Miner Deposita (2013) 48:137–153 DOI 10.1007/s00126-012-0424-5

ARTICLE

The geology and geochemistry of the Lumwana Cu (± Co ± U) deposits, NW Zambia Robin Bernau & Stephen Roberts & Mike Richards & Bruce Nisbet & Adrian Boyce & James Nowecki

Received: 13 December 2010 / Accepted: 11 May 2012 / Published online: 6 June 2012 # Springer-Verlag 2012

Abstract The Lumwana Cu (± Co ± U) deposits of NW Zambia are large, tabular, disseminated ore bodies, hosted within the Mwombezhi Dome of the Lufilian Arc. The host rocks to the Lumwana deposits are two mineralogically similar but texturally distinct gneisses, a granitic to pegmatitic gneiss and a banded to augen gneiss which both comprise quartz–feldspar ± biotite ± muscovite ± haematite ± amphibole and intervening quartz–feldspar ± biotite schist. The sulphide ore horizons are typically developed within a biotite–muscovite–quartz–kyanite schist, although mineralization locally occurs within internal gneiss units. Contacts between the ore and host rocks are transitional and characterized by a loss of feldspar. Kinematic indicators, such as

Editorial handling: H. Frimmel Bruce Nisbet is deceased. R. Bernau : S. Roberts (*) : J. Nowecki School of Ocean and Earth Science, National Oceanography Centre, University of Southampton, Southampton SO14 3ZH, UK e-mail: [email protected] M. Richards : B. Nisbet Equinox Minerals PLC, Ground Floor, Scott House, 46-50 Kings Park Road, West Perth, WA 6005, Australia A. Boyce SUERC, UK Scottish Enterprise Technology Park, Rankine Avenue, East Kilbride, Glasgow G75 OQF, UK Present Address: M. Richards Barrick (Australia Pacific) Limited, 2 Mill St, Perth 6000, Australia

S-C fabrics and pressure shadows on porphyroblasts, suggest a top to the north shear sense. The sulphides are deformed by a strong shear fabric, enclosed within kyanite or concentrated into low strain zones and pressure shadows around kyanite porphyroblasts. This suggests that the copper mineralization was introduced either syn- or pre-peak metamorphism. In addition to Cu and Co, the ores are also characterized by enrichments in U, V, Ni, Ba and S and small, discrete zones of uranium mineralization, occur adjacent to the hanging wall and footwall of the copper ore bodies or in the immediate footwall to the copper mineralization. Unlike typical Copperbelt mineralization, unmineralized units show very low background copper values. Whole rock geochemical analyses of the interlayered schist and ore schist, compared to the gneiss, show depletions in Ca, Na and Sr and enrichments in Mg and K, consistent with replacement of feldspar by biotite. The mineral chemistry of muscovite, biotite and chlorite reflect changes in the bulk rock chemistry and show consistent increases in XMg as the schists develop. δ34S for copper sulphides range from +2.3‰ to +18.5‰, with pyrite typically restricted to values between +3.9‰ and +6.2‰. These values are atypical of sulphides precipitated by bacteriogenic sulphate reduction. δ34S data for Chimiwungo (Cu + Co) show a broader range and increased δ34S values compared to the Malundwe (Cu) mineralization. The Lumwana deposits show many characteristics which distinguish them from classical Copperbelt mineralization and which suggests that they are formed by metasomatic alteration, mineralization and shearing of pre-Katangan basement. Although this style of mineralization is reported elsewhere in the Copperbelt, sometimes associated with the more widely reported stratiform ores of the Lower Roan, none of the previously reported occurrences have so far developed the tonnages of ore reported at Lumwana.

138

Introduction The Neoproterozoic, Central African Copperbelt, which straddles the border between Zambia and the Democratic Republic of Congo, is host to some of the Earth’s largest known concentrations of copper and cobalt mineralization. The Lumwana deposits are located 220 km northwest of the Copperbelt and 65 km from the provincial capital of Solwezi, in the North Western Province of Zambia and represent one of the most significant copper-resources outside of the Zambian Copperbelt (Fig. 1). Exploration activity in the Lumwana region began with reconnaissance mapping in the 1920–1930s, which outlined the regional geology of the area and discovered the Malundwe copper clearing and adjacent small outcrop of low-grade copper mineralization within the Lumwana East River. However, the mineralization was considered inconsequential. Stream and soil sediment sampling in the 1950–1960s by Roan Selection Trust identified three major copper anomalies: Malundwe, Chimiwungo and Lubwe. Two of these geochemical anomalies were associated with prominent copper clearings (areas where normal savannah woodland is replaced by sparse coverage of metal tolerant plants), and

Fig. 1 a Regional geological setting of the Zambian Copperbelt and Domes Region. Modified after Nyambe and Kawamya (2005). b Location map of major copper and cobalt deposits within Northern Zambia

