Applied Geochemistry 24 (2009) 1–15
Contents lists available at ScienceDirect
Applied Geochemistry
journal homepage: www.elsevier.com/locate/apgeochem
Mineralogy and environmental stability of slags from the Tsumeb smelter, Namibia
Vojteˇch Ettler a,*, Zdenek Johan b, Bohdan Krˇíbek c, Ondrˇej Šebek d, Martin Mihaljevicˇ a a Institute of Geochemistry, Mineralogy and Mineral Resources, Charles University, Albertov 6, 128 43 Prague 2, Czech Republic b Bureau des Recherches Géologiques et Minières (BRGM), av. Claude Guillemin, 45060 Orléans, cedex 2, France c Czech Geological Survey, Geologická 6, 152 00 Prague 5, Czech Republic d Laboratories of the Geological Institutes, Charles University, Albertov 6, 128 43 Prague 2, Czech Republic article info abstract
Article history: Three types of smelting slags originating from historically different smelting technologies in the Tsumeb Received 27 June 2008 area (Namibia) were studied: (i) slags from processing of carbonate/oxide ore in a Cu–Pb smelter (1907– Accepted 22 October 2008 1948), (ii) slags from Cu and Pb smelting of sulphide ores (1963–1970) and (iii) granulated Cu smelting Available online 30 October 2008 slags (1980–2000). Bulk chemical analyses of slags were combined with detailed mineralogical investi- gation using X-ray diffraction analysis (XRD), scanning electron microscopy (SEM/EDS) and electron Editorial handling by R. Fuge microprobe (EPMA). The slags are significantly enriched in metals and metalloids: Pb (0.97–18.4 wt.%), Cu (0.49–12.2 wt.%), Zn (2.82–12.09 wt.%), Cd (12–6940 mg/kg), As (930–75,870 mg/kg) and Sb (67– 2175 mg/kg). Slags from the oldest technology are composed of primary Ca- and Pb-bearing feldspars, spinels, complex Cu–Fe and Cu–Cr oxides, delafossite–mcconnellite phases and Ca–Pb arsenates. The presence of arsenates indicates that these slags underwent long-term alteration. More recent slags are composed of high-temperature phases: Ca–Fe alumosilicates (olivine, melilite), Pb- and Zn-rich glass, spi- nel oxides and small sulphide/metallic inclusions embedded in glass. XRD and SEM/EDS were used to study secondary alteration products developed on the surface of slags exposed for decades to weathering on the dumps. Highly soluble complex Cu–Pb–(Ca) arsenates (bayldonite, lammerite, olivenite, lavendu- lan) associated with litharge and hydrocerussite were detected. To determine the mineralogical and geo- chemical parameters governing the release of inorganic contaminants from slags, two standardized short-term batch leaching tests (European norm EN 12457 and USEPA TCLP), coupled with speciation- solubility modelling using PHREEQC-2 were performed. Arsenic in the leachate exceeded the EU regula- tory limit for hazardous waste materials (2.5 mg/L). The toxicity limits defined by USEPA for the TCLP test were exceeded for Cd, Pb and As. The PHREEQC-2 calculation predicted that complex arsenates are the most important solubility controls for metals and metalloids. Furthermore, these phases can readily dis- solve during the rainy season (October to March) and flush significant amounts of As, Pb and Cu into the environment in the vicinity of slag dumps. Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction investigations of slags are essential for understanding the position of potentially toxic elements in primary solid phases and represent Slags are the most important mineral wastes resulting from the first step in assessing their environmental impact (Ettler et al., pyrometallurgy. They correspond to the silicate melt produced 2000, 2001; Parsons et al., 2001; Lottermoser, 2002; Puziewicz during the pyrometallurgical recovery of base metals by fusion in et al., 2007). Furthermore, the textures and the chemical composi- a blast furnace and are produced in large amounts. Slags are gener- tion of primary phases in slags may also be used to estimate the ally deposited on dumps or, if considered to be unreactive materi- conditions of their formation, especially in relation to advances als, used for civil-engineering purposes such as road construction in smelting technologies (Ettler et al., 2000, in press; Manasse (Ettler et al., 2002, 2003a). However, these waste materials are of- et al., 2001; Manasse and Mellini, 2002a; Sáez et al., 2003; Haupt- ten enriched in toxic elements, in particular metals (Cu,Pb,Zn) and mann, 2007). in metalloids (As,Sb) that can be released into the environment Some recent monitoring studies have shown that the extensive through alteration processes and leaching (Parsons et al., 2001; Et- mining and ore processing activities in the Tsumeb district (Nami- tler, 2002; Piatak et al., 2004; Ettler et al., 2004, 2005; Lottermoser, bia) left important amounts of various mining and smelting waste 2005; Navarro et al., 2008; Costagliola et al., 2008). Mineralogical materials that can be considered as a serious problem in relation to environmental contamination (Ongopolo Mining and Processing
* Corresponding author. Tel.: +420 221 951 493; fax: +420 221 951 496. Limited, 2001; Krˇíbek and Kamona, 2005). Approximately E-mail address: [email protected] (V. Ettler). 24,550,280 t of ore were mined out during the modern history of
0883-2927/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.apgeochem.2008.10.003 2 V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 the Tsumeb Mine and it has been estimated that millions of tons of to produce granulated slag (Tsumeb Corporation Ltd., 1987; War- mining/processing wastes are stored in this area ( 200,000 t of tha and Genis, 1992). slag on dumps and at least 10 million t of material in tailing ponds) (Krˇíbek and Kamona, 2005). Because this has never been studied, 2.2. Slag sampling and processing this paper is focused on the Tsumeb smelter slags resulting from historically different smelting operations. Bulk chemical analyses Numerous slag samples were collected during the mission and phase/mineralogical analyses of slags were coupled with guided by the Czech Geological Survey in 2004 (Project of the experimental leaching and thermodynamic modelling, in order to Development Cooperation of the Czech Republic No. RP/20/2004). provide information on the solid speciation of pollutants in slags, Eleven slag samples were studied in this detailed investigation. weathering products and possible environmental impacts. They represent the available materials corresponding to the histor- ical evolution of smelting technologies in Tsumeb. The following 2. Materials and methods three groups of slags were collected (the detailed locations are gi- ven in Fig. 1): 2.1. Smelting site history (i) Slag type-I corresponds to historical slags produced from The Tsumeb deposit belongs to the northern Namibia sulphidic 1907 to 1948 in Pb–Cu blast furnaces processing mainly car- metallogenic province. The deposit lies in the upper part of the bonate and oxide ores and fired by coke from Germany and Otavi group, which consists of limestones and dolomites of Neo- South Africa. They are found as 20-cm large fragments of proterozoic age (Miller, 1983). The mineralization is characterised massive and heavy material of black and grey colour with by large-scale alteration (calcification and silicification of host a green crust of secondary phases (number of samples, rocks) and common hydrothermal carbonate veins. The deposit n = 1). contains a great diversity of ore minerals of Pb, Cu, Zn, Ag, As, Sb, (ii) Slag type-II corresponds to historical slags and mattes pro- Cd, Co, Ge, Ga, Au, Fe, Hg, Mo, Ni, Sn and W, as well as V, containing duced between 1963 and 1970 resulting from reverbatory about 11% Pb, 5% Cu and 4.3% Zn, with economic concentrations of furnaces (Cu smelting) and shaft furnaces (Pb smelting) fired Ag, Cd, Ge and As. The deposit was mined by a large open pit and with aerated pulverized coal. During this period, mainly sul- by several shafts. It was once the foremost producer of Pb in Africa phide ores were processed. The slags are found as up to 10- and, over its lifetime, has produced over 2 million t of Pb, some cm large fragments of black, vitreous material, often covered 500,000 t of Zn and over 1 million t of Cu (Frimmel et al., 1996; by white and blue secondary phases (n = 8). Chetty and Frimmel, 2000). (iii) Slag type-III corresponds to granulated slags from reverba- In 1907, two Pb–Cu blast furnaces were built in the Tsumeb tory furnaces produced by recent Ausmelt technology in area by the Otavi Minen- und Eisenbahn-Gesellschaft (OMEG) the Cu smelter from 1980 to 2000. In this case, the furnaces Company to smelt local ores. These furnaces were supplemented were fired with black oil and the furnace charge was pellet- by a third furnace in 1923; the smelter was then operated until ized with pulverized coal and fluxes (lime and quartz chert) the end of World War II, first processing the carbonate ore from prior to melting. The slags are found as granulated black the upper part of the Tsumeb ore body. After interruption of the material with fragments up to 3 cm in size or as milled pow- smelting activities in the 1950s, new smelters were constructed der (n = 2). in 1963, consisting of a Cu smelter with a reverbatory furnace and a Pb smelter with a shaft furnace. In this period, deeper parts 2.3. Bulk chemical analyses of the Tsumeb ore body composed of sulphide ores were processed (main minerals: chalcocite, enargite, galena, sphalerite). During the An aliquot of each sample (approximately 20 g) was crushed early 1980s, the Slag Mill was built to re-process old Cu reverba- and pulverized in an agate mortar using the Fritsch Pulverisette tory slags, which are milled and treated by flotation. The resulting apparatus and was used for bulk chemical analyses and phase concentrate is then passed to the smelters to recover Pb and Cu and identification using X-ray diffraction analysis. The bulk chemical
Fig. 1. Schematic sketch of the Tsumeb smelter and location of smelting slags. V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 3 composition was determined on pulverized slag samples using in Ettler (2002). A total set of 240 spot analyses was performed digestion in mineral acids and/or sintering and subsequent chem- by EPMA. ical analysis. Two types of dissolution techniques were employed prior to the analytical procedures. Dissolution procedure I (modified 2.5. Mineralogical analysis of secondary weathering products from CˇSN 72 0100, 1984 and Šulcek and Povondra, 1989) was used for the analysis of Ag, As, Ba, Co, Cr, Cu, K, Mn, Na, Ni, Pb, Zn (Varian The secondary weathering products were sampled on the sur- SpectrAA 280FS flame atomic absorption spectroscopy, FAAS), P face of the slags using a preparation needle under a binocular (spectrophotometry) and Sb, Bi, Ga, Ge (VG Elemental PQ3 induc- microscope. They were analyzed by XRD (same analytical condi- tively coupled plasma mass spectrometer, ICP-MS). A mass of tions as for bulk slag samples) and were examined using a JEOL
