sulfate release in fine tailings under suggesting that the main S-source, pyrite, different conditions of pH, acid medium, was completely or effectively depleted at this and temperature are shown in Figure 11a. point or soon after. The greater delay before For fine tailings at 25 °C, two release rate the accelerated rate in the H2SO4 reactors regimes were observed. In the experiment may be due to an extended lag phase after
with pH ~3 in HNO3, performed soon after approximately a year of storage, and/or + - collection of tailings, a typical pattern for addition of K and NO3 , possibly acting as bacterial leaching (e.g., Rossi, 1990) was nutrients for the micro-organisms, on the observed, with a fast release rate, or preceding sampling occasion (see above). “accelerated” release, after a lag phase of ~8 The similar release rates at pH ~2 and pH days (Fig. 11a). By contrast, in later ~3 in both the accelerated and non- experiments with the same tailings and pH, accelerated period suggests low to negligible 2- but in H2SO4 medium, the SO4 release rate pH-dependence of the dominant S-releasing remained approximately constant until ~Day process.
96, after which a similarly accelerated release In the reactor with coarse tailings in HNO3 2- was observed. A similar pattern was also (not shown; see Paper V), the final SO4 observed in the pH 2 H2SO4 reactor (Fig. release rate was by contrast somewhat lower 11a). After the period of high release rate in than the initial rate; the mass-normalised all reactors, approximately 70-90% of the S release rate [mol g-1 day-1] was also lower
29 S. Ursula Salmon TRITA-LWR PHD 1009
than the “non-accelerated” rate in the fine depletion of fine particles and/or more tailings. The coarse tailings did not exhibit a reactive sites on freshly exposed mineral 2- period of accelerated SO4 release in either surfaces (e.g., White and Brantley, 1995). 2+ HNO3 or H2SO4 medium. A possible Indications were that initial Ca release in explanation for the accelerated release coarse tailings was affected by gypsum period in the fine tailings but not in the dissolution (see Paper V); if this process coarse is that in the absence of other occurred in the fine tailings, it did not nitrogen sources, the low levels of nitrate in appear to affect Ca2+ release rates. the fine tailings reactors were utilised as a N For fine and coarse tailings and the sulfide source (Erlich, 1996, and references therein). concentrate, Zn2+ and Cu2+ release was At higher concentrations, however, nitrate initially fast, then decreased to level off has been reported to inhibit microbial before 100% was released to solution (e.g., activity (e.g., Harahuc et al., 2000); it is also 2+ 2- Zn in Fig. 11c), irrespective of SO4 reported to have greater toxicity at lower pH release pattern, acid medium (HNO3/ (Alexander et al., 1987), and may therefore H2SO4), or temperature (5/25°C). In the have had an inhibiting effect on microbial reactors with fine tailings at 25°C, the activity in the coarse tailings reactor with remaining Zn2+ was released to solution at HNO . Over the course of the coarse 2- 3 the same time as the accelerated SO4 tailings experiment (~100 days), less than release rate; the accumulated Cu2+ release 10% of the S in the coarse tailings was was not affected. Both Zn2+ and Cu2+ release released to solution. rates displayed low dependence on pH but
In the reactor with fine tailings (H2SO4) at high dependence on temperature. A 2- 5°C (Fig. 11a), the SO4 release rate was decrease in leaching rates with time is lower than that at 25°C, and constant over commonly observed in leaching experiments the ~340 days of the experiment. Sulfate of chalcopyrite and sphalerite, and is release rates in the sulfide concentrate suggested to be due to, for example, reactors (not shown, see Paper V) were decrease in surface area due to reduction of constant with time, over the ~40 days of particle size, or formation of a leached layer experiments. Release in the pure pyrite on the mineral surface such that dissolution reactors was constant for the first ~150 becomes transport-limited. A shrinking days, and also for the latter 300 days of the radius model relatively successfully explained experiments, but at a lower rate. the Zn2+ and Cu2+ release (Papers III and V). 2- The increase in SO4 release in all fine In summary, release rates of base cations tailings reactors was accompanied by a drop and Al3+ and Si were affected by pH, which 2- 2+ in pH, by up to 1 pH unit in rectors at pH 3, was not the case for release of SO4 , Zn , and an increase in redox potentials (not or Cu2+. Some fine tailings reactors exhibited shown). In reactors where pH decreased by periods of accelerated sulfate release, which up to one pH unit over the experiments, an was tentatively assigned to microbial increase in the release rate of base cations mediation. In experiments with relatively was observed (e.g., Fig. 11b; see also Fig.3 constant pH, initial release rates of, for in Paper V). However, in reactors at pH ~2, example, Mg2+ were higher than final release where the pH remained relatively constant, rates. Release rates [mol/g/day] were base cation release generally decreased with generally lower in coarse tailings than in fine time (Fig. 11b), a phenomenon commonly tailings, and at low temperature than at high observed in aluminosilicate weathering temperature. experiments (e.g., Malmström and Banwart, 1997; Brandt et al., 2003; Gustafsson and Puigomenech, 2003), and suggested to be due to processes such as preferential leaching of metal ions, weathering and
30 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
Table 6. Applied sterilisation technique and degree of microbial activity.
Degree of Phase Reactora Tailings treatment microbial activityc I G1 Phase I control - untreatedb high G2 Phase I control - untreated high G3 autoclave none G4 repeated heating to 70°C none G5 ethanol none II G6 antibiotics mixture medium G7 γ-irradiation – 10 kGy none G8 γ-irradiation – 25 kGy none G9 rinsed with distilled watere high G10 Phase II control- untreated high aReactor name in weathering experiments. b Solution was 0.05 M H2SO4, as opposed to 0.1 M HNO3 as used in the other reactors. cDetected after 30 days (Herbert et al., 2003).
6.3 Effect of sterilisation 6.3.2 Results In the sterilisation experiments, the BET 6.3.1 Experimental method surface area of the sterilised samples was Weathering experiments and BET surface found to be on average slightly lower than in area determinations were performed on the untreated samples, possibly due to, e.g., coarse tailings sub-samples that had been loss of fine particles during the sterilisation subjected to various traditional and novel treatments. However, the mineralogical and methods of sterilisation (see Table 6, Paper chemical analysis of the tailings, as IV and Herbert et al., 2003). Sterilisation was mentioned above, indicated that the bulk found to have little effect on the bulk composition did not vary more between mineralogy or total chemical composition; treated and untreated samples than between however, oxidised sulfur surface species and untreated controls. Release rates in the non-hydrocarbon surface C, as well as sterilisation experiments generally followed ascorbate-extractable metals, increased as a the same pattern as in the untreated controls result of some treatments (see Herbert et al., (coarse tailings reactors G1 and G2, see 2003). Chapter 6.2), with the exception of reactors Procedures for weathering experiments were where inspection by fluorescent microscopy similar to in the other experiments (Chapter revealed the presence of large amounts of 6.2.1) with the exception that measures were organic filaments, probably fungal. taken (see Paper IV) to prevent microbial Filaments were only observed in reactors contamination of reactors. Extra samples where sterilisation of tailings had not been were also collected from all reactors for performed or was reportedly unsuccessful staining with 4’6-diamidino-2-phenylidole (see Table 6). (DAPI). This process leads to fluorescence Final and initial element release rates, of DNA and organic material under UV obtained from linear regression on element light, and is generally used for total count of release data in the reactors containing viable bacteria in aqueous solution. sterilised (G3-9) and untreated controls and normalised to the BET-surface area of each sample, are given in Figure 12. In both treated and untreated samples, initial rates
31 S. Ursula Salmon TRITA-LWR PHD 1009
1.E-07
1.2E-05 ] -1 --- ] b) -1 a) initial rate day -2
day final rate -2
6.0E-06 5.E-08 rate Cu rel. [mol m rate S rel. [mol m 0.0E+00 0.E+00 G1 G2 G3 G4 G5 G6 G7 G8 G9 G1 G2 G3 G4 G5 G6 G7 G8 G9 G10 G10 2.0E-06 ] 9.0E-07 ] -1 c) -1 d) day day -2 -2
4.5E-07 1.0E-06 rate Zn rel. [mol m rate Mg rel. [mol m 0.0E+00 0.0E+00 G1 G2 G3 G4 G5 G6 G7 G8 G9 G1 G2 G3 G4 G5 G6 G7 G8 G9 G10 G10 Figure 12. Initial and final element release rates normalised to tailings surface area -2 -1 2- 2+ 2+ 2+ [mol m day ] for a) SO4 , b) Cu , c) Mg , and d) Zn .