Miner Deposita (2013) 48:137–153

subsequent exploration drilling over the next four decades by the Roan Selection Trust, Agip-Cogema, Phelps-Dodge and Equinox identified three copper deposits Malundwe, Chimiwungo and Lubwe; of which the first two have JORC/NI 43-101 compliant Mineral Resources. McGregor (1965) and Benham et al. (1976) described the geology of the deposits and ascribed them a syngenetic origin. However Benham et al. (1976) recognised that if typical Copperbelt rock types were subject to the prevailing metamorphic conditions of the Mwobezhi Dome, the resultant mineral assemblage would probably not be the mineral assemblage observed at Malundwe and Chimwungo. The published mineral resource for Malundwe shows a depleted mineral resource of 18.1 Mt at 1.0 % Cu in measured, 182.4 Mt at 0.71 % Cu in indicated and 57.5 Mt at 0.63 % Cu inferred. At Chimiwungo, the mineral resource consists of 82.5 Mt at 0.76 % Cu and 0.03 % Co in measured, 168.7 Mt at 0.58 % Cu and 0.0130 % Co in indicated and 553.9 Mt at 0.63 % Cu and 0.005 % Co in inferred (Equinox Minerals Limited TSX Technical Report, 2008 and 2011) Despite the delineation of this significant resource, the origin of the mineralization and host rocks at Lumwana is ambiguous, with transitional contacts between barren

Miner Deposita (2013) 48:137–153

quartz–feldspar ± biotite gneiss and a Cu ± Co mineralized quartz–biotite–muscovite–kyanite–sulphide–ore schist. The geological characteristics of the deposits certainly do not conform to the typical copper ores located in the Lower Roan stratigraphy of the Copperbelt. Furthermore, the deposits at Lumwana have yet to be subject to the same level of scientific scrutiny and as such are poorly understood. Typically, the Cu-Co mineralization of the Zambian Copperbelt is restricted to clastic sequences of the Katangan Supergroup, metamorphosed to greenschist and locally amphibolite facies (Mendelsohn 1961; Annels 1989; Selley et al. 2005). However, copper mineralization within the preKatangan basement of the Copperbelt is reported as vein fill along faults and disseminated within granitoid and sedimentary rocks (Pienaar 1961). Furthermore, occurrences of copper within the pre-Katangan basement to the Copperbelt deposits, includes the Samba Deposit with an estimated 50 million tonnes at 0.5 wt% Cu (Wakefield 1978), the Nchanga Red Granite with an estimated 1.5 million tonnes at 1.5 wt% Cu and the Muva phyllites south of the Nkana Basin with a 3.6-m-wide zone at 3.6 wt% Cu (Cailteux et al. 2005). The Lumwana deposits are located within high-grade metamorphic rocks and may represent mineralized and metamorphosed Lower Roan clastic sequences (Benham et al. 1976) or mineralized and metamorphosed pre-Katangan basement units. Here we present new geological, petrographic, whole rock geochemistry, electron microprobe and stable isotope analyses (δ34S) of ore and host rock assemblages of the Lumwana deposits, in an attempt to better understand the origin of the Cu-Co mineralization. The development of an increased understanding of the mechanism and timing of ore formation and associated wallrock alteration at Lumwana deposits may have significant implications for the exploration potential of preKatangan basement throughout the Zambian Copperbelt.

Regional setting The Lumwana Cu (±Co ± U) deposits are hosted within the Mwombezhi Dome, one of several inliers of Early to MidProterozoic age rocks within the Lufilian Arc, which is a Pan-African fold-thrust belt approximately 900 km in length and which hosts the Central African Copperbelt. The Mwombezhi Dome is located east of the Kabompo Dome and west of Solwezi and Luswishi domes (Fig. 1), which typically comprise granitic gneisses, migmatites and schists, unconformably overlain at their margins by rocks of the Neoproterozoic Katangan Supergroup (Cosi et al. 1992; Fig. 1). Mineralization within the Domes Region also includes the Kalumbila Co-Ni-Cu deposit of the Kabompo Dome (Steven and Armstrong 2003) and the vein-hosted

139

Kansanshi Cu deposit to the east of the Solwezi Dome (Broughton et al. 2002). The Mwombezhi Dome (Fig. 2) is interpreted, along with the other inliers of the region, as an antiformal thrust stack that developed above mid to lower crustal ramps, which are separated from the overlying Katangan Supergroup sediments by a major decollement (Daly et al. 1984; Cosi et al. 1992; John et al. 2004). Whiteschist assemblages in the Mwombezhi Dome indicate peak metamorphism of 750°C± 25°C and 13±1 kb, corresponding to burial depths of approximately 50 km (Cosi et al. 1992; John et al. 2004). This peak metamorphic assemblage gives U-Pb monazite ages of 524±3Ma to 532±2Ma (John et al. 2004). The age range and rock types of the Domes Region show a strong similarity with the basement rocks of the Kafue Anticline. Collectively they most likely belong to the Lufubu Metamorphic Complex, which has been interpreted as a Paleoproterozoic magmatic arc terrane (Rainaud et al. 2005). In Zambia and the Democratic Republic of Congo, the Lufubu Metamorphic Complex yields U-Pb SHRIMP zircon ages of between 1980 and 1874 Ma (John et al. 2004; Rainaud et al. 2005) and the Mufulira Pink Granite and the Chambishi Granite yield ages of 1994±7 Ma and 1983±8 Ma, respectively (Rainaud et al. 2005). U-Pb age determinations of detrital zircon grains on the flanks on the Kabompo Dome to the west of Mwombezhi give ages of 2004 to 1884 Ma, suggesting the source of detritus was the underlying granite gneiss of the Kabompo Dome (Steven and Armstrong 2003). The Lufilian Arc, of which the domes region is an integral part, most likely formed during the collision of the Angola-Kalahari and Congo-Tanzania Plates, with accompanying NE-directed thrusting at 560–550 Ma, during the assembly of the supercontinent Gondwana between 750 and 500 Ma (Porada and Berhorst 2000; Frimmel et al. 2011). The extent of rifting between the Congo and Kalahari cratons is debatable; however, the most recent studies suggest rifting started after 880 Ma and separated the CongoTanzania plate from the Angola-Kalahari plate, leading to the formation of an ocean basin and the development of a passive continental margin on the southern side of the Congo Craton (Porada and Berhorst 2000; John et al. 2003). During subsequent compression and convergence of the Kalahari and Congo cratons, the ocean basin closed subducting oceanic lithosphere and forming eclogites at 600 Ma (John et al. 2003). The final stage of convergence resulted in a continent–continent collision with the Kalahari Craton overriding the passive continental margin and colliding with the Congo Craton at ∼530 Ma. This stage of the Pan-African evolution is interpreted to be responsible for the formation of the talc–kyanite assemblages located in the Domes Region and Zambezi Belt. The final collision was followed by rapid erosion, tectonic uplift and cooling (John et al. 2004).