0.5 g was dissolved in a mixture of 10 mL HF (38%) and 2 mL HClO4 JSM 6400 scanning electron microscope equipped with a Kevex (70%) and was evaporated to near dryness. This procedure was re- Delta energy-dispersion spectrometer (SEM/EDS). peated and the residue was dissolved in 2% HCl or HNO3 (v/v), di- luted to 100 mL and stored in polypropylene (PP) AzlonÒ bottles 2.6. Leaching experiments until analysis. Dissolution procedure II (again modified from CˇSN 72 0100, 1984 and Šulcek and Povondra, 1989) was based on the To assess the leaching characteristics of these materials, leach- sintering of 0.2 g of sample with 0.6 g Na2CO3 and 0.05 g NaNO3 ing tests were performed on two slag samples from historically dif- overlapped with a thin layer of supplementary Na2CO3 in a Pt cru- ferent technologies: sample N2 (slag III, recent slag) and sample N6 cible at 750–820 °C for 3 h. The sintered sample was dissolved in a (slag I, 100-a-old material). Two standardized leaching tests were mixture of 10 mL H2O and 8 mL HCl (37%) at 90 °C. This solution performed: European norm EN 12457 and USEPA Toxicity Charac- was used for the determination of SiO2 (gravimetric analysis) and teristic Leaching Procedure (TCLP). All the experiments were per- further determinations of Al2O3,Fe2O3, FeO, MgO, CaO (volumetric formed in duplicate and with a procedural blank. The EN 12457 analysis) and TiO2 (spectrophotometry). Detailed methodologies of and TCLP procedures necessitate a reduction of the grain size of these analytical procedures are given in Johnson and Maxwell the waste material to <4 mm and <9.5 mm, respectively. The effect (1981). The standard solutions used were prepared with an ade- of the particle size, in particular the dust fraction, was found to be quate amount of the sintering mixture in order to minimize the crucial when interpreting the results, even for standardized leach- matrix effect. All the laboratory glassware was acid-washed and ing tests (Zandi et al., 2007). However, granulated slag N2, as sam- all the chemicals used were reagent grade (Lach-ner, Czech Repub- pled directly in the tailing ponds, was already crushed and milled lic and Merck, Germany). Procedural blanks were run simulta- with a grain size of <0.2 mm (Fig. 2c). For the sake of consistency, neously. The accuracy of the dissolution/analytical procedure was sample N6 was crushed and milled prior to the leaching test, controlled by G2 (granite) reference material certified by the US although it was known that this sample preparation would result Geological Survey and was generally better than <10% relative in a significantly higher reactivity. Both samples were subse- standard deviation (RSD) for all the elements. quently manually homogenized with a pestle in an agate mortar to obtain approximately the same grain size. The granulometry 2.4. Mineralogical analysis of primary phases of both samples was measured by the low-angle laser light scatter- ing (LALLS) method using a Fritsch Analyzette 22 NanoTec laser 2.4.1. X-ray diffraction analysis particle sizer, equipped with a He–Ne laser (k = 632.8 nm). About 0.5 g of pulverized sample was used for the X-ray powder Granulometric investigations of sample N2 yielded the following diffraction analysis (XRD), performed on a PANanalytical X’Pert Pro size distributions: <5 lm (30%), 5–10 lm (23%), 10–20 lm (33%), diffractometer using CuKa radiation, at 40 kV and 30 mA, over the 20–100 lm (14%). The granulometry of sample N6 was as follows: range 5–80° 2h with a step of 0.02° and counting time of 150 s in <5 lm (45%), 5–10 lm (24%), 10–20 lm (26%), 20–100 lm (5%). each step (X’Celerator detector). The X’Pert HighScore, version The obtained specific surfaces for N2 and N6 were 1.6 and 1.0d equipped with the JCPDS PDF-2 database (ICDD, 2003) was 2.2 m2/g, respectively. The results of the leaching test with these used for qualitative analysis of the diffractograms. fine-grained slags correspond to the ‘‘worst” leaching scenario, as could be expected in the tailing ponds filled with granulated and 2.4.2. Microscopy and electron microprobe analysis crushed/milled slag. The significantly lower specific surface area Slag samples were embedded in resin and prepared as polished of 20-cm-large slag fragments (sample N6) in the dumps will re- thin sections for microscopic observation. The polished thin sec- sult in slower release rates of contaminants than obtained by the tions were examined under a Leica DM LP polarizing microscope leaching test. in transmitted and reflected light and subsequently studied under The leaching experiment denoted as ‘‘EN 12457” was carried a CamScan scanning electron microscope (SEM) equipped with an out according to the experimental protocol of the static leaching Oxford Link energy dispersion spectrometer (EDS). Quantitative test described in detail by the European standard EN 12457 (part microanalyses were performed using a Cameca SX-100 electron 2) adopted recently by the Czech legislation (EN 12457, 2002). A microprobe (EPMA). For the silicate and oxide phases, the analyti- mass of 10 g of solid was placed in the high-density polyethylene cal conditions were: accelerating voltage 15 kV, beam current (HDPE) reactor and 100 mL of MilliQ+ deionised water was added 10 nA, and counting time 10 s. The following standards were used: to maintain a L/S (liquid/solid) ratio of 10. The leaching test was jadeite (Na), diopside (Mg, Ca), synthetic SiO2 (Si), synthetic Al2O3 performed at 22 ± 3 °C for 24 h and the reactors were gently agi- (Al), leucite (K), tugtupite (Cl), apatite (P), barite (S), rutile (Ti), tated (60 rpm). After the experiments, the reactors were centri- magnetite (Fe), spinel (Cr, Mg), cuprite (Cu), willemite (Zn), syn- fuged in order to settle fine slag particles at the bottom before thetic CdS (Cd), crocoite (Pb) and synthetic GaAs (As). For metals the filtration and measurements. After opening the reactor, the and sulphides, the analytical conditions were: accelerating voltage pH, Eh and specific conductivity were measured in the leachate. 20 kV, beam current 10 nA and counting time 10 s for all the ele- The pH and Eh values were determined using a Schott Handylab ments. The following standard set was used for metals and sulp- 1 pH meter equipped with a Schott L 7137 A combined electrode hides: pyrite (S, Fe), pure nickel (Ni), pure copper (Cu), sphalerite and a Schott PT 737 A (Pt–Ag/AgCl) redox electrode, respectively. (Zn), synthetic GaAs (As), pure silver (Ag), pure tin (Sn), pure anti- The temperature and specific conductivity were measured using mony (Sb) and galena (Pb). The calculation of Fe2O3/FeO ratio in a Schott Handylab LF 1 conductometer equipped with a LF 513 T spinels was performed using the methodology described in detail measuring cell and a temperature detector. The supernatant was 4 V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15
Fig. 2. Scanning electron micrographs in backscattered electrons of slags with emphasis on silicate and oxide phases: (a) zoned Cu-bearing oxides (Cu–Fe oxide, Cu–Cr oxide, mcconnelite, see corresponding analyses in Table 4) and feldspar crystals within the Pb-arsenate matrix (sample N6, Slag I); (b) glass inclusion with zoned spinel oxides (granulated sample N1, Slag III); (c) milled slag N2 (Slag III) composed of glassy fragments with small sulphide/metallic droplets and/or dendritic spinels and olivine laths; (d) cross-like wuestite dendrites and euhedral spinels within the melilite crystal associated with olivine (sample N9-2, Slag II); (e) olivine crystals within the glassy matrix with small wurtzite/galena droplets (sample N9-1, Slag II); (f) euhedral melilite within the matrix composed of olivine laths with trapped spinel crystals, residual glass and small metallic/sulphide inclusions (sample N9-2, Slag II). Abbreviations: mcc – mcconnelite, spl – spinel, ol – olivine, as – arsenate, f – feldspar, Pb-f – Pb feldspar, mel – melilite, gl – glass, w – wuestite, gn – galena, wz – wurtzite. then filtered to 0.45 lm (MilliporeÒ) using a Sartorius polycarbon- experimental protocol defined by USEPA (1994). With respect to ate filtering device and split into two aliquot parts, for cation and the alkalinity and buffering capacity of slags, measured according trace element analysis (diluted and acidified to pH <2 by HNO3), to the TCLP procedure, solution #1 with pH 4.93 ± 0.05 was used. and for anion analysis and alkalinity measurements. The obtained The leaching solution was prepared by adding 5.7 mL of acetic acid results were compared with regulatory levels for leachates defined (reagent grade, Merck, Germany) to 500 mL of deionised water for non-hazardous and hazardous waste by the EU legislation (MilliQ+), subsequent addition of 64.3 mL of 1 mol L 1 NaOH (re- (compiled in Van Gerven et al., 2005). agent grade, Lach-ner, CZ) and dilution to 1 L. An amount of 5 g The second leaching experiment denoted as ‘‘TCLP” was con- was placed in the reactor and 100 mL of leaching solution were ducted according to the toxicity characteristic leaching procedure added to attain an L/S ratio of 20. The experiment was conducted V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 5 for 18 h at 22 ± 3 °C and the reactors were agitated at 30 rpm. The 3. Results measurement of physico-chemical parameters and leachate pro- cessing (filtration, acidification) for analyses were analogous to 3.1. Bulk chemical composition those of the previously described leaching test. The obtained re- sults were compared with the maximum permissible concentra- The chemical compositions of the slag samples are reported in tions for toxicity characteristics defined by USEPA (2005). Table 1. These materials are mainly composed of SiO2 (9.82– Major cations were analysed by FAAS (Varian, SpectrAA 280 FS) 35.50 wt.%), Fe2O3 tot. (8.42–64.22 wt.%), CaO (3.85–17.44 wt.%) and trace elements (Al, As, Cd, Cr, Cu, Fe, Mn, Sb, Pb, Zn) were and Al2O3 (2.82–18.50 wt.%). The slag samples are also significantly determined by ICP-MS (VG Elemental PQ3). Quality control/quality enriched in metals and metalloids: Pb (0.97–18.38 wt.%), Cu (0.49– assurance (QC/QA) of the analytical measurements was controlled 12.18 wt.%), Zn (2.8–12.09 wt.%), As (0.09–7.59 wt.%), Sb (67– by a NIST 1640 standard reference material (Trace Elements in 2175 mg/kg). As some of metallic elements are present in various Water) and a Merck IV solution (ICP multielement standard IV, phases (silicates, glass, oxides, sulphides, metallic compounds), Merck). The accuracy of the measurement was generally better the sums of oxides are generally low and are not given in Table 1. than 10% relative standard deviation (RSD). The alkalinity of the Differences in the chemical compositions between the histori- samples was measured by back titration using the 0.05 M HCl (re- cally different slag groups reflect the variation in furnace charges agent grade Lach-ner, CZ) with a Schott TitroLine Easy automatic and smelting conditions over time. In contrast to other samples, 2 titrator. The concentrations of Cl and SO4 in the leachates were the oldest slag (Slag I, sample N6) is significantly enriched in Al measured by a Dionex ICS-2000 liquid chromatography instrument (18.5 wt.% Al2O3), impoverished in Fe and Ca and also has high con- (HPLC). centrations of Pb, Cu, Zn and As (Table 1). This chemical composi- tion indicates relatively low efficiency of metal recovery and 2.7. Thermodynamic modelling probably also the fact that carbonate ore was processed without Ca additives routinely used for sulphide ore processing in modern The PHREEQC-2 geochemical speciation-solubility code, version pyrometallurgy (Ettler et al., 2000, 2001). A low concentration of S 2.13.2 for Windows (Parkhurst and Appelo, 1999) was used to (<0.01 wt.%) is consistent with the processed carbonate and oxide determine the speciation of contaminants and the degree of satu- ores. With the exception of sample N8 (particularly enriched in ration of leachates with respect to the mineral phases. The Min- Fe and impoverished in Si), slags of group II exhibit similar chem- teq.v4.dat database (derived from MINTEQA2, version 4 released ical compositions to Pb and Cu slags from other smelting sites (Et- by USEPA in 1999) with the thermodynamic data of acetic com- tler et al., 2000; Manasse et al., 2001; Lottermoser, 2002). The S plexes was used for model calculations. The thermodynamic concentrations ranging from 0.7 to 3.85 wt.% indicate that metals database was supplemented by the solubility products of Ca- and dissolved in slag melt are present not only in silicate matrix and Pb-arsenates and antimonates recently compiled by Cornelis pure metallic phases, but can be also associated with S in the form et al. (2008). In addition, the solubility products of secondary of sulphides (PbS, ZnS, CuS) (Ettler et al., 2001; Manasse and Mel- Cu-, Zn- and Pb-arsenate minerals determined by Magalhães lini, 2002a). This was also confirmed by the mineralogical investi- et al. (1988) were included in the PHREEQC-2 calculations. gation of slags (see below). Particularly high S (and metallic
Table 1 Bulk chemistry of selected slag samples.