exceeded final rates by up to a factor of 35. experiments (> 30 days) are to be Higher initial release rates may be due to, for recommended for establishment of example, dissolution and depletion of fine weathering rates in untreated and sterilised particles with high specific surface area or of mill tailings samples. secondary minerals present in the tailings. The results also indicated similar release Final release rates of all elements were rates of elements between sterilised and similar in most of the reactors (Fig. 12), untreated samples at the end of the irrespective of the presence/absence of experiments. While other explanations were filaments and which/whether a sterilisation possible (see Paper IV), these results indicate technique had been applied (cf. Table 6), that biological mediation of sulphide with the exception of reactors where large oxidation is not favoured at the experimental amounts of filaments were observed, that is, conditions, possibly due to, e.g., nutrient G6, G9 and final Cu2+ release rate in G10 limitation or inhibition due to high nitrate (see Paper IV). The similarity of final release concentration at low pH (e.g., Alexander et rates from the majority of reactors suggested al., 1987), and that abiotic processes that the sterilisation methods, other than dominate the release of elements. treatment with antibiotics, do not mechanically or chemically affect the long- term weathering of the tailings, given the experimental conditions. Normalisation to sample weight instead of the BET-surface area of each sample does not affect this conclusion (compare Fig.12 with Fig. 2 in Paper IV). An implication of the decline in release rates with time is that long-term
32 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
given in Table 4 in Paper I, with the
7 MODELLING WEATHERING exception of chlorite, where Ea=60 kJ/mol RATES ON BATCH, COLUMN, was used, in keeping with recent chlorite AND FIELD SCALE weathering studies (Brandt et al., 2003, and references therein).
7.1 Model interpretation of element release rates 7.2 Mineral weathering rates in tailings Conceptual understanding of element release compared to monomineralic rates, as is important for interpretation of samples impoundment groundwater composition and prediction of its evolution with time, 7.2.1 Pyrite relies on identification of source processes. In Papers III and V, element release from Figure 13 shows the obtained pyrite tailings samples were thus attributed to, and weathering rates vs the average pH for the interpreted in terms of, specific mineral time interval over which the rate was weathering reactions. Element release rates determined. Despite a decrease in rate with time, both initial and final rates of pyrite (Rj´) from batch experiments were used as tracers for mineral weathering reactions weathering in the pure pyrite reactor (crosses (Reactions 1, 3, 5, and 7-9 in Table 1). in Fig. 13b) were within the range of Copper and Zn2+ were used as tracers for reported abiotic pyrite weathering rates in the chalcopyrite and sphalerite weathering, the literature. This indicates that the applied respectively, and Mg2+, K+ and Ca2+ (or Na+) experimental method can achieve as tracers for chlorite, muscovite and comparative rates to those reported in the plagioclase weathering, respectively. The literature. pyrite weathering rates was obtained from Determination of the pyrite weathering rate 2- release of SO4 after correction for using Equation 7 lead to rates from the contribution of ZnS and CuFeS2 weathering. coarse tailings, the non-accelerated release (see Papers III and V). From the obtained period of the fine tailings, and the sulfide rates [mol g-1 s-1] and the mineral surface concentrate that agreed internally within a area as estimated by Equation 6, the surface- factor ~4 (Fig. 13a). In addition, the rate in area normalised weathering rate of the ith the fine tailings reactor at low temperature -2 -1 (squares in Fig. 13a), when scaled up to mineral, ri [mol m s ] was obtained from: room temperature, was in the same range. ′ R j However, these “non-accelerated” rates (Fig. r = [mol m-2 s-1] (7). i Aη 13a) were up to an order of magnitude lower i i, j than the lowest abiotic rates reported in the where ηi,j is the stoichiometric coefficient of literature. Possible explanations may include tracer j in the weathering reaction of mineral that the pyrite weathering rates in the i. This approach, combined with check of literature are generally performed on freshly solubility equilibrium control of tracer exposed mineral surfaces, whereas the concentrations, correction for multiple reactivity of the pyrite in the tailings had source minerals where possible, and decreased after 40 years of exposure in the estimation of surface area through Equation tailings impoundment, resulting in slower 6, is also applied in consideration of rates dissolution rates in the laboratory from column experiments (see Chapter 7.3) experiments. Lower weathering rates in aged and the field study (Chapter 3 and 7.3). samples has been reported in a number of For comparison of results from experiments weathering studies (e.g., Anbeek, 1993, for at different temperatures, rates were feldspar), and is further supported by the converted to 23 °C using the Arrhenius decrease in the pure pyrite weathering rate with time in this study. equation and the activation energies, Ea,
33 S. Ursula Salmon TRITA-LWR PHD 1009
1.E-07 1.E-07 a) b) 1.E-08 1.E-08 ] ] -1 -1 s s -2 -2 1.E-09 1.E-09
[mol m 1.E-10 [mol m 1.E-10
1.E-11 1.E-11 1.5 2.5 3.5 4.5 1.52.53.54.5 pH pH
Fig 13: Pyrite oxidation rate [mol m-2s-1] vs pH, 23 °C, Po2=0.2 atm. a) Non-accelerated pyrite oxidation rates; and b) Accelerated pyrite oxidation rate. Data points from this study: pyrite sample
(°), sulfide concentrate (■), coarse tailings (HNO3-•; H2SO4-▲), fine tailings (HNO3, pH ~3-o;
H2SO4, at pH~3 -◊, pH~2-∆, and pH ~3, scaled up from 5 ºC-□). Small/large symbols indicated initial/final rates. Lines and bars indicate literature values. Abiotic rates in a: horizontal line, Nicholson (1994: 25 °C); diagonal line, Williamson and Rimstidt (1994; 25 °C); (_), Nicholson and Scharer, (1994; 22°C). Whole and dotted bars in b: Scharer et al. (1994; 30°C) and Olson (1991; 28 °C), respectively, indicating range of abiotic (lower) and biotic (upper) rates.