140

Miner Deposita (2013) 48:137–153

Fig. 2 a Simplified geology of the Mwombezhi Dome, with the location of the Chimiwungo and Malundwe deposits. Geology based on aeromagnetic interpretation and field studies. b Aeromagnetic map, showing the first vertical derivative, of the Mwombezhi Dome. Bright responses indicate the presence of highly magnetized Katangan rocks imbricated within the Dome

Geological setting of the mineralization The internal geology of the Mwombezhi Dome is complex, comprising granite, granite gneiss, schist and quartzite units with associated thrusts and folds (Fig. 2a). Limited field exposures and drill core preserve a strong, flat-lying planar fabric (S1) which has been overprinted by a later parallel (S2) fabric, defined by muscovite. In the vicinity of the mineralization, a strong N-S orientated stretching lineation (L) plunges gently to the south at 6°. Kinematic indicators on quartz and kyanite augen enclosed within the S1 fabric consistently indicate a top-to-the-north sense of displacement, parallel to the stretching lineation. Post Lufilian relaxation and/or younger extensional tectonics (Pavoni 1992) developed a series of NW-SE to E-W and N-S to NNE–SSW normal faults. The biotite gneiss, hornblende gneiss, granite gneiss, migmatites and schists of the dome give U-Pb ages of

1400–1200 Ma (Cosi et al. 1992), 207Pb–206Pb ages of basement zircon grains between 1700–1500 Ma and high initial 87Sr/86Sr ratios of 0.715, suggesting this was not the first event in the history of the pre-Katangan basement (Cosi et al. 1992). Recent studies by Rainaud et al. (2005) suggest that the basement rocks of the Mwombezhi Dome predominantly correlate with the 2050–1850 Ma Lufubu Metamorphic Complex and that the ages quoted by Cosi et al. (1992) are too young. These units are intruded by granites which have not been dated, but are likely to be synchronous with either the 877±11 Ma Nchanga Red Granite located in the Copperbelt, or the 559±18 and 566±5 Ma Hook Granite Complex to the south of Mwombezhi Dome, located on the northern contact of the Mwombezhi Shear Zone (Armstrong et al. 2005; Hanson et al. 1993). The rocks of the Mwombezhi Dome are separated from the overlying Katangan stratigraphy by a major decollement

Miner Deposita (2013) 48:137–153

marked by a quartzite and muscovite–quartz–talc–kyanite schist or whiteschist with lower strain zones of quartzite (Cosi et al. 1992; John et al. 2004). Whiteshchists are typically found in Alpine-type orogens, with the stability field of the talc–kyanite assemblage restricted to temperatures between ca. 600 and 800°C and pressures which range from 6 kb to ultra high pressure metamorphism (∼15 kb; Chopin 1984; Massonne and Schreyer 1989). The whiteschist and quartzite are overlain by carbonate-rich units, which correlate with the units above the Roan in the Zambian Copperbelt and Democratic Republic of Congo. The Chimiwungo and Malundwe deposits are large tabular low-grade ore bodies, with strike extents of 4 and 6 km, respectively, which dip gently to the south and west (Fig. 3). Hanging wall, ore schist and footwall units at the two deposits are broadly comparable (Fig. 3).