Sample N1 N2 N6 N7 N8 N9-1 N9-2 N10 N11 N12 N13 type Slag III Slag III Slag I Slag II Slag II Slag II Slag II Slag II Slag II Slag II Slag II wt.%
SiO2 14.68 35.50 27.68 20.86 9.82 28.52 29.12 23.24 21.60 23.20 24.30
TiO2 0.25 0.20 0.70 0.15 0.15 0.20 0.16 0.20 0.15 0.15 0.13
Al2O3 4.70 3.85 18.50 2.82 3.92 3.86 4.11 3.75 3.60 4.10 4.18 a Fe2O3 tot. 32.47 28.81 8.42 25.25 64.22 35.94 30.24 37.90 42.32 39.43 42.98 FeO 9.49 17.56 4.43 13.98 21.22 21.49 22.33 28.69 28.78 29.65 31.08 MnO 0.08 0.28 0.05 0.16 0.10 0.24 0.25 0.26 0.25 0.26 0.25 MgO 1.43 4.65 0.69 2.34 1.55 5.01 5.10 3.25 3.23 3.23 3.49 CaO 4.50 11.53 3.85 8.48 2.29 17.44 17.34 12.98 12.26 12.99 13.36
Na2O 0.26 1.87 0.25 0.74 4.31 1.57 1.57 0.81 0.80 0.80 0.84
K2O 0.30 0.35 0.93 0.27 0.10 0.61 0.61 0.63 0.57 0.58 0.61
P2O5 0.59 0.42 0.56 0.39 0.25 0.22 0.23 0.28 0.27 0.25 0.27 S tot. 6.51 0.24 <0.01 3.85 0.65 0.77 0.70 1.99 1.82 1.98 1.68 mg/kg As 24,060 17,445 28,750 75,865 13,135 2055 930 3420 3165 3210 935 Ba 690 1275 1280 825 240 3150 3075 1065 1395 1965 1060 Bi 56 0.54 78 3.5 0.36 0.37 <0.1 0.55 0.85 0.88 0.31 Cd 1364 65 6939 412 34 19 12 26 15 16 14 Co 149 216 27 129 4244 130 119 142 169 168 127 Cr 368 826 232 789 1351 252 267 201 1576 238 247 Cu 27,488 7938 106,200 121,850 47,650 12,763 4963 23,263 19,613 19,513 11,163 Ga 3.1 48 3.1 19 0.97 12 13 11 7.0 7.2 12 Ge 9.3 123 15 33 2.4 8.6 9.0 7.8 10 11 9.2 Ni 56 116 32 168 252 32 8.0 49 33 45 12 Pb 103,900 40,300 183,800 55,900 62,750 37,213 12,575 24,713 43,163 10,263 9763 Sb 2175 1071 2110 821 729 341 109 174 219 214 67 Zn 28,188 60,275 33,125 45,925 77,550 96,250 74,750 93,000 110,500 96,350 120,850
a Total Fe (including FeO + Fe2O3). 6 V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15
Table 2 Primary phases from the Tsumeb slags obtained by XRD and EPMA (normal – both methods, italics – only EPMA)a.
Group Phase Chemistry Slag type Slag I Slag II Slag III
Silicates Fayalite Fe2SiO4 ++ +
Monticellite CaMgSiO4 +++
Melilite Ca2(Mg,Fe,Zn)Si2O7 +++ +
Anorthite CaAl2Si2O8 +++ b Unnamed (PDF 087-1003) PbAl2Si2O8 ++ Amorphous glass Si–Ca–Fe–Al ++ +++
Oxides Spinel series (Zn,Mg,Fe,Cu)(Fe,Al)2O4 +++ + +++ Wuestite FeO + 1+ 3+ 3+ Delafossite–mcconnelite Cu (Fe ,Cr )O2 ++ Sulphides Galena PbS + + Wurtzite ZnS + Sphalerite ZnS +
Chalcopyrite CuFeS2 +
Pyrrhotite Fe1 xS + Cubanite CuFe2S3 + Covellite CuS + Elements Lead Pb + Copper Cu +
Others Domeykite a Cu3As +
Cu5Sb Cu5Sb +
Cu3(Sn,Sb) Cu3(Sn,Sb) +
Fe2As Fe2As + FeAs FeAs +
Unidentified Ca–Pb arsenates Ideal formula (Pb,Ca,Fe)3(AsO4)2 H2O ++
a Relative abundance: +++ dominant phase, ++ common phase, + trace phase. b Unnamed phase (Pb feldspar) according to Benna et al. (1996).
element) contents were observed for sample N1 (slag III) (Table 1), glass and sulphide-metallic inclusions (Ettler et al., 2001; Manasse indicating the lower efficiency of sulphide-bound metal recovery and Mellini, 2002b; Piatak et al., 2004). In general, spinels are the due to quenching of the slag melt by granulation. first crystallizing phases, forming large zoned crystals (up to 50 lm across; Fig. 2b) or small dendrites (Fig. 2c) or euhedral crys- 3.2. Mineralogy tals (Fig. 2d). Occasionally, dendritic wuestite aggregates form at the beginning of the slag melt solidification (Fig. 2d). The crystalli- 3.2.1. Slag petrography sation sequence follows with the formation of large euhedral meli- In the present paper, mineral names are given for phases having lites (Fig. 2d and f) or skeletal to lath-like olivines (Fig. 2e,f). Glass natural equivalents. Significantly distinct phase compositions were solidifies last and contains small inclusions of sulphides (mainly observed for slag I compared to younger slags as revealed by XRD galena, wurtzite, bornite, digenite) or intermetallic compounds analysis (Table 2), indicating both different conditions of slag for- (Fig. 2c, e, and f). Mattes (sulphide-rich materials) and speiss (arse- mation (e.g., the chemical composition of the slag melt, the cooling nide-rich materials) were present as droplets and fragments of var- regime), and also possible alteration processes at the dumping site ious sizes composed of symplectitic intergrowths of sulphides, (the presence of secondary phases). Interestingly, slag I is mainly metals and intermetallic compounds, and embedded within the composed of feldspars (with anorthite and Pb-feldspar structures), silicate slag (Fig. 3a–c). The rare occurrence of massive matte Cu-bearing spinels, delafossite, an As-bearing matrix phase (corre- material associated with silicate slags (sample N7) was also ob- sponding to arsenate) and traces of galena (Table 2). In contrast, served in one case (Fig. 3d): this heavy material is composed of sec- slags II and III are predominantly composed of typical high-tem- ondary litharge (PbO) filling the cavities between the pure Pb perature Ca–Fe silicates (olivine-type phases, melilite), amorphous crystals and intermetallic compounds (Cu3As). silicate glass, spinel-type oxides, wuestite and also trace sulphides (e.g., galena, wurtzite/sphalerite, bornite), pure metals (Pb, Cu) and 3.2.2. Crystal chemistry various intermetallic compounds (Table 2). 3.2.2.1. Olivines. The chemical composition of olivine-type phases
Slag I. Microscopic investigation shows that feldspars forming varies from the nearly pure fayalite (Fe2SiO4) to kirschsteinite (CaF- 50- to 300-lm large euhedral crystals are the dominant phase in eSiO4) – monticellite (CaMgSiO4) solid solution as revealed by these slags (Fig. 2a). Oxides belonging to the Cu-spinel and delafos- EPMA (Table 3). They occur as a dominant phase in slags II and site–mcconnelite series form euhedral crystals up to 80 lm in size III. Olivine crystals are slightly zoned (Fig. 2e and f) with bright and are often zoned (Fig. 2a). Another generation of prismatic feld- rims corresponding to Ca-poor olivine and dark cores correspond- spar crystals (brighter on SEM images than the previous ones) is ing to Ca-rich kirschsteinite (up to 28.51 wt.% CaO). All the ob- associated with Cu-spinels and the slag matrix. The latter is com- served olivines are Zn-bearing; the fayalite end-members are posed of As-bearing phases, often enriched in Pb, Cd, Cu and Ca generally most enriched (up to 8.74 wt.%). Zinc can enter into octa- (Fig. 2a), probably in the form of arsenates, indicating that the slag hedral sites of the olivine structure and substitutes for Fe2+ (Ettler underwent some alteration process. Occasional inclusions of gale- et al., 2000, 2001). Previous works on smelting slags showed that na (PbS) are observed in the slag matrix (Fig. 2a). these phases are the most common silicates that form large skele- Slags II and III are composed of phases commonly reported in tal crystals or laths in slowly crystallizing slags, and dendrites in papers devoted to smelter slag mineralogy: spinel oxides, silicates, quenched slags (Ettler et al., 2001, in press; Manasse et al., 2001). V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 7
Fig. 3. Scanning electron micrographs in backscattered electrons of sulphide phases: (a) Cu sulphide inclusions associated with sphalerite and overgrown with galena in myrmekites (sample N7, Slag II); (b) inclusion composed of bornite with inclusions of wurtzite and pure Pb droplets (sample N11, Slag II); (c) large sulphide/metallic inclusion composed of galena-CuS myrmekite, crystals of covellite, sphalerite, Cu3As and pure Pb droplets (sample N9-1, Slag II); (d) metallic Pb altered to litharge (PbO) with residual crystals of chalcocite and Cu3As intermetallic phase (matte associated with slag sample N7, Slag II). Abbreviations: gl – glass, CuS – copper sulphide, gn – galena, Pb – metallic Pb, wz – wurtzite, bn – bornite, sph – sphalerite, co – covellite, cc – chalcocite.