The fact that pyrite weathering rates in the normalisation reported in Nordstrom and sulfide concentrate, which was a relatively Alpers, 1999); it is possible that the rates are well-defined size fraction, also were low limited by the same factor as limiting abiotic compared to the reported abiotic literature rates, by low concentrations of nutrients, or rates, suggests that the relatively lower rates by another, unknown factor. The low rates in the batch tailings experiments are not due relative to reported biotic rates are to overestimation of the pyrite surface area consistent with the findings described in by Equation 6, which for example may be Chapter 6.3 and Paper IV, where similar the case if pyrite was predominantly present element release rates were observed in in coarse size fractions. A further possibility sterilised and untreated tailings samples, is that the surface area normalised release leading to the conclusion that microbial rate is not constant over the different mediation was not favoured under the particle size fractions, as is assumed in experimental conditions. application of literature rate laws, determined on one particle size, to a particle 7.2.2 Sphalerite and chalcopyrite size distribution such as in tailings. As for pyrite, utilisation of Equation 7 gave Figure 13 highlights that the pyrite weathering rates from all samples that were weathering in the tailings and its sulfide the same to within a factor of approximately concentrate exhibited low to negligible pH- 10 for sphalerite (Fig. 14a). Figure 14 also dependence for 2 < pH < 4.5, as has highlights the low pH-dependence of the previously been reported in the literature for weathering rates of these minerals and that monomineralic weathering studies of pyrite conversion of sphalerite release from 5 °C (e.g., Nicholson, 1994; Williamson and using reported activation energies gave Rimstidt, 1994). similar rates to those at 25 °C (squares in The accelerated rates observed in fine Fig. 14). For chalcopyrite, rates agree within tailings reactors (Fig. 13b) at 25 °C may be a factor 14, and rates for coarse tailings and due to microbial mediation. However, rates fine reactors in H2SO4 at 25 °C agree within are lower than reported microbially mediated a factor 8 (Fig. 14b). The lower rate for rates (e.g., Olson, 1991; surface area chalcopyrite at 5 °C may indicate that the
34 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
1.E-07 1.E-07 a) b) 1.E-08 1.E-08 ] ] -1
1.E-09 -1 1.E-09 s s -2 1.E-10 -2 1.E-10 1.E-11 1.E-11 [mol m [mol [mol m 1.E-12 1.E-12
1.E-13 1.E-13 1.5 2.5 3.5 4.5 1.5 2 2.5 3 3.5 4 pH pH
Fig 14: Weathering rates [mol m-2s-1] vs pH at 23 ºC. a) sphalerite; and b) chalcopyrite. Markers denote fine tailings (HNO3, pH ~3-o; H2SO4, at pH~3 -◊, pH~2-∆, and pH ~3, scaled up from
5 ºC-□), coarse tailings (HNO3-•; H2SO4-▲), and sulfide concentrate (■). Bars indicate the range of rates reported in the literature (Scharer et al., 1994; 30°C); dotted bar - sphalerite/chalcopyrite alone, dashed bar - sphalerite/chalcopyrite in the presence of pyrite, indicating range of abiotic (lower) and biotic (upper) rates.
employed literature value of Ea is not area by Equation 6 and therefore representative of chalcopyrite oxidation in overestimation of the rate in Equation 7. tailings, or that another process, with a 2+ higher Ea, controls Cu release. Release was 7.2.3 Aluminosilicate minerals also higher in the reactor with fine tailings in Normalisation of aluminosilicate weathering HNO3 (circle in Fig. 14b); of note may be rates to the BET surface area using Equation that this greater release rate occurred at the 2- 6 and 7 lead to generally higher initial rates same time as the accelerated SO4 release in for coarse tailings for chlorite (Fig. 15a) and this reactor. plagioclase (Fig. 15c), whereas rates of In contrast to the pyrite rates, the muscovite dissolution were similar in coarse chalcopyrite and sphalerite weathering rates and fine tailings (Fig. 15e,f). For the determined in the batch experiments were plagioclase rate, which is derived from the orders of magnitude higher than reported Ca2+ release rate, this would appear to be abiotic rates in the literature, although lower due to gypsum dissolution in the initial than reported biotic (Fig. 14) The observed stages of the coarse experiments, and is rates were close to reported abiotic rates for resolved only by considering later rates of mineral dissolution in a mixture with pyrite. Ca2+ release or plagioclase weathering based As mentioned in Section 2, it was observed on Na release (Figure 15c,d). Higher chlorite that opaque minerals in the tailings (i.e., weathering was also found in the fine tailings predominantly sulfides) are often reactor at pH ~2; this suggests that the intergrown; accelerated dissolution of higher release is a function of preferential sphalerite, and other minerals with lower leaching at low pH (e.g., Schnoor, 1990; rest potentials than pyrite in the Malmström and Banwart, 1997; Gustafsson electrochemical series, has been reported in and Puigdomenech, 2003), and that Mg2+ the presence of pyrite (e.g., Kwong, 1993, release at low pH does not necessarily reflect 2001), and may be the cause of the relatively chlorite weathering. higher rates. Further possibilities may While final weathering rates were generally include oxidation of these minerals by 1-2 orders of magnitude lower than those aqueous Fe(III), or predominance of these reported in the literature, the pH- minerals in the fine tailings fraction, which dependence of mineral weathering in general would lead to underestimation of surface displays the same trends as reported in the
35 S. Ursula Salmon TRITA-LWR PHD 1009
1.E-10 1.E-10 a) initial chlorite b) final chlorite ]
-1 1.E-11 1.E-11 s -2
1.E-12 1.E-12
1.E-13 1.E-13 rate [mol m
1.E-14 1.E-14 1.5 2 2.5 3 3.5 4 1.5 2 2.5 3 3.5 4 pH
1.E-09 1.E-09 c) initial plagioclase d) final plagioclase ] ] -1 -1 1.E-10 s 1.E-10 s -2 -2 1.E-11 1.E-11
1.E-12 1.E-12 rate [mol m rate [mol m
1.E-13 1.E-13 1.522.533.54 1.5 2 2.5 3 3.5 4 pH pH
1.E-11 1.E-11 e) initial muscovite f) final muscovite ] ] -1 -1 s s
-2 1.E-12 -2 1.E-12
1.E-13 1.E-13 rate [mol m rate [mol m
1.E-14 1.E-14 1.522.533.54 1.522.533.54 pH pH
Fig 15: Initial (left panel) and final (right panel) weathering rates [mol m-2s-1] (23 ºC) vs pH. a, b) chlorite; c, d) plagioclase; and e, f) muscovite; Markers denote fine tailings (HNO3, pH ~3- o; H2SO4, at pH~3 -◊, pH~2-∆, and pH ~3, scaled up from 5 ºC-□) and coarse tailings (HNO3 -•; H2SO4 -▲). Additional symbols in panel c and d indicate weathering rates derived from Na release for coarse tailings (HNO3-; H2SO4 -■) and fine tailings (H2SO4, pH~2 -°). Dashed lines indicate rates from literature rate laws (chlorite – Malmström et al., 1995; plagioclase, Oxburgh et al., 1994; muscovite – Knauss and Wolery, 1989). Solid line in b: regression on fine tailings points. Error bars indicate uncertainty due to unknown tracer stoichiometry (see Table 1 in Paper V).
literature (e.g., Fig.15b).
36 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
source of solutes in the impoundment 7.3 Comparison: Batch, column and groundwater (see also Banwart and field rates Malmström, 2001). In order to further assess the 1-2 orders of Comparison of data in Figure 16, where magnitude discrepancy between field and laboratory rates for batch experiments on laboratory weathering rates observed in fine tailings at 25 and 5 °C and coarse Papers I and III, experimental mineral tailings, as well as rates in columns, are weathering rates obtained on different scales normalised to field rates, indicates that for were compared directly, before and after most components, rates determined in batch correction for field conditions (Papers III experiments overestimate field rates by 1-5 and V). Element release rates determined on orders of magnitude, which is consistent a mesoscale in water saturated columns with the commonly observed scale (Table 2 in Paper VI), using similar coarse dependence of mineral weathering rates. In and fine tailings as used in the batch contrast, rates from columns underestimated experiments, were preliminarily interpreted the majority of components. in terms of weathering rates and utilised in the comparison between different scales. As 7.4 Prediction of field weathering rates the tailings surface area-normalised element from laboratory experiments release rates obtained for the two columns In order to predict field element release agreed within a factor 3, results from the rates, kinetic data from the laboratory coarse tailings column are used here. The experiments were scaled with factors effluent pH from the column over 11 accounting for the effect of differences in month-experiment was similar to that in the temperature, mineral surface area (as field, ~5. estimated by Equation 6), and, where Element release rates from the field and appropriate, pH and Po2, between the scales column experiment were assessed as on the weathering rates, in an approach concentration x flowrate and were similar to that proposed by Malmström et al. normalised to the mass of material present (2000): (see Papers V and VI). For Impoundment 1, the mass of tailings in the unsaturated zone B−F F B−F F B T pH Po2 was used, assuming, as in the modelling R j = ηi, j Ai ri = ηi, j Ai ri β i β i β i study, that the unsaturated zone is the major (8)
6 5 4 3 ) F 2 1 log (R/R 0 -1 -2 -3 SFe-Cu Zn Mg Na K Ca Al- pyr mod
Figure 16. Comparison of element release rates [mol g-1 day-1] in batch reactors (fine tailings, 25 °C, HNO3, pH~3, ο; fine tailings, 5 °C, H2SO4, pH~3, ∆; coarse tailings, 25 °C, HNO3, pH~2 ◊) and coarse tailings column (□), normalised to field rates.