Chimiwungo The Chimiwungo ore deposit comprises three lenticular, sulphide mineralized ore schist units separated by unmineralized to weakly mineralized laterally discontinuous units of quartz–biotite–feldspar banded gneiss or amphibolite (Fig. 3). Ore schist horizons continue to the east and west of the current resource area but are narrower and lower grade. The ore body is open to the south. The deposit is intersected by a series of later, E-W to NW-SE normal Fig. 3 Simplified cross sections through the Malundwe and Chimiwungo deposits which illustrate the intercalation of ore and gneiss units, and at Chimiwungu, the displacement of ore horizons by later normal faults

141

faults, of which the most important is the Chimiwungo South Fault (Fig. 3). The hanging wall of the Chimiwungo ore is dominated by two mineralogically similar but texturally distinct rock types, a granitic to pegmatitic gneiss and a banded to augen gneiss; both comprise quartz–feldspar ± biotite ± haematite. The granitic gneiss is a quartz–feldspar gneiss with minor muscovite and or biotite and varying proportions of euhedral to anhedral magnetite and haematite crystals (Fig. 4a). These rocks typically show the development of a weak fabric delineated by the alignment of biotite. In contrast, the remainder of the hanging wall comprises a quartz–biotite–feldspar gneiss, with garnet, amphibole and minor muscovite, a higher biotite content, more pronounced millimetre- to centimetre-scale compositional banding and local quartz and feldspar augen (Fig. 4b). Contacts between the banded gneiss and the granite gneiss are typically sharp, although they can be ambiguous if the granite gneiss has been altered at the contact. Granite gneiss is generally restricted to the hanging wall of the ore schist whereas the banded gneiss is found in both hanging wall and footwall locations. Beneath the lower ore schist, the Chimiwungo footwall consists primarily of quartz–biotite–feldspar banded gneiss with minor amphibolite and schist units. Ore schist units There are three main laterally continuous ore schist units at Chimiwungo (Fig. 3). The upper ore schist has an average

142

Miner Deposita (2013) 48:137–153

Fig. 4 Representative core pieces from Chimiwungo, which outline the relationships between gneiss schist and ore horizons. a Hanging wall gneiss. b Hanging wall gneiss with onset of incipient fabric defined by biotite. c Kyanite schist with sulphides aligned and deformed within the prevailing fabric. d Kyanite bearing ore schist, with well developed kyanite porphyroblasts which locally enclose sulphides. e Ore schist, with quartz–biotite and late chlorite–sericite

thickness of 10 m and comprises a muscovite–biotite– quartz–kyanite schist with chalcopyrite and pyrite ± pyrrhotite mineralization (Fig. 4C). The central ore schist has an average thickness of 60 m and is mineralogically similar to the upper ore schist, although biotite (± muscovite)–quartz– kyanite schist with chalcopyrite and bornite zones occur towards the base of the unit. Locally the central ore schist contains small lenses of quartz–mica and quartz–feldspar gneiss. These are typically not continuous between drill holes and interpreted as remnants of pre-shearing basement preserved in lower strain domains. The lower ore schist has an average width of 12 m and is dominantly a biotite (±muscovite)–quartz–kyanite schist. The lower ore schist is dominated by a bornite zone (±chalcopyrite) although zones of chalcopyrite–pyrite are noted. In all of the ore schist horizons, sulphides are aligned with the dominant S1 fabric or occur as inclusions in kyanite, in pressure shadows adjacent to kyanite porphyroblasts and in the quartz–mica bands (Fig. 4D). The upper and central ore schist also contain discrete zones of cobaltiferous sulphides including: carrollite (CuCo2S4), bravoite ((Fe,Ni,Co)S2) seigenite ((Co,Ni)4S3) and cobaltiferous pentlandite.

Malundwe The gross geological characteristics of the Malundwe deposit are comparable to Chimiwungo; however, there are some notable differences. The Malundwe deposit consists of hanging wall granite gneiss with schist interlayers overlying sulphide mineralized kyanite schist (ore schist; Figs. 3 and 5), with a footwall sequence predominantly composed of unmineralized kyanite schists in addition to muscovite–quartz–talc– kyanite schist with quartzite lenses, which overlie a marble, amphibolite and metapelite sequence, interpreted as a faultbound sliver of highly deformed Katangan rocks (e.g. Cosi et al. 1992). The Malundwe ore schist generally lacks the continuous internal gneiss units observed at Chimiwungo, with only minor internal gneiss units of limited lateral extent (Fig. 3). Furthermore, the cobalt tenor is significantly lower than at Chimiwungo with only rare zones of cobaltiferous sulphides (carrolite–bravoite–seigenite) observed.

Internal gneiss units The main units of ore schist units at Chimiwungo are separated by 16 to 60m thick domains of quartz–feldspar to quartz–feldspar–biotite gneiss. These typically lack welldeveloped schistose fabrics and sulphides are minor or absent.

Fig. 5 Image of the pit face at Malundwe, which shows the contact between the hanging wall gneiss and ore schist