3.2.2.2. Melilite. The composition of melilite varies significantly and presence of feldspars in slags is rather scarce and was observed often corresponds to a solid solution of predominant åkermanite only by Sáez et al. (2003), who detected plagioclase in Cu-smelting
(CaMgSi2O7) and hardystonite (CaZnSi2O7) with minor amounts slags from Spain, and by Puziewicz et al. (2007), who described Ca, of Na-melilite (NaCaAlSi2O7), gehlenite (Ca2Al2SiO7) and ferro- Ba and Pb feldspars in Zn-slags from Poland. The shapes of feldspar åkermanite (CaFeSi2O7)(Table 3). Thus, melilite seems to effi- crystals from Namibian slags seem to indicate that these phases ciently accumulate Zn, exhibiting ZnO concentrations up to crystallized directly from the melt. In other cases, feldspars can 11.64 wt.%. Similar Zn contents in melilite were observed in other be residual, i.e. from unmelted gangue, especially in medieval slags smelting slags resulting from similar technological processes (Et- such as those studied by Sáez et al. (2003). The presence of angular tler et al., 2001, 2002). grains of residual feldspars indicate that (i) the temperatures in the smelting furnaces were not sufficiently high and/or (ii) the dura-
3.2.2.3. Feldspars. Anorthite (CaAl2Si2O8) and an unnamed phase tion of melting was not sufficient to completely melt the furnace corresponding to a ‘‘Pb-feldspar” (PbAl2Si2O8), as reported by Ben- charge containing gangue minerals (Hauptmann, 2007; Ettler na et al. (1996), were observed only in slag I. According to EPMA, et al., in press). their composition is rather variable and the Pb concentration can be as high as 34 wt.% PbO (Table 3). They certainly crystallized 3.2.2.4. Silicate glass. The presence of glass was observed mainly in from the slag melt (large euhedral crystals) and represent (to- slags II and III and confirms the rapid cooling of the slag melt (gran- gether with spinels) the early phases crystallizing in the slag I. ulation). Its composition is rather variable and, similar to other sil- The presence of two generations of chemically different feldspars icates in slags; the glass is enriched in Ca and Fe (Table 4). indicates significant changes in the slag melt during solidification. Furthermore, it is the principal carrier of contaminants such as The first generation of large euhedral feldspars is Pb-depleted Pb (up to 11.8 wt.% PbO), Zn (up to 16.2 wt.% ZnO), Cu (up to
(analyses 18/1 and 28/1 in Table 3) and the second generation of 1.78 wt.% CuO) and As (up to 2.99 wt.% As2O3). Low analytical to- smaller Pb-rich feldspars associated with spinels is probably the tals of some EPMA indicate the presence of H2O or possibly the product of crystallisation from the melt strongly enriched in Pb, presence of trivalent Fe in the glass structure (Table 4). Si and Al (analyses 17/1, 20/1, 24/1 and 30/1 in Table 3). Benna et al. (1996) showed that the disordered structure of synthetic 3.2.2.5. Oxides. Oxides in Namibian slags are mainly represented by Pb-feldspar is a product of rapid cooling (quenching), as it can oc- spinel-type compounds for slags II and III. EPMA revealed that their cur during the slag melt solidification. It should be recalled that the composition is rather variable (Table 5). Whereas the compositions 8 V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15
Table 3 Selected microprobe analyses of primary silicates (olivines, melilites and feldspars). Structural formulae (apfu – atoms per formula unit) were calculated on the basis of four (olivine), seven (melilite) and eight oxygens (feldspars).
Spot analysis 58/1 59/1 67/1 46/1 49/1 57/1 17/1 18/1 20/1 24/1 30/1 28/1 Sample N9-2 N9-2 N9-2 N7 N8 N9-2 N6 N6 N6 N6 N6 N6 Slag type Slag II Slag II Slag II Slag II Slag II Slag II Slag I Slag I Slag I Slag I Slag I Slag I Phase Olivine Olivine Olivine Melilite Melilite Melilite Feldspar Feldspar Feldspar Feldspar Feldspar Feldspar
SiO2 wt.% 34.07 30.46 32.11 SiO2 40.72 40.14 40.36 SiO2 47.67 44.23 44.09 44.04 39.02 44.33
TiO2 – 0.02 – TiO2 –––TiO2 0.02 – 0.05 0.14 0.04 –
Al2O3 – 0.02 0.11 Al2O3 1.48 2.58 3.62 Al2O3 22.58 32.34 23.96 18.78 20.39 31.22 FeO 21.68 50.23 42.10 FeO 2.99 3.91 4.75 FeO 1.03 0.89 1.24 0.78 0.41 0.70 MnO 0.48 0.63 0.79 MnO – 0.06 0.04 MnO – – – – 0.03 0.07 MgO 10.74 6.55 13.19 MgO 5.98 3.01 3.90 MgO – – 0.04 0.11 0.04 – CaO 28.51 4.04 4.13 CaO 37.14 33.95 34.90 CaO 3.56 16.21 5.93 2.38 1.39 14.83
Na2O 0.17 0.32 0.23 Na2O 0.63 2.91 2.15 Na2O 0.62 0.67 0.86 0.43 0.27 0.84
K2O 0.01 0.02 0.01 K2O 0.06 – 0.11 K2O 6.41 0.45 2.56 2.26 2.78 0.38
P2O5 0.01 0.06 0.14 P2O5 0.01 0.18 0.05 P2O5 0.03 0.05 0.18 0.15 0.06 0.03
Cr2O3 0.06 – 0.05 Cr2O3 0.07 – 0.12 Cr2O3 – 0.06 0.13 0.07 – 0.11 PbO – – – PbO 0.14 0.10 0.05 PbO 16.54 3.66 19.33 29.54 34.00 5.68 ZnO 4.87 8.74 7.62 ZnO 10.37 11.64 8.94 ZnO – 0.05 0.13 0.27 0.43 – CuO – – – CuO – – – CuO 0.30 0.39 0.71 0.91 0.49 0.55 CdO 0.10 – 0.04 CdO – – 0.07 CdO 0.05 0.15 0.14 0.14 0.16 0.27
As2O3 –––As2O3 –––As2O3 –– 0.32 0.61 0.03 –
SO2 –––SO2 0.02 – – SO2 –––––– Cl–––Cl–––Cl0.02 – 0.02 – 0.03 – Total 100.70 101.06 100.54 Total 99.61 98.48 99.06 Total 98.83 99.14 99.67 100.58 99.55 99.00 Si apfu 0.996 0.983 0.989 Si 1.985 2.000 1.974 Si 2.554 2.131 2.411 2.605 2.441 2.170 Fe 0.530 1.355 1.084 Al 0.085 0.152 0.209 Al 1.426 1.836 1.544 1.310 1.504 1.801 Mn 0.012 0.017 0.021 Fe 0.122 0.163 0.194 Fe 0.046 0.036 0.057 0.038 0.021 0.029 Mg 0.468 0.315 0.606 Mn 0.000 0.003 0.002 Ca 0.204 0.837 0.347 0.151 0.093 0.777 Ca 0.893 0.140 0.136 Mg 0.435 0.224 0.284 Na 0.065 0.063 0.091 0.049 0.033 0.080 Cr 0.001 0.000 0.001 Ca 1.940 1.813 1.829 K 0.438 0.028 0.179 0.170 0.222 0.024 Zn 0.105 0.208 0.173 Na 0.060 0.281 0.204 Pb 0.239 0.047 0.285 0.470 0.573 0.075 K 0.004 0.000 0.007 Zn 0.000 0.002 0.005 0.012 0.020 0.000 Proportion of end-members (mol.%) Pb 0.002 0.001 0.001 Cu 0.012 0.041 0.029 0.041 0.023 0.020 La 47 8 7 Zn 0.373 0.428 0.323 Cd 0.001 0.003 0.003 0.004 0.005 0.006 Fo 25 17 33 Fa + Te 28 75 60 Proportion of end members (mol.%) Proportion of end members (mol.%) Ha 37 40 31 Ab–Or 50 9 27 24 26 10 Ak 43 21 27 An 50 91 73 76 74 90 FeAk 12 15 19 Gh4910 SM 5 14 14
Symbols used: – not detected; La larnite (Ca2SiO4), Fo forsterite (Mg2SiO4), Fa fayalite (Fe2SiO4), Te tephroite (Mn2SiO4), Ha hardystonite (Ca2ZnSi2O7), Ak åkermanite
(Ca2MgSi2O7), FeAk ferro-åkermanite (Ca2FeSi2O7), Gh gehlenite (Ca2Al2SiO7), SM soda-melilite ((Ca,Na)AlSi2O7), Ab albite (NaAlSi3O8), Or orthoclase (KAlSi3O8), An anorthite
(CaAl2Si2O8).
Table 4 Selected microprobe analyses and average compositions of all the studied glasses.
Spot analysis 1/1 8/1 38/1 43/1 50/1 53/1 54/1 62/1 73/1 n =16 Sample N1 N2 N7 N7 N8 N8 N8 N9-2 N9-2 Min Max Mean Slag type Slag III Slag III Slag II Slag II Slag II Slag II Slag II Slag II Slag II
SiO2 25.14 40.22 38.35 28.21 39.62 35.72 37.05 40.14 39.57 25.14 45.49 37.50
TiO2 0.25 0.12 – 0.28 – 0.10 0.03 – – – 0.28 0.07
Al2O3 4.48 1.97 5.97 4.31 12.93 0.77 6.67 5.36 7.27 0.65 12.93 5.55 FeO 24.71 32.47 21.86 21.84 11.80 18.50 20.96 5.07 5.02 5.02 58.48 20.96 MnO 0.10 0.22 1.32 0.36 0.07 0.14 0.16 0.14 – – 1.32 0.24 MgO 3.44 0.32 0.27 5.34 2.84 1.31 2.16 0.93 0.65 0.03 5.34 1.51 CaO 19.88 5.16 8.30 18.75 1.77 15.95 2.32 32.06 30.70 0.81 32.06 12.01
Na2O 0.96 1.01 3.53 1.31 15.18 12.25 10.08 3.92 5.00 0.40 15.18 5.50
K2O 0.07 0.21 1.37 0.64 0.14 0.27 0.13 0.10 0.14 0.07 6.74 1.04
Cr2O3 – 0.09 0.06 0.36 0.04 – – 0.07 – – 0.36 0.04
P2O5 0.32 0.09 1.48 0.36 0.12 1.05 0.33 0.29 0.71 0.08 1.48 0.55 PbO 7.85 11.80 5.26 1.30 0.06 0.32 0.25 0.14 0.08 – 11.80 2.29 ZnO 9.12 3.23 11.42 16.21 14.70 9.19 10.94 11.62 10.51 2.42 16.21 9.29 CuO 1.78 – – 0.08 – 0.09 – – – – 1.78 0.15 CdO 0.05 0.03 – 0.10 – – 0.03 0.07 – – 0.10 0.03
As2O3 1.56 0.56 0.58 – 0.07 1.69 0.16 – – – 2.99 0.54
SO2 0.01 0.50 0.05 0.90 – 0.38 0.05 0.01 0.01 – 0.90 0.24 Cl 0.01 – 0.02 0.01 – 0.02 – 0.02 0.01 – 0.20 0.03 Total 99.73 97.99 99.83 100.36 99.35 97.71 91.31 99.95 99.66 90.39 102.79 97.55
Symbols used: – not detected. V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 9
Table 5 Selected microprobe analyses of primary oxides.