37 S. Ursula Salmon TRITA-LWR PHD 1009
6 6 5 a) fine tailings, 25 °C 5 b) fine tailings, 5 °C 4 4
) 3 ) 3 F F 2 2 1 1 log (R/R log (R/R 0 0 -1 -1 -2 -2 -3 -3 SFe-Cu Zn Mg Na K Ca Al SFe-Cu Zn Mg Na K Ca Al- pyr pyr mod
6 6 5 c) coarse tailings 5 d) column 4 4 ) F
) 3 3 F 2 2 1 1 log (R/R 0 log (R/R 0 -1 -1 -2 -2 -3 -3 SFe-Cu Zn Mg Na K Ca Al SFe-Cu Zn Mg Na K Ca Al- pyr pyr mod
Figure 17. Prediction of field-scale element release rates ([mol g-1 day-1], normalised to field rates), based on batch experiments in this study (Papers III-V), a) fine tailings at 25 °C in HNO3, b) fine tailings, at 5 °C, c) coarse tailings at 25 °C. d) Release rates from a column study on similar coarse tailings materials (Paper VI).
T rather than prevailing partly water- where βi is the correction factor for temperature based on the Arrhenius unsaturated conditions at the time of field sampling. Aluminium release was predicted β pH β Po2 equation, i and i are correction for from aluminosilicate mineral weathering pH and Po2 based on the dependencies rates and stoichiometry (i.e., the aluminium given in rate laws in Table 1 (see Paper V for model applied in Papers III and V, see, e.g., details), and superscripts B, F, and B-F Equation 8 and Figure 8 in Paper V); as the denote batch, field, and batch scaled to field. modelling study indicated a low degree of The quantification of s caling factors for iron oxidation or immobilisation in Po2, pH, and temperature are thus Impoundment 1, Fe release was predicted independent of the laboratory experiments 2- from SO4 release and pyrite stoichiometry. presented in this thesis. An analogous Adjustment for field conditions (Fig. 17a-c) approach was applied for adjustment of the brought batch experiment-derived rates for column rates, with the exception for the all elements, except Ca2+, closer to those correction for field Po2. Given the observed in the field, particularly for those sensitivity of the release rate of S, Fe, Cu2+ 2+ derived at low temperature (Fig. 17b) and and Zn to the partial pressure of oxygen from the coarse tailings (Fig. 17c), which (e.g., Table 1) and that this quantity was not agreed with field rates to within a factor 10, measured in the columns, release rates of except for Cu2+, K+ and Ca2+. Sulfate release these elements were not corrected for field (and thus, Fe release) is overestimated by the Po Po (i.e. β 2 =1 for all minerals). The 2- 2 i accelerated SO4 release rate observed in the predicted field rates from the column studies 25 °C fine batch reactor in HNO3 medium thus applies to water saturated conditions in (Fig. 17a; cf. Fig. 11a). This result is the field, such as was aimed for with the consistent with conclusions of the earlier applied remediation measures at the site, modelling study (Chapters 4-5) that
38 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale microbial processes do not contribute Field rates were also predicted from mineral greatly to the geochemical processes weathering rate laws from the literature occurring in Impoundment 1. Consideration (Paper V, Figure 9d). However, in general, of microbial conditions when deriving prediction using kinetic data of the site laboratory sulfide oxidation rates to specific tailings resulted in better prediction extrapolate to other scales is thus of of element release rates in the field. This importance, particularly as model results indicates that factors that contribute to were sensitive to small changes in the pyrite discrepancies between laboratory weathering weathering rate (Chapter 5.5.1). rates of minerals in the tailings and in Calcium release is underestimated by the monomineralic museum specimens (see adjusted plagioclase rates in all reactors; this Chapter 7.2) also are important in the field. is consistent with control of calcium release in the field by dissolution of gypsum rather than plagioclase weathering, as was also concluded from the modelling study (cf. Figure 5). The prediction based on batch experiments for Cu2+, Zn2+, K+ and Al3+, and, from coarse tailings, Mg2+, overestimated the field release rates, indicating a remaining unresolved difference between the laboratory results and field rates. Possible reasons for remaining discrepancies include that additional processes affect the component discharge rate at the somewhat higher pH in the field. This suggests a need for experiments at pH values that are similar to those that will occur in the field, long term experiments so that mineral weathering reaches steady state, and further investigation and quantification of potential immobilising processes in the field. 2- 2+ 2+ Lower rates of SO4 , Fe, Cu and Zn as predicted from columns-derived rates are due to the low oxygen availability in these experiments; low aluminosilicate dissolution may be due to the decreased proton release (e.g., Banwart et al., 2002), or greater preferential flow in the column than in the field. This implies that rates of contaminant release will be lower at water saturated conditions, which is consistent with modelling studies and reported preliminary post-remediation studies of Impoundment 1 (Chapter 5.2.2). However, it also indicates that prediction of unremediated conditions (ie with the presence of an unsaturated zone) from saturated columns will be difficult without consideration of the dependence of aluminosilicate weathering rates on acid production rather than just pH.