Miner Deposita (2013) 48:137–153

The contact between the hanging wall gneiss and ore schist is particularly well preserved at Malundwe, with a gradual transition from leucocratic, feldspar–quartz–muscovite–biotite banded gneiss to a more biotite-rich gneiss which correlates with a reduction in the mode of feldspar, an increased mode of biotite and increased schistosity development (Fig. 5). The ore schist at Malundwe is approximately 15 m thick and comprises a biotite–muscovite–quartz–bornite–chalcopyrite–kyanite schist with minor quartz-mica gneiss units. A distinctively textured, poikiloblastic kyanite biotite– quartz–chlorite schist, termed the mottled schist, forms the immediate footwall to the Malundwe ore schist. Kyanite porphyroblasts within the mottled schist vary in size from 1 to 5 cm and contain inclusions of biotite and quartz. Chlorite is located both within the fabric as well as partially replacing the kyanite porphyroblasts. Within the mottled schist, there is a distinctive, highly fissile muscovite–quartz–epidote schist locally called the “epidote schist” (Fig. 3). The epidote schist is only located below the Malundwe ore schist but can vary in its absolute position relative to the ore schist footwall contact. Below the epidote schist the kyanite porphyroblasts gradually decrease in size and muscovite content increases until a muscovite– quartz–talc–kyanite schist±quartzite is encountered. Below Fig. 6 Photomicrographs which illustrate hanging wall gneiss, kyanite shist and ore observed at Lumwana. a Hanging wall gneiss with quartz and abundant microcline, with local alteration of feldspar to sericite (XPL). b Kyanite porphyroblast within fabric defined by biotite wrapping around porphyroblast. c Sulphides enclosed in kyanite porphyroblasts and within the cleavage planes of phyllosilicates (XPL). d Reflected light photomicrograph of chalcopyrite and bornite intergrowths commonly observed at Lumwana. e Late fan-like chlorite (XPL). f Opaque sulphides within kyanite porphyroblasts (XPL)

143

this horizon a whiteschist marks the location of a major thrust zone between the Cu-Co mineralized host rocks of the Dome and a succession of highly sheared and altered Katangan rocks (Cosi et al. 1992; John et al. 2004).

Petrology of host rocks and mineralization Hanging wall units The units of granite gneiss at both Chimiwungo and Malundwe, predominantly comprise quartz (∼45 vol%), feldspar (∼35 vol%) and biotite (∼5 vol%) with a planar fabric defined by quartz bands, which vary in thickness between 100–200 μm and 3–4 mm, separated by interlocking subhedral feldspar (microcline and plagioclase; Fig. (6a)). Quartz occurs as anhedral aggregates interlocking with feldspar and as subhedral recrystallized bands which exhibit a weakly sigmoidal texture. The feldspar is predominantly microcline with cross hatched twinning and moderate sericitization (Fig. 6a). Muscovite (∼4 vol%) forms anhedral laths which crosscut both feldspar and quartz; accessory minerals include magnetite, haematite and epidote.

144

The unmineralized schist units in the hanging wall of Chimiwungo and Malundwe show a pronounced reduction in feldspar, compared to the gneiss units, and comprise quartz (∼40 vol%), muscovite (∼40 vol%) and biotite (∼15 vol%) which locally exhibit C/S shear fabrics. The quartz forms large elongate aggregates up to 4 mm long and 2 mm wide in low strain zones separated by muscovite and lesser amounts of biotite. Kyanite is not observed in the hanging wall at Malundwe, although occurs as augen or porphyroblasts retrogressed to biotite/chlorite in low strain zones in the hanging wall schist to gneiss units at Chimiwungo. These units also tend to contain rutile, which forms in concentrated bands within muscovite-dominated areas and titanite which ranges in morphology from diamond to lozenge shaped grains, up to 80 μm in diameter. There are no copper sulphides in the hanging wall schist and gneiss units at Chimiwungo and Malundwe. However, minor pyrite mineralization occurs as occasional, narrow mica–quartz±kyanite schist lenses, locally named hanging wall ore schist which has poor lateral continuity. Secondary malachite and azurite are present along fracture surfaces at both Malundwe and Chimiwungo close to the contact with the ore schist. Ore schist units Typically the Cu-Co mineralized zones at Chimiwungo and Malundwe comprise schists which show a high mode of muscovite–biotite and kyanite, separated by quartz bands. The muscovite is often aligned in discrete bands which overprint the earlier biotite fabric but which also anastomose around quartz and kyanite augen (Fig. 6b). The higher-grade ore horizons typically comprise bornite and chalcopyrite and commonly show pairs of copper sulphides: pyrite–chalcopyrite, chalcopyrite–carrollite (Chimiwungo only) and chalcopyrite–bornite. Chalcopyrite shows either intergrowths with or replaces bornite (Fig. 6d). Other sulphide phases only observed at Chimiwungo include cubanite and pyrrhotite. Secondary minerals include digenite after bornite and chalcopyrite, which in turn altered to covellite. Digenite is also located as individual grains which are in many places crosscut by a lattice of haematite. Chrysocolla, malachite and azurite occur within the ore schist at both deposits. The sulphides are coarse-grained, commonly elongated within a planar fabric and readily observed as inclusions within kyanite (Fig. 6c, f) and as pressure shadows around kyanite and quartz augen, indicating that they predate later deformation and peak metamorphism. Uranium mineralization occurs within the ore schist at Chimiwungo and Malundwe. Uraninite (UO2) and disseminated brannerite (U,Ca,Ce)(Ti,Fe)2O6 occur as disseminated aggregates, aligned parallel to the main ore schist fabric but confined to relatively small, discrete zones, typically