Spot 2/1. 3/1. 7/1. 42/1. 44/1. 61/1. 68/1. 66/1. 13/1. 35/1.c 11/1.d 3/2.e 4/2.f 13/2.g analysis Sample N1 N1 N2 N7 N7 N9-2 N9-2 N9-2 N6 N6 N6 N6 N6 N6 Slag type Slag III Slag III Slag III Slag II Slag II Slag II Slag II Slag II Slag I Slag I Slag I Slag I Slag I Slag I Phase Spinel Spinel Spinel Spinel Spinel Spinel Spinel Wuestite Mcconnelite Cu–Fe Cu–Cr Cu–Fe Zn–Cu–Al–Fe Cu–Fe oxide oxide oxide oxide oxide
SiO2 wt.% 0.18 0.09 1.88 0.16 – 0.39 0.26 0.38 – 0.05 – 0.06 0.54 0.14
TiO2 0.25 0.14 0.26 0.36 0.16 0.88 2.00 0.41 0.35 5.98 0.55 4.89 0.23 0.63
Al2O3 9.30 5.74 0.78 48.77 14.08 36.97 25.94 0.67 9.21 3.40 13.41 4.90 38.57 9.66
Cr2O3 20.14 1.02 0.36 3.26 52.67 0.08 0.68 – 33.73 – 45.56 0.12 0.08 3.61
Fe2O3 31.98 50.28 62.86 9.13 5.09 21.58 32.77 – 7.99 27.15 8.09 32.03 13.25 52.79 FeO 15.31 23.13 33.31 6.83 8.40 12.84 18.98 87.09 – – 5.42 – 7.68 – MnO – 0.29 0.04 0.04 – 0.06 0.21 0.40 – – – – 0.11 0.14 MgO 6.36 5.39 0.34 6.58 13.17 3.68 4.18 0.35 0.08 0.71 6.28 0.49 2.46 0.99 ZnO 15.63 13.93 1.26 25.41 6.17 21.31 15.69 8.42 – 0.63 3.65 0.29 20.20 4.90
CuO (Cu2O) 0.20 0.25 0.06 – 0.10 0.05 – – (50.19) 55.20 13.20 54.40 10.05 20.92 CdO – – – – – – – – – – 0.08 0.06 – 0.24
Na2O 0.42 0.37 0.07 0.67 0.10 0.72 0.43 0.19 – – 0.14 – 0.49 0.10
K2O – – – – 0.02 0.01 – 0.01 – – – – 0.03 0.04 CaO 0.17 0.29 0.08 0.03 0.02 0.36 0.18 0.42 – 0.05 0.02 0.02 0.03 0.03 PbO – – – – – 0.07 – – – 0.35 0.08 0.12 0.12 –
P2O5 –– 0.03 – – – 0.01 – – 0.01 – 0.02 0.01 –
SO2 – – – – – 0.02 – – – 0.02 – – – –
As2O3 – – – – – 0.04 – – – 0.15 – 0.12 – – Total 99.94 100.92 101.33 101.23 99.97 99.06 101.32 98.33 101.54 93.68 96.47 97.51 93.84 94.16 a H2O 6.32 3.53 2.49 6.16 5.84 Si apfu 0.007 0.003 0.071 0.005 0.000 0.012 0.008 0.009 0.000 0.004 0.000 0.004 0.019 0.020 Ti 0.007 0.004 0.007 0.008 0.004 0.021 0.050 0.007 0.006 0.303 0.014 0.234 0.006 0.069 Al 0.394 0.254 0.034 1.696 0.533 1.406 1.011 0.019 0.249 0.270 0.553 0.367 1.602 1.655 Cr 0.573 0.030 0.011 0.076 1.336 0.002 0.018 0.000 0.613 0.000 1.261 0.006 0.002 0.414 Fe(3) 0.867 1.424 1.782 0.203 0.123 0.525 0.817 0.000 0.139 1.379 0.214 1.532 0.352 5.787 Fe(2) 0.460 0.727 1.047 0.169 0.225 0.347 0.525 1.774 0.000 0.000 0.159 0.000 0.226 0.000 Zn 0.415 0.386 0.035 0.554 0.146 0.508 0.383 0.151 0.000 0.031 0.094 0.014 0.526 0.526 Mg 0.341 0.302 0.019 0.290 0.630 0.177 0.206 0.013 0.003 0.071 0.327 0.046 0.129 0.214 Cu 0.006 0.007 0.002 0.000 0.002 0.001 0.000 0.000 0.969 2.811 0.349 2.608 0.267 2.298 Ab 1.222 1.422 1.103 1.012 1.004 1.033 1.114 1.938 0.971 2.913 0.929 2.668 1.148 3.037 Bb 1.847 1.716 1.905 1.988 1.996 1.967 1.904 0.036 1.007 1.956 2.042 2.142 1.982 7.946 sum 3.070 3.138 3.008 3.000 3.000 3.000 3.019 1.974 1.978 4.869 2.972 4.810 3.130 10.983 oxygens 4 4 4 4 4 4 4 1 2 6 4 6 4 15
H2O 2.84 0.82 1.05 1.45 5.66
Symbols used: – not detected. a H2O by difference. b A and B correspond to structural positions in the oxide structure (e.g., AB2O4 for the spinel structure). c Chemical interpretation: Cu3Fe2O6 3H2O (3CuO Fe2O3 3H2O). d Chemical interpretation: (Cu,Mg)Cr2O4 H2O (CuO Cr2O3 H2O). e Chemical interpretation: Cu3Fe2O6 H2O (3CuO Fe2O3 H2O). f Chemical interpretation: (Zn,Cu,Mg)(Al,Fe)2O4 1.5H2O. g Chemical interpretation: Cu3Fe8O15 6H2O (3CuO 4Fe2O3 6H2O).
of spinel-type phases from slag III correspond either to Fe–Cr spi- 3CuO Fe2O3 3H2O, 3CuO Fe2O3 H2O, 3CuO 4Fe2O3 6H2O, CuO nels or to magnetite (Fe3O4), those in slag II exhibit a composition Cr2O3 H2O. A more complex spinel-like structure ideally corre- close to gahnite (ZnAl2O4) or Cr-bearing spinels (Table 5). The den- sponding to (Zn,Cu,Mg)(Al,Fe)2O4 1.5H2O was also detected in slag drites of Zn-rich wuestite (FeO) are also reported for slag II (Table I. With respect to the stable spinel-type phases without Cu, the Cu- 5; Fig. 2d). Spinel-type oxides from slag I are complex solid solu- bearing oxides seem to be more susceptible to hydration, subse- tions of Cu–Fe–Cr-end-members. The presence of a phase of the quent weathering and also probable release of Cu. delafossite (CuFeO2) – mcconnellite (CuCrO2) system, correspond- 1þ ing to the stoichiometric formula ðCu0:969Mg0:003Þ0:972ðAl0:249Cr0:613- 3.2.2.6. Sulphides and intermetallic compounds. The most commonly 3þ Fe0:139Ti0:006Þ1:007O2 was observed in this type of slag (analysis 13/1 observed sulphides are galena (PbS), wurtzite/sphalerite (ZnS), in Table 5, Fig. 2a). This phase has been commonly found in Cu- pyrrhotite (Fe1 xS) and various Cu or Cu–Fe sulphides (digenite bearing slags in Spain (Sáez et al., 2003) and its presence (with (Cu,Fe)9S5, cubanite CuFe2S3, covellite CuS, chalcocite Cu2S). Cu1+ in the delafossite or mcconnellite structures) indicates ex- Wurtzite is a high-temperature phase (>1020 °C) (Ettler and Johan, treme reducing conditions in the slag melt. This is compatible with 2003) and contains significant amounts of Fe (up to 15.4 wt.%) and the historical fact that blast furnaces at the beginning of the 20th minor amounts of Cu (up to 1 wt.%). Other sulphides are also en- century were fired with German or South African coke, inducing riched in Cu. Up to 2 wt.% Cu was observed in pyrrhotite (Fe1 xS) a highly reducing environment during smelting of the ore (Tsumeb and up to 2.98 wt.% Cu was detected in galena (PbS) (data not Corporation Ltd., 1987). Other complex Cu-bearing oxides are di- shown). Unlike the other Pb–Zn slags (Ettler et al., 2001), those rectly associated with mcconnellite-like phases (Fig. 2a). Lower to- from Tsumeb contain higher amounts of Cu-rich sulphides, as tals of their EPMA indicate the presence water in these compounds Cu-rich ores were treated in the Tsumeb smelter. Intermetallic (1–6 moles per formula unit). The EPMA of these compounds exhi- compounds mainly belonging to the Cu–As, Cu–Sb, Cu–Sn and bit the following structural formulae (see also Table 5): Fe–As binary systems were also observed: domeykite a (Cu3As), 10 V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15
Cu3(Sn,Sb), Cu5Sb (Fe,Cu)2As, FeAs. The elemental substitutions in ence of the majority of SEM-observed phases was also confirmed their structures correspond to the substitutions observed by Ettler by XRD as indicated in Fig. 4. In some places, slag I was covered and Johan (2003) in matte phases from primary Pb metallurgical by green-coloured phases corresponding to Cu and Cu–Pb arse- processes. nates (lammerite Cu3(AsO4)2 and bayldonite Cu3Pb(AsO4)2(OH)2; Fig. 4a and b). Other Cu-bearing secondary alteration products 3.2.2.7. Arsenates. According to EPMA, the As-bearing matrix phases were observed for the most of the Cu-rich slags of type II (N7, observed in the oldest Tsumeb slag (slag I) probably correspond to see also bulk chemical composition in Table 1): green olivenite complex Ca–Pb arsenates. An example of EPMA analysis of such a (Cu2AsO4OH) aggregates covering the altered Cu-rich parts of the phase is (Pb1.273Ca1.056Cu0.382Cd0.171Zn0.112K0.050Fe0.016Na0.010)3.070 slags (probably droplets composed of Cu sulphides and/or Cu-arse- [(As1.641P0.311Si0.029)1.981O4] H2O, corresponding to the ideal for- nides) (Fig. 4c) or a blue continuous crust formed by lavendulan mula close to (Pb,Ca,Fe,Cu,Zn,Cd)3(AsO4)2 H2O. Such phases would (NaCaCu5(AsO4)4Cl 5H2O) associated with white crystals of gyp- probably be chemically similar to tsumcorite [FePbZn(AsO4)2 H2O] sum (CaSO4 2H2O). Other slags from group II contain significantly and other arsenates, generally found in the oxidation zones of the less Cu and As and their alteration products appear mainly on Tsumeb ore deposit, and certainly indicate alteration processes in Pb-rich parts of slags forming platy white crystals of hydrocerus- the slags (see further section on weathering products). site (Pb3(CO3)2(OH)) (Fig. 4e). In addition, the presence of litharge (PbO) associated with minor ZnO and Cu-arsenates was also de- 3.3. Natural weathering products tected in these samples (Fig. 4f and g). The presence of secondary PbO was previously confirmed by EPMA in a polished section from In addition to alteration phases observed in polished sections of matte-rich parts of slag N7 (Fig. 3d). With the exception of rare slag I (sample N6) and described above (matrix arsenates and hy- rims of anglesite (PbSO4, micrograph not shown) developed on drated Cu oxides), the investigation of slag surfaces under a binoc- Pb droplets in slags III, no other alteration products were detected, ular microscope showed the development of secondary phases of probably reflecting the fact that these slags were exposed to dark green to yellow green, blue, red and white colours. The pres- weathering for a short period of time.