39 S. Ursula Salmon TRITA-LWR PHD 1009
40 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
impoundment. Uncalibrated, absolute 8 CONCLUSIONS AND weathering rates and thereby also resulting IMPLICATIONS model concentrations were however 1-2 orders of magnitude higher than observed in A methodology is presented for studying the field. For all major components, a single and quantifying geochemical processes that -2 calibration factor, Xr=10 , applied similarly contribute to AMD generation in base metal to all weathering rate expressions, sufficed to sulfide ore mill tailings. The methodology account for this (commonly observed) builds on first principles regarding apparent scale dependence of mineral geochemical processes, using site-specific weathering rates, from laboratory values quantification of factors that are commonly reported in the literature (uncalibrated observed to exert a large influence on model) to field relevant rate values mineral weathering rates, such as (calibrated model). temperature, pH, and mineral surface area. Similar modelling approaches have been Subsequent laboratory studies of mineral applied to waste rock and other geochemical weathering rates in site-specific tailings systems in the past, however the bulk of material taken from the field indicated that literature on prediction of AMD from mill weathering rates for pyrite and tailings currently focuses on static and aluminosilicates, normalised to mineral surface area, were in fact 1-2 orders of kinetic laboratory testing. Such testing is reported to be associated with a high degree magnitude lower in the ~50-year-old tailings of uncertainty, and furthermore largely than the rates reported in the literature, neglects coupling between geochemical which are usually determined on freshly processes and oxygen availability. The exposed mineral surfaces. Observations of transparent modelling methodology and similarly low, surface area-normalised results presented in this thesis will hopefully weathering rates in a sulfide concentrate contribute to increased general (where extraction of this fraction from the understanding of dominant processes tailings removed a number of assumptions affecting AMD generation, increased otherwise required for determination of capacity to quantify these processes, and mineral surface area in tailings) indicated wider application of such quantification, that the applied method of sulfide surface with improved mitigation of the area estimation in the experimental tailings environmental impact of mining as the material was a reasonable approximation, ultimate result. and that the observed difference in weathering rates was likely due to Application of the modelling methodology differences in mineral properties between to the case study site, Impoundment 1 of the freshly exposed minerals and old tailings Kristineberg mine site in northern Sweden, material. Exact reasons for the observed involved consideration of site-specific decrease with time of weathering rates tailings mineralogy, quantification of slow normalised to mineral surface area is a topic weathering processes using rate laws of ongoing scientific discussion (e.g., White obtained from the literature, and coupling to and Brantley, 1995; Hodson and Langan, fast equilibrium processes, such as aqueous 1999), and remains yet to be explicitly speciation and precipitation/dissolution of quantified in predictive geochemical models. secondary phases. This approach The results of the present laboratory study, successfully reproduced the pH at the site; however, indicated that model results of the major processes contributing to the dominant processes in tailings balance of proton production and impoundments were valid after the scaling consumption were found to be pyrite down of all weathering rates in the model by oxidation and aluminosilicate and Fe(OH)3(s) the single factor ~102. Furthermore, the dissolution. Relative release rates of most laboratory study indicted that pyrite and major ions were also the same as in the aluminosilicate minerals exhibited the same
41 S. Ursula Salmon TRITA-LWR PHD 1009
dependence on pH and temperature, and for that further investigation is also needed of some minerals, also on surface area, as other factors, such as nutrient availability, reported in the literature. This observation which may limit microbial mediation in the further supported the hypothesis that field. To this end, the present laboratory literature weathering rate laws, obtained experiments indicate that commonly applied from monomineralic samples, are applicable sterilisation methods do not affect long-term for tailings, once allowance is made for the (>30 days) abiotic element release rates in scale dependence in absolute rate values. tailings. The modelling study demonstrated the In contrast to findings for pyrite and importance of kinetic representation of aluminosilicate weathering rates, heavy metal aluminosilicate dissolution and iron redox (Zn2+, Cu2+) release rates derived here from reactions for prediction of pH, which results batch laboratory experiments exceeded from the balance of proton production and reported literature rates for oxidation of attenuation, and major ion composition of sphalerite and chalcopyrite by 1-3 orders of the impoundment groundwater. At the case magnitude. For Zn2+, uncalibrated study site, kinetic aluminosilicate dissolution extrapolation of this batch rate from was found to provide the bulk of proton laboratory to field conditions lead to attenuation and, as a result, considerably reproduction of the field rate within a factor affect the rate of depletion of fast-reacting 10; for Cu2+ release, however, the pH-buffering minerals, such as calcite, which corresponding field rate was overestimated are able to keep pH at a circum neutral level. by 2-4 orders of magnitude. Possible Inclusion of the kinetics of aluminosilicate differences in the mechanisms of release and dissolution and of the feedbacks between immobilisation of Zn2+ and Cu2+ (and also slow and fast processes is thus potentially Al3+ and K+) between the laboratory and the crucial for prediction of the proton balance field require further investigation. and the long-term evolution of pH in This study suggests that within certain limits, general. which appear narrower than recognised in Simulations further indicated that the iron current prediction practices in industry, it is redox cycling, which may be accelerated by possible to predict the weathering behaviour acidophilic bacteria, was low at the case of major minerals, and hence proton release study site, where conditions of low and attenuation, in base metal tailings under temperature and near-neutral pH prevail. relevant field environmental conditions. However, the sensitivity of the modelled Remaining issues to resolve for further groundwater composition and pH to sulfide improved confidence in predictive modelling and iron redox reactions indicated that of leachate composition from tailings quantification of microbial mediation of impoundment include changes in mineral these reactions at other sites may be weathering rates with time and the necessary for predicting the overall oxidation mechanisms controlling heavy metal release rate. Limited oxygen availability has in and retention. The herein proposed previous modelling studies been assumed to methodology of integrating model be the overall limiting factor on AMD quantification, geochemical characterisation quality; however, even for a given, low and observations on different scales may oxygen flux, the modelled pH was sensitive considerably aid in such issue resolution. to the balance between pyrite oxidation Furthermore, more sophisticated pathways. Consideration of the possibility of mathematical-numerical models than used in microbial mediation and other factors that this thesis provide means to couple more lead to uncertainty in the iron redox cycling detailed representations of transport and therefore require closer investigation. mass transfer processes with the Comparison of rates in the field and in the geochemical processes and principles laboratory studies, however, also indicated focussed upon here. Important
42 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale complementary studies to those presented in this thesis include prediction of AMD potential already during the mining exploration stage (e.g., Kwong, 1993, 2003), and assessment of the effect of subsurface heterogeneity on soil cover remediation of mine wastes (e.g., Werner, 2000) and on downstream contaminant transport (e.g., Berglund et al., 2003), which all indicate that the improved source term quantification attempted here will contribute to improved prediction reliability of the environmental impacts of AMD. Such improved reliability will in turn assist in optimisation of remediation expenditure on both site (O’Kane and Wels, 2003) and water catchment (Baresel et al., 2003) scales, and thereby facilitate efficient attainment of environmental water quality goals.
43 S. Ursula Salmon TRITA-LWR PHD 1009
44 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
9 ACKNOWLEDGEMENTS
Financial support from MISTRA, the Swedish Foundation for Strategic Environmental Research, through the MiMi programme ("Mitigation of the environmental impact from mining waste"), and from the EU project ERMITE ("Environmental regulation of mine waters in the EU"), funded within the European Commission Fifth Framework Programme under Contract No EVK1-CT-2000-0078, is gratefully acknowledged. Participation in the MiMi and ERMITE programmes has furthermore allowed interdisciplinary problem formulation and collaboration, and formation of friendships, with other PhD students and various experts in their (diverse) fields, for which I am also very grateful.
I wish to sincerely thank Georgia Destouni, my official supervisor, for giving me the chance to work with the environmental impact of mining. As a student of Gia’s I’ve also had the good fortune to witness at first hand application of critical thought to, among other things, research, as well as masterful written communication of research, lessons that I value greatly. I will also forever be indebted to my assisting supervisor Maria Malmström, who has been a pillar of technical and moral support throughout this postgraduate study period. Maria’s scientific curiosity, rigour, and skill have lit up the path of my wanderings through geochemistry, and without Maria’s help, the work in this thesis probably never would have happened. Furthermore, I probably still wouldn’t be able to hold a conversation in Swedish. TACK, Maria!
I’ve been very lucky to be surrounded by a great bunch of “usual suspects” at the Department of Land and Water Resources Engineering. Despite my poor attendance at “fika” over the last couple of years, your smiling faces in the corridors have always inspired. Special thanks must also go to Monica Löwén and Ann Fylkner for help in the LWR laboratory, and to Aira Saarelainen for help with all administrative matters, no matter how short the notice.
To my friends and family far away – thanks for being here in spirit (if that’s what SMSs and emails are called) and quite often in person too. To Patrik’s family in Karlstad and our friends in Stockholm, and especially Cate and fam, Jad, and Mike: thanks for considering me as one of the family, keeping me sane, and making Stockholm so much more than just a pretty place to do a PhD.
To Mum, Dad, and Rob: thanks for not mentioning too often how far away I am, and for always being there.
To Patrik: Thankyou for all your patience and support, all these years. I don’t know how other people get through PhDs without you.