Miner Deposita (2013) 48:137–153

adjacent to the hanging wall and footwall contact and the immediate footwall. Within the weathered zone, local remobilized secondary uranium mineralization is in places observed. Quartz veins cross cut the ore schist and internal gneiss units and locally host massive grains of the copper sulphide assemblage, carbonate veins are also evident. A more detailed account of the uranium mineralization and its relationship to the prevailing fabrics is to be found in Cosi et al. (1992). Within the ore schist kyanite and garnet locally break down to chlorite and quartz. Also present are two distinct types of chlorite, one with decussate texture and berlin blue interference colours, the other forming fine laths that are commonly located where sulphides are crosscut by muscovite (Fig. 6e). Footwall units The Chimiwungo footwall consists of augen to banded gneiss. However, the footwall at Malundwe contains amphibolite, mottled kyanite schist and an epidote schist which overlies a quartzite and talc–kyanite, whiteschist. In the mottled schist, kyanite exhibits two texturally distinct morphologies, occurring either as augen (∼3 mm diameter) in a fabric dominated by quartz and biotite (Fig. 6b) or as kyanite porphyroblasts which poikilitically enclose quartz and biotite inclusions and which locally exhibit simple twinning. Kyanite is in many places crosscut by later biotite, muscovite and chlorite (Fig. 6f). Scapolite is a common phase, replacing the kyanite in the mottled schist units at Malundwe.

Geochemistry methods XRF A total of 110 whole rock multi-element XRF analyses were completed on core samples from the Chimiwungo and Malundwe deposit using a Philips Magix Pro WD-XRF at the School of Ocean and Earth Science, National Oceanography Centre. Accuracy and precision were evaluated through analysis of international standards (including JA-1) and repeat analyses of standards and unknowns. These data show detection limits for major elements of 0.005 % and trace elements typically to 3ppm. Accuracy is well within acceptable limits, typically replicating major values to between 3 % and 5 % and trace element concentrations to better than 10 %. Repeat analyses of a Montserrat andesite sample, measured ten times, indicates that analytical results compare favourably and within 5% relative standard deviation for both major and minor elements.

Miner Deposita (2013) 48:137–153

Electron microprobe

145

EQ-CHI-062

Wt % 0.50

1.00

1.50

2.00

2.50

3.00

3.50 4.00

100.00

Biotite, muscovite and chlorite were analysed on a Cameca SX50 WDS electron microprobe at the Natural History Museum, London using polished thin sections. Analyses were conducted at 15 kV and 20 nA and counting times ranged from 10 to 50 s for spot analysis. Using pure metal and mineral standards for calibration, a detection limit of ∼0.01 wt% was achieved for most elements. Sulphur isotope analyses δ34S analyses were completed on representative sulphide samples from Chimiwungo and Malundwe. Conventional sulphur isotope analyses of hand-picked sulphide separates followed standard techniques, as well as extraction of coarser sulphides using a dentist’s drill. In addition, use of an in situ laser combustion system allowed greater resolution of sampling to sub-millimetric levels on polished samples. For these analyses, samples were rastered under a laser beam to combust individual zones extending around 500 by 100 μm following the technique described in detail by Kelley and Fallick (1990) and Wagner et al. (2002). Determination of the sulphur isotope composition of the purified SO2 gas from both techniques was completed on a VG SIRA II gas mass spectrometer. The reference gas in the mass spectrometer was calibrated by regularly analysing a suite of international and lab standards, including NBS-123, IAEA-S-3 and SUERC standard CP-1. These gave δ34S values of +6.3‰, −36.5‰ and −12.8‰ respectively, with 1σ reproducibility around ±0.2‰. Data are reported in δ34S notation as per mil (‰) variations from the Vienna Canyon Diablo Troilite (V-CDT) standard. A mineral specific sulphur isotope fractionation occurs between the host mineral and the SO2 gas produced via laser combustion (Kelley and Fallick 1990). Fractionation corrections for pyrite (raw gas δ34S +0‰), chalcopyrite (raw gas δ34S +0‰) and bornite (raw gas δ34S +0.3‰) were determined experimentally through comparison with conventionally analysed material recovered from the same sample block.

Whole rock analyses Copper, cobalt and sulphur assay values show the highest downhole concentrations in the biotite–muscovite–quartz– kyanite ore schist, although not all schist units are mineralized (Fig. 7). Notably, the hanging wall and intercalated gneiss units are essentially devoid of copper mineralization, which is in contrast to the downhole whole rock geochemical data for Nchanga, which showed a consistent geochemical anomaly for Cu throughout the sampled core (McGowan et al. 2006). There is a correlation between the

Copper 120.00

Cobalt 140.00

Sulfur

160.00

180.00

200.00

220.00

240.00

260.00

280.00

Gneiss Schist Fig. 7 Downhole assay data for diamond drill hole EQ-CHI-62. Note almost complete lack of copper grades within the gneiss units

copper and cobalt mineralization, although Co maxima are in places slightly offset from Cu maxima, an observation reported from within the Copperbelt ores (Annels and Simmonds 1984; McGowan et al. 2006). Plots of Zr versus Nb, Cr and TiO2 for rocks from Lumwana suggest a relationship between the gneisses, ore schists and amphibolites (Fig. 8). The schistose rocks tend to show lower Zr and Nb contents and higher Cr values than the gneisses with relatively constant TiO2. The trends are consistent with the more mafic character of the schists and amphibolites compared to the gneisses. In addition, the considerable compositional overlap between the schists and gneisses is as expected given core and thin section studies that suggest the schists developed by progressive deformation (shearing) and alteration of the gneisses. Included on these diagrams is a whole rock data set for all rock types observed at the Nchanga Mine, enabling a comparison with host rocks from a typical Lower Roan-hosted Zambian Copperbelt deposit, and which show significantly lower Zr and Cr values than the rocks from Lumwana (Fig. 8). The application of Grant diagrams (Grant 1986) using Shearcalc TM (Sturm 2003) to compare ore schist at Chimiwungo and Malundwe with hanging wall gneiss units