Fig. 4. Scanning electron micrographs of secondary alteration products developed on slag surfaces with corresponding EDS spectra and interpretations based on microanalysis and XRD results. V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 11
3.4. Leaching behaviour hazardous wastes (Fig. 5a). However, As exceeds the limit for non-hazardous waste (0.2 mg/L) for both slags and, in the case of 3.4.1. Leachability and speciation of contaminants sample N2, even the limit for hazardous waste (2.5 mg/L). For slag The results of leaching of selected metals (Cd,Cu,Pb,Zn) and N2, Sb leaching was higher than the limit for non-hazardous waste metalloids (As, Sb) from two slag samples using both leaching tests (0.07 mg/L) (Fig. 5a). and comparison with regulatory levels are reported in Fig. 5. Other The TCLP method showed that concentrations of some contam- possible metallic contaminants were present in low concentra- inants, in particular for slag N6, exceeded the regulatory levels gi- tions; for example Cr, although found in some slags in relatively ven by USEPA for Cd, Pb and As (1 mg/L, 5 mg/L and 5 mg/L, high concentrations (Table 1), yielded concentrations in leachates respectively). Unfortunately, these maximum levels are not de- close to the detection limit of the analytical method used fined for Cu, Zn and Sb. As an alternative method for evaluation (0.07 lg/L). This fact is probably related to the tight binding of Cr of contaminant leaching from slags, Piatak et al. (2004) compared in less soluble phases such as spinels. In general, the concentra- the concentrations in the obtained leachates with the levels for tions of metals in leachates were higher for the TCLP leaching test acute and chronic toxicity given by USEPA in the National Recom- than for the EN 12457 tests due to the lower equilibrium pH of the mended Water Quality Criteria (2006). If the same criteria are ap- solution (5.71 for N2 and 4.93 for N6) and probably also the effect plied to the present leachates, the concentrations in the TCLP of enhanced extraction by acetate. A higher equilibrium pH was leachates (and also in some EN 12457 leachates) are significantly observed for the EN 12457 leachates (6.78 for N2 and 6.54 for higher than the limit values in the acute toxicity guidelines for N6). The concentration levels of As and Sb were similar for both aquatic habitats defined by this USEPA regulation (in lg/L: As tests, probably due to the lower efficiency of the TCLP method 340, Cd 2, Cu 13, Pb 65, Zn 120, Sb not determined). compared to the EN 12457 protocol in extracting oxyanions from PHREEQC-2 speciation modelling indicates that the metal speci- wastes reported in the literature (e.g., Ghosh et al., 2004). ation varies according to the pH and concentrations of other ligands Comparison of the EN 12457 leaching results with the EU regu- (acetate, sulphate, carbonate). The prevailing species were the free latory levels shows that the concentrations of the metals are gen- ionic forms (Cd: 77–98%; Cu: 40–99%, Pb: 47–96%, Zn: of total speci- erally below the limits imposed for both non-hazardous and ation). Sulphate complexes accounted for up to 22 (Cd), 15 (Cu), 29 (Pb) and 20 (Zn)% of the total speciation and were important only in the EN 12457 leachates, i.e. at higher pH. Carbonate complexes 0 þ corresponding to the sum of the MeCO3 and MeHCO3 species were a 100 important only in Cu and Pb speciation with up to 35% and 44% of hazardous waste EN 12457 the total speciation. A very low fraction of acetate complexes was non-hazardous waste 20 mg/L observed (only 0.1–1.9% of the total speciation, with Me-acetate+ 10 mg/L 10 5 mg/L 5 mg/L 5 mg/L as the main species). The PHREEQC-2 predictions indicate that the 2 prevailing As species are H2AsO4 or HAsO4 according to the pH va- V 1 mg/L 2.5 mg/L lue of the leachate and Sb is mostly present in its oxidized form (Sb ) 1 0.5 mg/L 0.5 mg/L as a the SbO3 complex (i.e., SbðOHÞ6 ). 0.2 mg/L Waters sampled in situ in the smelter complex during the rainy 0.1 mg/L 0.07 mg/L season (Table 6) are slightly alkaline (pH 8) and exhibit particu- 0.1 concentration (mg/L) 0.01 Table 6 Chemical compositions of surface waters sampled in the Tsumeb smelter area during the rainy season (data from Walmsley Environmental Consultants, 2001).
0.001 Parameter Units Run-off from the smelter Water from evaporation Cd Cu Pb Zn As Sb and slag deposits area dams in the smelter pH Standard 8.0 8.2 b 1000 units TCLP ECa mS/m 75 507 nd b nd Alkalinity mg/L 227 160 As mg/L 6.0 16 100 Al mg/L <0.1 <0.1 Ca mg/L 91 469 Cd mg/L 1.8 0.13 10 Cl mg/L 16 634 5 mg/L 5 mg/L N2 Co mg/L <0.025 <0.025 N6 Cr mg/L <0.025 0.048 1 mg/L 1 Cu mg/L 0.60 0.14 nd F mg/L 1.5 2.2 Fe mg/L 0.21 0.13 concentration (mg/L) Hg mg/L <0.002 <0.002 0.1 K mg/L 3.4 19 Mg mg/L 33 205 Mn mg/L 0.16 0.56 0.01 Na mg/L 23 620 Cd Cu Pb Zn As Sb Ni mg/L <0.025 <0.025 NO3 mg/L <0.2 <0.2 element Pb mg/L 1.7 0.18 Se mg/L 0.010 0.022 2 Fig. 5. Results of experimental leaching procedures (EN 12457 and TCLP) and SO4 mg/L 157 1930 comparison with regulatory concentration levels (EU regulatory levels for non- Zn mg/L 2.2 0.25 hazardous and hazardous waste materials are taken from Van Gerven et al. (2005) a and TCLP maximum concentration levels are defined by USEPA (2005)). Sample N2 EC – electrical conductivity. b corresponds to recent slag and sample N6 corresponds to 100-a-old slag. Expressed as CaCO3. 12 V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 larly high concentrations of As (6–16 mg/L). Other toxic elements products on slag surfaces (Fig. 4). Typical phases observed in slags (Cd, Cu, Pb, Zn) also exceed the acute toxicity guidelines for aquatic and acting as efficient sorbents (hydrous ferric oxides, HFO and hy- habitats defined by the USEPA National Recommended Water drous Al oxides, HAlO) also yield positive saturation indices, Quality Criteria (USEPA, 2006). although the concentrations of Fe and Al were low in the leachates (<0.07 mg/L and <0.17 mg/L, respectively). However, their forma- 3.4.2. Solubility controls tion would require prolonged release of Fe from the slag, as ob- The results of the PHREEQC-2 calculation of the saturation in- served in other studies (Ettler et al., 2005). Especially in very old dex for possible solubility-controlling phases are given in Table slags (slag I), no HFO or HAlO were observed as secondary alter- 7. All the leachates and surface waters from the vicinity of the Tsu- ation products, because of the low Fe contents in the original slag meb smelter are strongly oversaturated with respect to numerous materials. In slags II and III, Fe is mainly bound in insoluble spinels Cu-, CuPb- or Pb-bearing arsenates, some of which are also ob- and is not released as extensively. served as secondary alteration products on the slag surfaces: bay- ldonite (PbCu3(AsO4)2(OH)2), conichalcite (CaCuAsO4(OH)), duftite 4. Discussion (PbCuAsO4(OH)) and mimetite (Pb5(AsO4)3Cl). Schultenite exhibits saturation indices close to 0 (EN 12457 leachates) or >0 (TCLP 4.1. Phase formation in slags and solid speciation of metals and leachates) in agreement with Magalhães and Silva (2003), indicat- metalloids ing its predominance at low pH values and in Cl-free environ- ments. Zinc-bearing arsenates (adamite, Zn2AsO4(OH); austinite, Mineralogical investigation of the Tsumeb slags indicates that CaZnAsO4(OH); legrandite, Zn2AsO4(OH) H2O) tend to be oversatu- two distinct groups of materials were produced in this district, rated only in leachates with high concentrations of Zn (TCLP, sam- according to the differences in the smelting technology, primary ple N2; Fig. 5, Table 7). Similarly, only leachates from the TCLP ore (or concentrate) compositions and secondary alteration pro- leaching test are oversaturated with respect to the other Cu arse- cesses. The studied slag samples do not contain clinopyroxene, nates (clinoclase, Cu3AsO4(OH)3; cornubite, Cu5(AsO4)2 2H2O; which is an indicator of relatively slow cooling of the slag melt (Et- Cu3(AsO4)2 2H2O, euchroite, Cu2AsO4(OH) 3H2O), probably due to tler et al., 2000, 2001). Nevertheless, the cooling regime is different the high concentrations of Cu and As. In contrast to the investiga- for these two distinct slag groups, indicating that the initial com- tion of secondary alteration products, the leachates are not over- positions of the melt were probably also significantly different. saturated with respect to more common Pb-controlling phases, The oldest technology used at the beginning of the 20th century such hydrocerussite (Pb3(CO3)2(OH)), cerussite (PbCO3), litharge produced a slag melt that cooled very slowly in ladles and was (PbO) or anglesite (PbSO4). Although the pH and bulk chemical highly enriched in Pb, Cu and As (slag I). The crystallisation of composition of leachates are favourable for precipitation of car- Ca–Pb feldspars indicates that the slag melt was poor in alkalis bonates, oxides and sulphates, to efficiently control the Pb concen- and enriched in Al. In addition, feldspar formation completely re- trations in solution, these phases would be formed only under moved silica from the melt and prevented the formation of residual conditions of longer-term slag/water interaction (Ettler et al., glass. Further solidification of the slag melt continued through the 2003a). Due to the higher pH, waters from the smelter area are formation of Cu-rich oxides (spinel and mcconnellite-family oversaturated with respect to metallic carbonates, such as phases). Their presence shows that the redox conditions in the slag (hydro)cerussite (Table 7), typically found as newly formed melt were locally variable, probably due to incomplete mixing of
Table 7 Saturation index of selected solubility-controlling phases as calculated by PHREEQC-2 for the leachates and the surface waters sampled in the Tsumeb smelter area. Oversaturation of the solutions with respect to the solids is indicated in bold.