45 S. Ursula Salmon TRITA-LWR PHD 1009
46 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
Development, Contract No EVK1-CT-2000- 0078, University of Oviedo. 10 REFERENCES Berglund, S. Malmström M., Järsjö, J., Destouni, G., 2003. Effects of spatially variable flow on Alexander, B., Leach, S., Ingledew, W. J., 1987. the attenuation of acid mine drainage in The relationship between chemiosmotic groundwater. In: Proc. Sudbury 2003, Mining parameters and sensitivity to anions and and the Environment, Sudbury, Canada, May organic acids in the acidophile, Thiobacillus 26-28, 2003 (CD-rom). ferrooxidans. J Gen Microbiol, 133, 1171- Blowes, D.W., Ptacek, C.J., 1994. Acid- 1179. neutralization mechanisms in inactive mine Alpers, C.N., Nordstrom, D.K., 1999. tailings. In: Jambor, J.L., Blowes, D.W. (Eds), Geochemical modelling of water-rock Env. Geochem of Sulphide Mine-Wastes, interactions in mining environments. In: G.S. Min. Assoc. Can., Nepean, Canada, pp. 271- Plumlee and M.J. Logsdon (Editors), The 292. environmental geochemistry of mineral Brandt, F., Bosbach, D., Krawczyk-Barsch, E., deposits. Reviews in Economic Geology, vol. Arnold, T., Bernhard, G., 2003. Chlorite 6A., Soc. Ec. Geologists, Chelsea, USA, dissolution in the acid pH-range: A combined pp.289-323. microscopic and macroscopic approach. Anbeek, C., 1993. The effect of natural Geochimica et Cosmochimica Acta, 67 (8), weathering on dissolution rates. Geochimica 1451-1461. Cosmochimica Acta, 57, 4963-4975. Brown, P.L., Lowson, R.T., 1997. The use of Axelsson, C.-L., Karlqvist, L., Lintu, Y., Olsson, kinetic modelling as a tool in the assessment T., 1986. Gruvindustrins restproduktupplag - of contaminant release during rehabilitation fältundersökningar med vattenbalansstudie i of a uranium mine. J. Cont. Hydrol. 26, 27- Kristineberg. Uppsala Geosystem AB, 34. Uppsala, Sweden (in Swedish). Brown, P.L., Ritchie, I.M., Bennett, J.W., 2000. Bain, J.G., Blowes D.W., Robertson W.D., Frind Geochemical kinetic modelling of acidic rock E.O., 2000. Modelling of sulfide oxidation drainage. In: Proc. ICARD 2000, 5th Int. with reactive transport at a mine drainage Conf. on Acid Rock Drainage, Denver, site. J. Cont. Hydrol. 41, 73-94. Colorado, U.S.A., May 21-24, 2000, vol. 1. Ball, J.W., Nordstrom, D.K., 1991. User’s pp. 289-296. manual for WATEQ4F, with revised Carlsson, E., Thunberg, J., Öhlander, B., thermodynamic database and test cases for Holmström, H., 2002. Sequential extraction calculating speciation of major, trace and of sulfide-rich tailings remediated by the redox elements in natural waters. U.S. application of till cover, Kristineberg mine, Geological Survey Open-File Report 91-183, northern Sweden. Sci. Total Environ. 299, 189p. (Revised and reprinted August 1992.) 207-226. Banwart, S., Malmström, M.E., 2001. Carlsson, E., Öhlander, B., Holmström, H. 2003. Hydrochemical modelling for preliminary Geochemistry of the infiltrating water in the assessment of minewater contamination. J. vadose zone of a remediated tailings Geochem. Expl. 74, 73-97. impoundment, Kristineberg mine, northern Banwart, S.A., Evans K.A., Croxford, S., 2002. Sweden. Appl. Geochem. 18(5), 659-674. Predicting mineral weathering rates at field Corrége, O., Carlsson, E., Öhlander, B., 2001. scale for mine water risk assessment. In: P.L. Geochemical investigation of the Younger and N.S. Robins (Editors), Mine groundwater in sulphide-bearing tailings water hydrogeology and geochemistry. The remediated by applying till cover, Geological Society of London, Bath, U.K. Kristineberg, northern Sweden. In: Proc. Baresel, C., Larsén, K., Destouni, G., Gren, I.- Securing the Future, Int. Conf. on Mining M., 2003. Economic Analysis of Mine Water and the Environment, Skellefteå, Sweden, Pollution Abatement on a Catchment Scale. June 25-July 1, 2001, pp. 97-114. ERMITE Report: D5; the European Edwards, K.J., Bond, P.L., Druschel, G.K., Commission Fifth Framework Programme, McGuire, M.M., Hamers, R.J., Banfield, J.F., Energy, Environment and Sustainable 2000. Geochemical and biological aspects of sulfide mineral dissolution: Lessons from
47 S. Ursula Salmon TRITA-LWR PHD 1009
Iron Mountain, California. Chem. Geol. 169, sulfide ores. Biotechnology and 383-397. Bioengineering, 69 (2), 196-203. Ehrlich, H.L., 1996. Geomicrobiology, 3rd ed. Herbert, R.B.Jr., Malmström, M., Ebenå, G., Marcel Dekker, New York, N.Y. Salmo,n S.U., Ferrow, E., Fuchs, M., 2003. Ekstav, A., Qvarfort, U., 1989. Metallbalans Sterilisation of mine tailings for the Kristineberg. Quaternary Geology, Dept. quantification of abiotic oxidation rates (in Earth Sciences, Uppsala universitet, Uppsala, preparation). Sweden (in Swedish). Hodson, M.E., Langan, S.J., 1999. The influence Elberling, B., Nicholson, R.V., 1996. Field of soil age on calculated mineral weathering determination of sulphide oxidation rates in rates. Applied Geochemistry, 14, 387-394. mine tailings. Water Resour. Res. 32, 1773- Holmström, H., Salmon, U.J., Carlsson, E., 1784. Paraskev, P., Öhlander, B., 2001. Elberling, B., Nicholson, R.V., Scharer, J.M., Geochemical investigations of sulfide-bearing 1994. A combined kinetic and diffusion tailings at Kristineberg, northern Sweden, a model for pyrite oxidation in tailings: a few years after remediation. Sci. Tot. change in controls with time. J. Hydrol. 157, Environ. 273, 111-133. 47-60. Jambor, J.L., 1994. Mineralogy of sulfide-rich Elberling, B., Schippers, A., Sand, W., 2000. tailings and their oxidation products. In: Bacterial and chemical oxidation of pyritic Environmental Geochemistry of Sulphide mine tailings at low temperatures. J. Contam. Mine-Wastes Jambor J.L. and Blowes D.W., Hydrol. 41, 225-238. (Eds) Short course handbook, vol.22, Mineral Association of Canada, pp.59-102. Evangelou, V.P., Zhang, Y.L. 1995. A review: Pyrite oxidation mechanisms and acid mine Jaynes, D.B., Rodowski, A.S., Pionke, H.B., drainage prevention. Crit. Rev. Env. Sci. 1984. Acid mine drainage from reclaimed Tech., 25, 141-199. coal strip mines 1. Model description. Water Resour. Res. 20, 233-242. Fowler, T.A., Holmes, P.R., Crundwell, F.K., 2001. On the kinetics and mechanism of the Johnson, R.H., Blowes, D.W., Robertson, W.D., dissolution of pyrite in the presence of Jambor, J.L., 2000. The hydrogeochemistry Thiobacillus ferrooxidans. Hydrometallurgy of the Nickel Rim mine tailings 59, 257-270. impoundment, Sudbury, Ontario. Journal of Contaminant Hydrology, 41 (1-2), 49-80. Furrer, G., Westall, J., Sollins, P., 1989. The study of soil chemistry through quasi-steady- Jurjovec, J., Ptacek, C. J., Blowes, D. W., 2002. state models: I. Mathematical definition of Acid neutralization mechanisms and metal model. Geochim. Cosmochim. Acta 53, 595- release in mine tailings: A laboratory column 601. experiment, Geochimica et Cosmochimica Acta, 66, 1511-1523. Gustafsson, Å.B., Puigdomenech, I., 2003. The effect of pH on chlorite dissolution rates at Kirby, C.S., Thomas, H.M., Southam, G., 25ºC Mat. Res. Soc. Symp. Proc. Vol. 757, Donald, R., 1999. Relative contributions of Materials Research Society II3.16.1-7. abiotic and biological factors in Fe(II) oxidation in mine drainage. Appl. Geochem. Gustafsson H.E., Lundgren T., Lindfred M., 14, 511-530. Lindahl L.-Å., Eriksson N., Jönsson H., Broman P.G., Göransson, T., 1999. The Kwong, Y.T.J., 1993. Prediction and Prevention Swedish acid min drainage experience: of Acid Rock Drainage from a Geological Research, development and practice. In: J.M. and Mineralogical Perspective. MEND Azcue (Editor), Environmental Impacts of Report 1.32.1, MEND (Mine neutral Mining Activities: Emphasis on Mitigation drainage) programme, Ottawa, ON, Canada. and Remedial Measures. Springer-Verlag Kwong, Y.T.J., 2001. Metal leaching through Environmental Science Series, Berlin, galvanic interaction -Implications for mine Germany, pp.203-228. waste management. In: Proc. Securing the Harahuc, L., Lizama, H.M., Suzuki, I., 2000. Future, Int. Conf. on Mining and the Effect of anions on selective solubilization of Environment, Skellefteå, Sweden, June 25- zinc and copper in bacterial leaching of July 1, 2001, pp.374-383.