146

Miner Deposita (2013) 48:137–153 50

30

45

Co

Cu Ni

I

Rb

V

40

Cv

Cr

Cf Ore Schist

Nb (ppm)

Cm

Ga

Ba

35 30 25 20 15

Sc

Al2O3

Cd

Zn

Bi

Y U

Se Na2O

0

400

500

Zr (ppm)

30

Sr

30

15

600

250

La

Malundwe

CaO

0

300

P2O5

Ce

5 200

Nb Zr

MnO K2O

100

Sb

Pb Th

S Ti2O

0

SiO2

15

10

0

Fe2O3

MgO Hf

Cu

P2O5

Co Hanging Wall Gneiss I Hf Co Ba

K2 O

Cm

Cl

Cv

Cr

Cf Ore Schist

Cr (ppm)

200

150

100

V

Zr

U

TiO2

300

400

500

Zr (ppm) Chimiwungu Gneiss Malundwe Gneiss Malundwe Ore Chimiwungu Ore Amphibolites Nchanga Data

1.80

TiO2 (wt%)

1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 0

100

200

300

400

500

600

Zr (ppm) Fig. 8 Whole rock analyses of Lumwana and Nchanga rocks. Gneiss rocks undergoing alteration to schists which show lower Zr and Nb with increased concentrations of Cr and TiO2. Whole rock analyses of Nchanga rocks are shown for comparison

Y

Se

Th La Ce

Sr Na2O

Chimiwungu

CaO

15

30

Co Hanging Wall Gneiss

600

2.00

Zn Bi Sb Pb

0

200

Al2O3

Nb

0

100

Cd

15

50

0

MnO Ga

Rb Fe2O3

Ni Sc

S

0

MgO

Fig. 9 Grant (1986) diagrams which illustrate the typical enrichments and depletions observed when comparing a hanging wall gneiss with ore schist. Compared to hanging wall gneiss, the ore schists are typically enriched in Cu, Co, Ni, V, Ti, Sc and Ba, and depleted in Na, Ca, Sr and Y. I isocon, Cm constant mass, Cv constant volume

Maynard 2005) and Nchanga (McGowan et al. 2006). Investigating the Ba enrichment associated with copper ore at Konkola, Sutton and Maynard (2005) reported Ba-rich Kfeldspar overgrowths with up to 4 wt% Ba. However the location of the Ba-enrichment at Lumwana is less obvious. The enhanced Ni values likely correlate with the presence of the cobalt sulphide phases siegenite ((Ni,Co)3S4), bravoite ((Fe,Ni,Co)S2), cobalt pentlandite ((Co,Ni,Fe)9S8) and carrollite (Cu(Co,Ni)2S4) identified within the Chimiwungo ore schist.

Mineral chemistry of the phyllosilicate phases (Fig. 9) shows that at both deposits the ore schist is substantively depleted in CaO, Sr and Na2O, and enriched in MgO and K2O, consistent with replacement of feldspar by biotite and with the mode of feldspar reducing from around 35 vol% to 0 % in the ore schist. The diagrams also confirm the enrichment of the ore schist units in Cr and depletion in Zr and Nb (Fig 8). The ore schist also shows increases in Cu, Co, V, Ni, Sc, Ba and S contents. This enrichment suite is typical of Copperbelt ores at Konkola (Sutton and

Muscovite Muscovite compositions were determined from samples of hanging wall gneiss, schist and the ore schist. All muscovite data show an inverse covariation between octahedral Al and octahedral cations which define a trend with a slope of approximately -1, which follows the Tschermak substitution (Fig. 10). Muscovites from the gneisses consistently show

Miner Deposita (2013) 48:137–153

Muscovite 0.22

4

0.17 3.5

X Na

Octahedral Al

Fig. 10 Electron microprobe analyses of muscovite from the Lumwana system. Typically the ore schists show increased XMg values and lower total cations compared to the gneiss units

147

0.12

3 0.07 0.02

2.5 0.2

0.4

0.6

0.8

1

0.2

0.4

0.6

X Mg Mg+Fe+Ti+Mn Cations

0.21

Ti Cations

0.8

1

X Mg

0.16 0.11 0.06

1.52

1.02

0.52

0.02

0.01 0

0.2

0.4

0.6

0.8

1

2.5

3

3.5

4

Octahedral Al Cations

X Mg

Ba Cations

0.04

HW Gneiss

0.03

HW/FW Schist Ore Schist

0.02

0.01 0.2

0.4

0.6

0.8

1

X Mg

the lowest XMg values with low Al, Na and elevated total cations and Ti compared to the schists and ore schist. In contrast, muscovites from the ore schist show significantly higher XMg values, increased Al and consequently, low total cations. Biotite The biotites analysed from Lumwana are typically phlogopite with XMg >0.5, and with low Ti (<0.35 apfu) and halogen contents (Fig. 11). Solid solution trends in the biotites are similar to those observed in the muscovite data. Within the gneisses, the biotite compositions show a limited range with XMg of 0.5 and Na <0.2 wt%; whereas within the schist units, the XMg is more variable 0.4–0.9. In particular, the ore schists tend to show highest octahedral Al values and lowest total cations on the octahedral site. However, compared to the muscovite data the biotite data less readily distinguish the ore schist from hanging wall schists.