Solution EN 12457 TCLP Surface Evaporation leachate leachate run-off a dama N2 (Slag III) N6 (Slag I) N2 (Slag III) N6 (Slag I) (Smelter and slag area) (Smelter area) pH 6.78 6.54 5.71 4.93 8.0 8.2 Phase Composition
Adamite Zn2AsO4(OH) 4.92 5.35 1.22 3.49 3.07 1.17
Anglesite PbSO4 2.54 3.85 0.98 0.64 2.10 2.27
Austinite CaZnAsO4(OH) 2.13 3.51 0.64 3.91 3.91 3.61
Bayldonite PbCu3(AsO4)2(OH)2 3.66 2.59 8.91 9.31 9.39 7.51
Brochantite Cu4(OH)6SO4 4.31 5.02 0.33 1.44 1.69 1.18
Cerussite PbCO3 1.21 1.24 – – 1.47 0.41
Clinoclase Cu3AsO4(OH)3 1.04 1.33 2.73 2.14 3.51 2.52
Conichalcite CaCuAsO4(OH) 4.19 2.83 5.38 3.16 7.42 7.68
Cornubite Cu5(AsO4)2(OH)4 1.10 1.77 5.65 5.01 5.83 4.07
Cu3(AsO4)2 2H2OCu3(AsO4)2 2H2O 3.96 4.82 1.23 1.65 1.39 2.69
Duftite PbCuAsO4(OH) 3.42 2.73 5.68 6.14 6.78 5.66
Euchroite Cu2AsO4(OH) 3H2O 1.04 1.42 1.95 1.89 1.34 0.57
Ferrihydrite Fe(OH)3 1.45 1.24 0.38 0.36 3.61 3.45
Gypsum CaSO4 2H2O 1.68 3.67 1.20 3.54 1.38 0.17
Hydrocerrusite Pb3(CO3)2(OH) 3.90 4.15 – – 4.65 1.95
Legrandite Zn2AsO4(OH) H2O 5.18 5.61 0.96 3.75 2.81 0.91 Litharge PbO 6.67 6.88 6.61 6.62 3.50 4.08
Mimetite Pb5(AsO4)3Cl 12.40 10.72 19.56 23.33 22.55 20.07
Pb3(AsO4)2 Pb3(AsO4)2 6.04 7.81 3.02 1.06 0.51 2.87
SbO2 SbO2 1.62 2.23 1.41 0.08 – –
Schultenite PbHAsO4 0.04 0.74 1.52 2.51 1.22 0.33 Tenorite CuO 0.83 0.73 0.04 0.57 1.35 1.13 Zincite ZnO 2.88 5.74 5.24 5.17 0.25 1.05
Symbols used: – not determined. a Calculated from analyses reported in Table 6 (data from Walmsley Environmental Consultants, 2001). V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 13
first-class cokes used for firing in the furnaces and ore concentrate hydrocerussite and litharge. Oxides and carbonates were com- (Cu1+ M Cu2+,Fe2+ M Fe3+ redox pairs). Magnetite, generally monly observed as important secondary alteration products in formed by the reaction with O2-bearing shaft gases, is lacking in slags and mattes (Ettler et al., 2003a,b; Seignez et al., 2007) and these old slags, also indicating quite reducing conditions (Biswas can precipitate from solution at pH >5. PHREEQC-2 predicts that 3+ and Davenport, 1976). The Fe present in the melt is then en- mimetite (Pb5(AsO4)3Cl) and schultenite (PbHAsO4) precipitate trapped in the delafossite–mcconnellite phases and other Cu-bear- from the obtained leachates (Table 7). Although it is suggested that ing oxides. The matrix of the slag is composed of Ca–Pb arsenates, they are important solubility-controlling phases in Pb- and As-rich which are often enriched in Cu, Zn and other metallic elements. It systems (Magalhães and Silva, 2003), the high activity of CO2 orig- does not seem to be possible that these phases would be the last inating from the atmosphere probably causes the preferred precip- step in the solidification of the slag melt, because of the reducing itation of Pb carbonates and the absence of mimetite and conditions in the shaft furnace. They probably correspond to alter- schultenite at the slag surfaces. Carbonates commonly form in ation products of arsenides or arsenites and result from weathering the vicinity of Pb-rich inclusions composed of galena and/or metal- of the slag over many decades. There is also some analytical evi- lic Pb and efficiently immobilize Pb (Ettler et al., 2003a; Seignez dence that Cu-rich oxides can also be partly altered and form hy- et al., 2007). However, the short-term leachates obtained using drated phases (indicated by low totals of EPMA; Table 5). EN 12457 and TCLP normalized leaching tests were not oversatu- The mineralogical compositions of 40-a-old and more recent rated with respect to these Pb-bearing phases. Longer-term slag/ slags (slags II and III) exhibit a typical crystallisation sequence (see water interaction resulting in higher Pb concentrations in the e.g., Ettler et al., 2001; Lottermoser, 2002; Puziewicz et al., 2007) leachate would be necessary to precipitate these phases (Ettler and indicate the predominance of silicate glass. The spinel-family et al., 2003a). oxides with compositions corresponding to gahnite–ferrochrom- It has been suggested that secondary HFO and HAlO are the ite–magnetite solid solutions are the first phases crystallizing from most important phases controlling the release and mobility of the slag melt. In contrast to the 100-a-old technology, the slag melt potentially toxic elements from slags through sorption and/or was enriched in alkalis and Ca (lime was probably used as a flux coprecipitation (Parsons et al., 2001; Ettler et al., 2003a, 2005; Pia- agent) and the Ca–Fe alumosilicates, such as olivine and melilite, tak et al., 2004). Although precipitation of these phases was pre- formed as the major slag constituents. Furthermore, the melt was, dicted by PHREEQC-2, they were not observed by SEM, as the in general, poorer in Cu (as indicated by the formation of residual secondary alteration products developed directly on the slag sur- Cu-sulphide droplets embedded in the silicate phases and the ab- faces. According to Ettler et al. (2005), the key factors influencing sence of Cu oxides) and enriched in Zn. Zinc is significantly concen- the precipitation of HFO are (i) the effective dissolution of primary trated in all the oxides, silicates and silicate glass occurring in these Fe sulphides (pyrrhotite, Fe-bearing sphalerite/wurtzite) and, to a slags. The initial concentration of Pb in the melts was probably also lesser extent, of primary silicates, oxides and glass accompanied lower than for melts produced at the beginning of the 20th century; by a release of Fe2+ into the solution and (ii) the time necessary it is concentrated only in the residual glass (solidifying at the end of for oxidation of dissolved Fe2+. However, relatively few Fe-bearing the crystallisation sequence) and in small galena droplets. sulphides were observed in the slags from Tsumeb, preventing any significant release of Fe during the weathering. With the exception 4.2. Environmental implications of Cu–Fe oxides, the spinels, which are the most important Fe car- riers, are not significantly weathered. The Fe concentrations in the The presence of secondary phases indicates that slags from Tsu- leachates obtained by EN 12457 and TCLP methods were also very meb, especially those exposed to weathering on dumps for several low, ranging from 0.012 to 0.07 mg/L, thus preventing massive pre- decades, may be important sources of potentially toxic elements cipitation of HFO. Furthermore, recent experimental investigation (metals and metalloids) that can be released into the environment. of Pb slag alteration by Seignez et al. (2007) did not revealed any The Ca–Pb arsenates forming the matrix of the oldest slags reveal HFO and suggested that carbonates only are the key Pb-controlling that the material underwent significant alteration processes phases. (Fig. 2a). The spinel phases observed in polymetallic slags from The leaching results indicated that only As may be considered to other smelting sites are generally considered to be the most be a serious problem exceeding the EU regulatory limits for haz- weathering-resistant phases (Ettler et al., 2002, 2003a; Seignez ardous waste. According to the TCLP test, the regulatory levels et al., 2007). In contrast, the Cu-bearing spinels and delafossite-like were exceeded for Cd, Pb and As for the oldest slag sample (N6). phases from the oldest Tsumeb slags are often highly weathered, In comparison with the TCLP test, the EN 12457 leaching test forming various Cu–Fe or Cu–Cr hydrated oxide compounds (Table seems to be more appropriate for quick evaluation of the hazard- 5). This alteration may be partly responsible for the small release of ous properties of the slags, because no organic complexing agents Cu into the environment during the long-term weathering of slags (such as acetate) can be expected in the environment of slag in the dumps, although the observed Cu concentrations in leach- dumps and tailing ponds. The leaching results are, however, ates and waters from the smelter area were relatively low (Fig. 5, strongly influenced by the presence of fine dust fractions (Zandi Table 6). et al., 2007) and the release of contaminants from slags will be The investigation of secondary weathering products revealed a much slower in reality. In particular, metal-bearing sulphides can- predominance of complex Cu- and Pb-arsenates locally developed not be dissolved in the short-term (24-h) regulatory leaching tests on the slag surfaces. The large variety of these species indicates lo- used in this study. For this purpose, long-term leaching with the cal differences in the chemical microenvironments on the weath- coarse-grained slag fraction (e.g., 2–5 mm, as performed by Ettler ered slag surfaces. Whereas lammerite is generally formed under et al., 2003a) could be useful for estimation of more realistic con- highly acidic conditions (pH <3), olivenite and bayldonite precipi- taminant release from the slag dumps. tate under slightly acidic to circum-neutral conditions (up to pH Seasonal variations between the dry and wet periods can lead to 6or7)(Magalhães et al., 1988; Inegbenebor et al., 1989; Magalhães the dissolution of primary slag phases and formation of secondary and Silva, 2003). These complex Cu- and Pb-bearing arsenates were efflorescence minerals, which are highly soluble, as documented at also predicted by PHREEQC-2 calculations for the slag leachates to other smelting sites (Lottermoser, 2005). As the secondary weath- be the main solubility-controlling phases for Cu, Pb and As. The ering products of the Tsumeb slags are composed of soluble arse- SEM and XRD study of weathered slag surfaces revealed that, in nates (see the high values of the solubility products in Magalhães addition to bayldonite, Pb is controlled by the precipitation of et al., 1988), some rain events during the rainy season in Namibia 14 V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 can flush out a significant amount of As from the slag dumps. In Chetty, D., Frimmel, H.E., 2000. The role of evaporates in the genesis of base Tsumeb, the annual average rainfall is 470 mm and flash floods sulphide mineralization in the Northern Platform of the Pan-African Samara Belt, Namibia: geochemical and fluid inclusion evidence from carbonate wall occasionally occur from October to March with up to 50 mm rain- rock alteration. Mineral. Depos. 