48 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
Kwong, Y.T.J., 2003. Thoughts on practical Mayer, K.U., Frind, E.O., Blowes D.W., 2003. tools for early identification of environmental Multicomponent reactive transport modelling challenges associated with a mining project. in variably saturated porous media using a In: Proc. ICARD 2003, 6th Int. Conf. on Acid generalized formulation for kinetically Rock Drainage, Cairns, Australia, July 14-17, controlled reactions. Water. Resour. Res. 2003, p.601-604. 38(9), 1174. Lacey, D.T., Lawson F., 1970. Kinetics of liquid- MEND, 2000. MEND Manual, Volume 3 - phase oxidation of acid ferrous sulfate by the Prediction. MEND 5.4.2c, Mine Neutral bacterium Thiobacillus ferrooxidans. Drainage Programme, Ottowa, Canada. Biotechnol. Bioeng. 12, 29-50. MiMi, 2002. Progamme plan for the year 2003. Lichtner, P.C., 1996. Continuum formulation of MiMi Report, Mitigation of the multicomponent-multiphase reactive environmental impact from mining waste transport. In: P.C. Lichtner, C.I. Steefel, and programme (MiMi), Stockholm, Sweden. E.H. Oelkers (Editors.), Reactive transport in Nemati, M., Webb, C., 1997. A kinetic model for porous media. Rev. Mineral., vol. 34., biological oxidation of ferrous iron by Mineralogical Society of Amererica, pp. 1-81. Thiobacillus ferrooxidans. Biotechnol. Lindvall, M., Eriksson N., Ljungberg J., 1999. Bioeng. 53, 477-486. Decommissioning at Kristineberg Mine, Nemati, M., Harrison, S.T.L., Hansford, G.S., Sweden. In: Proc. Sudbury '99, Conf. on Webb, C., 1998. Biological oxidation of Mining and the Environment II, Sudbury, ferrous sulphate by Thiobacillus Canada, Sept. 13 - 15, 1999, pp. 855-862. ferrooxidans: a review on the kinetic aspects. Lindvall, M., Eriksson, N., 2003. Investigation of Biochem. Eng. J. 1, 171-190. weathering properties of tailings sand from Nesbitt, H.W., Jambor, J.L., 1998. Role of mafic Boliden’s Aitik copper mine, Sweden – A minerals in neutralizing ARD, demonstrated summar of twelve years of investigations. In: using a chemical weathering methodology. Proc. ICARD 2003, 6th Int. Conf. on Acid In: L.J. Cabri and D.J. Vaughan (Editors), Rock Drainage, Cairns, Australia, July 14-17, Modern approaches to ore and 2003, p.725-731. environmental mineralogy. Short course Malmstrom M, Banwart S., 1997. Biotite handbook, vol.27, Mineralogical Association dissolution at 25 degrees C: The pH of Canada, pp.403-421. dependence of dissolution rate and Nicholson R.V., 1994. Iron-sulphide oxidation stoichiometry. Geochimica Et Cosmochimica mechanisms: Laboratory studies. In: Acta 61, (14), 2779-2799. Environmental Geochemistry of Sulphide Malmström, M., Banwart, S., Duro, L., Wersin, Mine-Wastes. J.L. Jambor and D.W. Blowes P., Bruno, J., 1995. Biotite and chlorite (Editors), Short course handbook, vol.22, weathering at 25°C: The dependence of pH Mineralogical Association of Canada, pp.163- and (bi)carbonate on weathering kinetics, 183. dissolution stoichiometry and solubility; and Nicholson, R.V., Gillham, R.W., Cherry, J.A., the relation to redox conditions in granitic Reardon, E.J., 1989. Reduction of acid aquifers. Technical Report 95-01, The generation in mine tailings through the use of Swedish Nuclear Fuel and Waste Co. (SKB), moisture-retaining cover layers as oxygen Stockholm, Sweden. barriers. Can. Geotech. J. 26, 1-8. Malmström, M.E., Destouni, G., Banwart, S.A., Nordstrom, D.K., Alpers, C.N., 1999. Strömberg, B.H.E., 2000. Resolving the Geochemistry of acid mine waters. In: G.S. scale-dependence of mineral weathering Plumlee and M.J. Logsdon (Editors), The rates. Env. Sci. Technol. 34, 1375-1378. environmental geochemistry of mineral Malmström, M., Werner, K., Salmon, U., deposits. Reviews in Economic Geology, Berglund, S., 2001. Hydrogeology and vol. 6A, Soc. Ec. Geologists, USA, pp.133- geochemistry of mill tailings impoundment 1, 160. Kristineberg, Sweden: Compilation and Nordstrom, D.K., Southam, G., 1997. interpretation of pre-remediation data. MiMi Geomicrobiology of sulfide mineral 2001:4, Mitigation of the environmental oxidation. In: J.F. Banfield and K.H. Nealson impact from mining waste programme (Editors), Geomicrobiology: interactions (MiMi), Stockholm, Sweden. between microbes and minerals. Rev.