35, 364–376. fall. This phenomenon is also confirmed by high concentrations of Cornelis, G., Johnson, C.A., Van Gerven, T., Vandecasteele, C., 2008. Leaching As in run-off from the smelter area and slag dumps and in water mechanisms of oxyanionic metalloid and metal species in alkaline solid wastes: a review. Appl. Geochem. 23, 955–976. from evaporation dams in the Tsumeb smelter, also indicating that Costagliola, P., Benvenuti, M., Chiarantini, L., Bianchi, S., Di Benedetti, F., Paolieri, M., arsenate phases can be more soluble under slightly alkaline condi- Rossato, L., 2008. Impact of ancient metal smelting on arsenic pollution in the tions (pH 8; Table 6). To better understand the cycling of toxic Pecora River Valley, Southern Tuscany, Italy. Appl. Geochem. 23, 1241–1259. CˇSN 72 0100, 1984. Basic Methods of Analysis of Silicates. Czech Technical elements in the vicinity of the dumps during the rainy season, it Standards, Czech Standard Institute, Prague, Czech Republic (in Czech). would be necessary to perform in situ long-term monitoring stud- EN 12457, 2002. Characterisation of Waste – Leaching – Compliance Test for ies in the Tsumeb area using the groundwater and surface water Leaching of Granular Waste Materials and Sludges. Part 2. One Stage Batch Test as a Liquid to Solid Ratio of 10 for Materials with Particle Size Below 4 mm sampling methodologies reported from other smelting sites (Lot- (Without or With Size Reduction). Czech Standard Institute, Prague. termoser, 2002; Parsons et al., 2001; Navarro et al., 2008). It is Ettler, V., 2002. Etude du potentiel polluant de rejets anciens et actuels de la important to point out that finely ground, reprocessed slags depos- métallurgie du plomb dans le district de Prˇíbram (République Tchèque). ited in tailing ponds – if exposed to weathering – can be expected Document du BRGM 301, ISBN 2-7159-0924-01 Orléans, France. Ettler, V., Johan, Z., 2003. Mineralogy of metallic phases in sulphide mattes from to release large amounts of potentially toxic elements due to their primary lead smelting. C. R. Geosci. 335, 1005–1020. higher reactive surface and can be considered to be a serious prob- Ettler, V., Johan, Z., Touray, J.C., Jelínek, E., 2000. Zinc partitioning between glass and lem for the environment in the Tsumeb area. silicate phases in historical and modern lead–zinc metallurgical slags from the Prˇíbram district, Czech Republic. C. R. Acad. Sci. Paris 331, 245–250. Ettler, V., Legendre, O., Bodénan, F., Touray, J.C., 2001. Primary phases and natural weathering of old lead–zinc pyrometallurgical slag from Prˇíbram, Czech 5. Conclusions Republic. Can. Mineral. 39, 873–888. Ettler, V., Mihaljevicˇ, M., Piantone, P., Touray, J.C., 2002. Leaching of polished The smelting slags from the Tsumeb area resulting from histor- sections: an integrated approach for studying the liberation of heavy metals from lead–zinc metallurgical slags. Bull. Soc. Geol. France 173, 161–169. ically different technologies are chemically and mineralogically Ettler, V., Piantone, P., Touray, J.C., 2003a. Mineralogical control on inorganic complex waste materials. Up to 100-a-old slags are composed of contaminant mobility in leachate from lead–zinc metallurgical slag: anorthite and Pb-feldspars, Cu-spinels and other Cu–Cr–Fe oxides experimental approach and long-term assessment. Mineral. Mag. 67, 1269–1283. and matrix Ca–Pb arsenates. More recent granulated slags are Ettler, V., Johan, Z., Hradil, D., 2003b. Natural alteration products of sulphide mattes from primary lead smelting. C. R. Geosci. 335, 1013–1020. mostly composed of glass, high-temperature Ca–Fe alumosilicates, Ettler, V., Komárková, M., Jehlicˇka, J., Coufal, P., Hradil, D., Machovicˇ, V., Delorme, F., spinel-family oxides and sulphide/metallic inclusions. Numerous 2004. Leaching of lead metallurgical slag in citric solutions – implications for natural weathering features were observed at the slag surface, disposal and weathering in soil environments. Chemosphere 57, 567–577. Ettler, V., Jehlicˇka, J., Mašek, V., Hruška, J., 2005. The leaching behaviour of lead indicating that these materials underwent alteration during their metallurgical slag in high-molecular-weight (HMW) organic solutions. Mineral. long-term disposal on the dumps. The secondary phases identified Mag. 69, 737–747. ˇ by SEM and XRD were mainly complex Cu–Pb–(Ca) arsenates. Ettler, V., Cervinka, R., Johan, Z., in press. Mineralogy of medieval slags from Pb and Ag smelting (Bohutín, Prˇíbram district, Czech Republic): towards estimation of According to batch leaching experiments coupled with speciation historical smelting conditions. Archaeometry. modelling, these phases were also suggested to be the main solu- Frimmel, H.E., Deane, J.G., Chadwick, P.J., 1996. Pan-African tectonism and the bility controls on the mobility of metals and metalloids in slag genesis of base metal sulfide deposits in the foreland of the Damara Orogen, Namibia. In: Sangster, D.F. (Ed.), Carbonate-Hosted Lead–Zinc Deposits. Society leachates. The presence of these highly soluble phases might be of Economic Geologists, Littleton, pp. 204–217 (Spec. Publ. No. 4). responsible for the significant release of related potentially toxic Ghosh, A., Mukiibi, M., Ela, W., 2004. TCLP underestimates leaching of arsenic from elements (As, Cu, Pb) from slag dumps, through their rapid dissolu- solid residuals under landfill conditions. Environ. Sci. Technol. 38, 4677–4682. Hauptmann, A., 2007. The Archaeometallurgy of Copper. Springer, Berlin, tion during thunderstorm events occurring in the Tsumeb area be- Heidelberg, New York. tween October and March. ICDD, 2003. JCPDS PDF-2 Database. ICDD, Newton Square, PA, USA. Inegbenebor, A.I., Thomas, J.H., Williams, P.A., 1989. The chemical stability of mimetite and distribution coefficient for pyromorphite–mimetite solid- Acknowledgements solutions. Mineral. Mag. 53, 363–371. Johnson, W.M., Maxwell, J.A., 1981. Rock and Mineral Analysis. Wiley, New York, This study was supported by Project of the Development Coop- USA. Krˇíbek, B., Kamona, F. (Eds.), 2005. Assessment of the mining and processing of ores eration of the Czech Republic No. RP/20/2004 (‘‘Assessment of min- on the environment in mining districts of Namibia. Final report of the project of ing and processing of ores on the environment in the mining the development assistance programme of the Czech Republic to the Republic districts of Namibia”) and was partly financed by a Czech Science of Namibia for the year 2004 – Tsumeb area. MS Czech Geol. Surv., Record Office ˇ – File Report No. 2/2005, Prague, Czech Republic. Foundation project (GACR 205/06/0298). The institutional funding Lottermoser, B.G., 2002. Mobilization of heavy metals from historical smelting slag was provided by Ministry of Education, Youth and Sports of the dumps, north Queensland, Australia. Mineral. Mag. 66, 475–490. Czech Republic (MSM 0021620855). Dr. Ladislav Strnad (Charles Lottermoser, B.G., 2005. Evaporative mineral precipitates from a historical smelting slag dump, Río Tinto, Spain. Neues Jahrb. Mineral. Abh. 181, 183–190. University) assisted with the ICP-MS measurements of trace ele- Magalhães, M.C.F., Silva, M.C.M., 2003. Stability of lead(II) arsenates. Monatsh. ments in digests, Radek Procházka (Charles University), Annick Chem. 134, 735–743. Genty (Université d’Orléans) with SEM/EDS measurement and Magalhães, M.C.F., Pedrosa de Jesus, J.D., Williams, P.A., 1988. The chemistry of formation of some secondary arsenate minerals of Cu(II), Zn(II) and Pb(II). Anna Langrová (Geological Institute, Czech Academy of Science) Mineral. Mag. 52, 679–690. with EPMA measurements. Dr. Eva Gregorová (Institute of Chemi- Manasse, A., Mellini, M., 2002a. Chemical and textural characterization of medieval cal Technology, Czech Republic) is thanked for the LALLS granulo- slags from the Massa Marittima smelting sites (Tuscany, Italy). J. Cult. Herit. 3, 187–198. metric analyses. The thorough reviews of Dr. Nadine M. Piatak (US Manasse, A., Mellini, M., 2002b. Archaeometallurgic slags from Kutná Hora. Neues Geological Survey) and two anonymous referees helped to improve Jahrb. Mineral. Mh. 2002 (8), 369–384. the final version of the manuscript. Manasse, A., Mellini, M., Viti, C., 2001. The copper slags of the Capattoli Valley, Campiglia Marittima, Italy. Eur. J. Mineral. 13, 946–960. Miller, R. McG., 1983. The Pan-African Damara orogen of South West Africa/ References Namibia. In: Miller, R. McG. (Ed.), Evolution of the Damara Orogen of South West Africa/Namibia. Geol. Soc. South Africa, Pietermaritzburg, pp. 431–515 Benna, P., Tribaudino, M., Bruno, E., 1996. The structure of ordered and disordered (Spec. Publ. 11).
Pb feldspar (PbAl2Si2O8). Am. Mineral. 81, 1337–1343. Navarro, A., Cardellach, E., Mendoza, J.L., Corbella, M., Domènech, L.M., 2008. Metal Biswas, A.K., Davenport, W.G., 1976. Extractive Metallurgy of Copper. Pergamon mobilization from base-metal smelting slag dumps in Sierra Almagrera Press, Oxford, UK. (Almería, Spain). Appl. Geochem. 23, 895–913. V. Ettler et al. / Applied Geochemistry 24 (2009) 1–15 15
Ongopolo Mining and Processing Limited, 2001. Site Assessment at the Tsumeb Šulcek, Z., Povondra, P., 1989. Methods of Decomposition in Inorganic Analysis. CRC Smelter. Phase 1. MS Report No. JW49/01/7818. Jones and Wagener Consulting Press, Boca Raton, FL, USA. Civil Engineers, Johannesburg, South Africa. Tsumeb Corporation Ltd, 1987. Technical Data Sheets. Tsumeb Smelter, Namibia. Parkhurst, D.L., Appelo, C.A.J., 1999. User’s guide to PHREEQC (version 2) – a USEPA, 1994. Toxicity Characteristic Leaching Procedure (TCLP), Method 1311, Solid computer program for speciation, batch-reaction, one-dimensional transport Waste Characterization Manual SW-848, Test Methods for Evaluating Solid and inverse geochemical calculations. US Geol. Surv. Report 99-4259. Waste, Physical/Chemical Methods, Athens, Georgia, USA