49 S. Ursula Salmon TRITA-LWR PHD 1009
Mineral., vol.35, Mineralogical Soc. Amer., Romano, C.G., Mayer, K.U., Jones, D.R., USA, pp. 361-390. Ellerbroek, D.A., Blowes, D.W., 2003. Olson, G.J., 1991. Rate of pyrite bioleaching by Effectiveness of various cover scenarios on Thiobacillus ferrooxidans: Results of an the rate of sulfide oxidation of mine tailings. interlaboratory comparison. Appl. Environ. Journal of Hydrology, 271 (1-4), 171-187. Microb. 57, 642-644. Rossi, G., 1990. Biohydrometallurgy. McGraw O´Kane, M., Wels, C., 2003. Mine Waste Cover Hill, Hamburg, Germany. System Design — Linking Predicted Sand, W., Gehrke, T., Jozsa P.-G., Schippers, A., Performance to Groundwater and Surface 2001. (Bio)chemistry of bacterial leaching – Water Impacts. In: Proc. ICARD 2003, 6th direct vs. indirect bioleaching. Int. Conf. on Acid Rock Drainage, Cairns, Hydrometallurgy 59, 159-175. Australia, July 14-17, 2003, p.341-349. Salmon, S.U., 2000. Biogeochemical processes in Parker G., Robertson A., 1999. Acid Drainage. mill tailings: Modelling and assessment of Australian Minerals & Energy Environment remediation effects. Licentiate thesis TRITA- Foundation, Melbourne, Australia. AMI LIC 2053, Water Resources Parkhurst, D.L., 1995. User’s guide to Engineering, Dept. Civil and Environmental PHREEQC – a computer program for Engineering, Royal Institute of Technology, speciation, reaction-path, advective transport Stockholm, Sweden. and inverse geochemical calculations. U.S. Salmon, S.U., Malmström, M., 2002. Steady Geological Survey Water-Resources state, geochemical box model of a tailings Investigations Report 95-4227, 143p. impoundment: Application to Impoundment Pentreath, R.J., 1994. The Discharge of Waters 1, Kristineberg, and prediction of effect of from Active and Abandoned Mines. In: R.E. remediation. MiMi 2002:2, Mitigation of the Hester and R.M. Harrison (Editors), Mining environmental impact from mining waste and its environmental impact. Royal Society programme (MiMi), Stockholm, Sweden. of Chemistry, U.K, pp.121-131. Salomons, W., 1995. Environmental-impact of Plumlee, G.S., Logsdon, M.J., 1999. An earth- metals derived from mining activities - system science toolkit for environmentally processes, predictions, prevention. Journal of friendly mineral resource development. In: Geochemical Exploration, 52 (1-2), 5-23. G.S. Plumlee and M.J. Logsdon (Editors), SCB, 2000a. Naturmiljön i siffror 2000. Statistics The environmental geochemistry of mineral Sweden, Stockholm (in Swedish). deposits. Reviews in Economic Geology, SCB, 2000b. Vattenräkenskaper. Rapport 2000:6. vol. 6A, Soc. Ec. Geologists, USA, pp.1-27. Statistics Sweden, Stockholm (in Swedish). Pulles, W., 2003. The status of water pollution in Scharer, J.M., Nicholson, R.V., Halbert, B., South Africa. In: Proc. ICARD 2003, 6th Int. Snodgrass, W.J., 1994. A computer program Conf. on Acid Rock Drainage, Cairns, to assess acid generation in pyritic tailings. In: Australia, July 14-17, 2003, p.21-27. C.N. Alpers and D.W. Blowes (Editors), Qvarfort, U., 1983. En kartläggning av Environmental geochemistry of sulfide sandmagasin från sulfidmalmsbrytning. oxidation. ACS symposium series, vol. 550, Naturvårdsverkets Rapport SNV PM 1699, Washington D.C., USA, pp. 132-152. The Swedish national environment Schnoor, J., 1990. Kinetics of chemical protection board, Stockholm, Sweden (in weathering: A comparison of laboratory and Swedish). field weathering rates. In: W. Stumm W. du Rietz, T., 1953. Geology and ores of the (Editor), Aquatic Chemical Kinetics. John Kristineberg deposit, Vesterbotten, Sweden. Wiley and Sons, New York, U.S.A. Sveriges Geologiska Undersökning, Ser C. 524, SEPA, 1986. Gruvavfall. Naturvårdsverket Årsbok 45(5) (1951). informera. Broschyr. Swedish Environmental Robertson, W.D., 1994. The physical Protection Agency, Stockholm. hydrogeology of mill-tailings impoundments. SEPA, 1995. Branschkartläggningen: En In: Environmental Geochemistry of Sulphide översiktlig kartläggning av efterbehandlings- Mine-Wastes. J.L. Jambor and D.W. Blowes behovet i Sverige. Rapport 4393. Swedish (Editors), Short course handbook, vol.22, Environmental Protection Agency, Mineral Association of Canada, pp.1-17. Stockholm.
50 Geochemical modelling of acid mine drainage in mill tailings: Quantification of kinetic processes from laboratory to field scale
SEPA, 1998. Gruvavfall - Miljöeffekter och Summary of remediation programme and behov av åtgäder. Rapport 4948. Swedish identification of potentially critical Environmental Protection Agency, assumptions. MiMi 2001:3, Mitigation of the Stockholm. environmental impact from mining waste Singer, P.C., Stumm, W., 1970. Acid mine programme (MiMi), Stockholm, Sweden. drainage: The rate-determining step. Science Werner, K., Carlsson, E., Berglund, S., 2001. 167, 1121-1123. Oxygen and water fluxes into a soil-cover Smith, K.S., Huyck, H.L.O., 1999. An overview remediated mill tailings deposit: Evaluation of the abundance, relative mobility, of field data from the Kristineberg mine site, bioavailability, and human toxicity of metals. northern Sweden. In: Proc. Securing the In: G.S. Plumlee and M.J. Logsdon (Editors), Future, Int. Conf. on Mining and the The environmental geochemistry of mineral Environment, Skellefteå, Sweden, June 25- deposits. Reviews in Economic Geology, July 1, 2001, pp. 896-905. vol. 6A, Soc. Ec. Geologists, USA, pp.29-70. White, A.F., Brantley, S.L., 1995. Chemical Strömberg, B., Banwart, S., 1994. Kinetic weathering rates of silicate minerals. Rev. modelling of geochemical processes at the Mineral., vol.31., Mineralogical Soc. Amer., Aitik mining waste rock site in northern USA. Sweden. Appl. Geochem. 9, 583-595. White A.F., Peterson M.L., 1990. Role of Strömberg, B., Banwart, S., 1999. Weathering reactive-surface area characterization in kinetics of waste rock from the Aitik copper geochemical kinetic models. In: Melchoir mine, Sweden: Scale-dependent rate factors D.L., Bassett R.L. (Eds.), Chemical Modelling and pH controls in large column of Aqueous Systems II, vol. 416. ACS experiments. J. Contam. Hydrol. 39, 59-89. symposium series, Washington D.C., USA, pp. 461-475 (Chapter 35). Tamura, H., Goto, K., Nagayama, M., 1976. The effect of ferric hydroxide on the oxygenation White, W.W. III, Lappakko, K., Cox, R.L., 1999. of ferrous ions in neutral solutions. Static-test methods most commonly used to Corrosion Sci. 16, 197-207. predict acid-mine drainage: Practical guidelines for use and interpretation. . In: Wachtmeister, C.A., Sundstsröm, G., 1986. G.S. Plumlee and M.J. Logsdon (Editors), Miljögifter och miljöföroreningar – en The environmental geochemistry of mineral översikt med exemple från yttre miljö och deposits. Reviews in Economic Geology, arbetsmiljö. In: T.-H. Iversen (Editor), vol. 6A, Soc. Ec. Geologists, USA, pp.325- Kjemiske miljØgifter: Effekter på biologiske 338. systemer, 2nd edition. Universitetsforlaget AS, Olso, Norway (in Swedish). Williamson, M.A., Rimstidt, J.D., 1994. The kinetics and electrochemical rate-determining Wehrli, B., 1990. Redox reactions of metal ions step of aqueous pyrite oxidation. Geochim. at mineral surfaces. In: Stumm, W. (Ed.), Cosmochim. Acta 58, 5443-545. Aquatic chemical kinetics. Wiley, New York, USA, pp. 311-336. Wunderly, M.D., Blowes, D.W., Frind, E.O., Ptacek, C.J., 1996. Sulfide mineral oxidation Werner, K., 2000. Soil-cover remediation of mill and subsequent reactive transport of tailings deposits: Effects on oxygen transport oxidation products in mine tailing and hydrological conditions. Licentiate thesis, impoundments: A numerical model. Water Dept. of Civil and Environmental Resour. Res. 32, 3173-3187. Engineering, Div. of Water Resources Engineering, Royal Institute of Technology, Yu, J.-Y., McGenity, T.J., Coleman, M.L., 2001. Stockholm, Sweden. Solution chemistry during the lag phase and exponential phase of pyrite oxidation by Werner, K., Berglund S., 1999. Effects of spatial Thiobacillus ferrooxidans. Chem. Geol. 175, variability on oxygen transport in sulphidic 307-317. mining waste deposits. In: Proc. Sudbury '99, Conf. on Mining and the Environment II, Sudbury, Canada, Sept. 13-17, 1999, pp. 243- 252. Werner, K., Salmon, S., 2001. Mine waste deposits and mines in Kristineberg, Sweden:
51