Chlorite Chlorite is present as sheridanite and ripidolite, with solid solutions in keeping with the previous mica analyses. Octahedral Al content varies inversely with Fe + Mg and the Si content varies between 5.2 and 5.6 apfu, tending to show little systematic variation between units (Fig. 12). Application of the chlorite geothermometers of Zang and Fyfe (1995) and Kranidiotis and MacLean (1987) indicates temperatures of around 230–250°C, consistent with the late textural setting of chlorite phases. Sulphur isotope data Sulphides from Chimiwungo and Malundwe display a broad range of values from δ34S +2.3‰ to +18.5‰ (Fig. 13) which extend earlier data of Dechow and Jensen (1965) who reported δ34S values of +6‰ to +13.3‰. These δ34S data are consistent with reported values from the west of the

148

Biotite Fe+Mg+Ti+Mn Cations

1

Octahedral Aluminium

Fig. 11 Electron microprobe analyses of biotite from the Lumwana deposits. Legend as previous figure. As for the muscovite data, hanging wall schist and ore schist show greater XMg values compared to gneisses

Miner Deposita (2013) 48:137–153

0.8 0.6 0.4 0.2 0

6.5 6 5.5 5 4.5 4 3.5 3

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0

1

0.2

0.4

0.4

0.3

0.3

Cl (wt%)

Ti Cations

0.4

0.6

0.8

1

1.2

Octahedral Aluminium

XMg

0.2

0.2

0.1

0.1

0

0 0

0.2

0.4

0.6

0.8

1

0.2

0.3

0.4

0.5

XMg

0.6

0.7

0.8

0.9

1

0.7

0.8

0.9

1

XMg 1

11

Na (wt%)

K (wt%)

10.5 10 9.5 9

0.75

0.5 0.25

8.5 0

8 0

0.2

0.4

0.6

0.8

XMg

1

0.2

0.3

0.4

0.5

0.6

XMg

HW Gneiss HW/FW Schist Ore Schist

Copperbelt; with Kansanshi showing δ34S between +1.8‰ to +7.9‰ and mineralization in the Solwezi Dome giving values of +0.7‰ to +7.9‰ (Dechow and Jensen 1965). Significant variations in δ34S are observed within individual samples. For example, Eq-Mal-094-4-3 shows δ34S values between +7.3‰ and +12.4‰, which suggest that the initial δ34S signatures are preserved despite subsequent metamorphism. At Chimiwungo, a wider range of δ34S values are observed than at Malundwe, with a significant cluster of samples in the range δ34S+12‰ to +18‰, which are not apparent at Malundwe (Fig. 13).

Discussion The setting of the Cu (±Co ± U) mineralization described at Lumwana is in contrast to the widely reported Lower Roanhosted deposits of the Copperbelt. The metamorphic grade of the host rock to the mineralization, strong fabric development, transitional nature of contacts between gneiss and ore schist and interleaving of ore schist and slivers of gneissic basement strongly suggest the Lumwana deposits are hosted in basement to Lower Roan stratigraphy. However, other features of the Lumwana mineralization are

Miner Deposita (2013) 48:137–153

149 7

Chlorite Data

Chimiwungu

9.60

a

6

9.50 9.40

5 4

9.20

Count

Fe+Mg

9.30

9.10 9.00 8.90

3 2

8.80 1

8.70 8.60 2.40

2.60

2.80

0

3.00

2

Octahedral Al 0.60

6

8

10

12

b

0.50

14

16

18

20

Malundwe

6

0.40

Count

Fe/Fe+Mg

4

8

0.30

4

0.20 2

0.10 0.00 5.00

5.20

5.40

5.60

0

5.80

6

10

c

Fig. 12 Electron microprobe analyses of chlorites from Lumwana

12

14

16

18

20

Hangingwall Gneiss

2.0

1.5

Count

common to the Copperbelt, for example: the observed suite of ore minerals; typical metal enrichment (Cu, Co, V, U, Ni); presence of biotite as an alteration mineral; and the δ34S data, routinely >0‰, lacking the low δ34S values typically associated with sedimentary copper deposits. The host rock to the mineralization exhibits a schistose fabric defined by aligned biotite grains, with S-C fabrics and asymmetric augen suggesting a top to the north sense of shear. The presence of transitional contacts between host rock, schist and ore schist appears to be the result of a gneiss undergoing progressive deformation and alteration to form schist. The alteration is observed in thin-section, and reflected in the whole rock geochemistry, by the loss of feldspar resulting in depletions of Ca, Na, Sr with Mg and K partially buffered by the increased mode of muscovite and biotite in the ore zone. Although the host rocks to the Lumwana copper sulphide mineralization have been correlated with the Lower Roan by previous authors (Benham et al. 1976; McGregor 1965), the observation of progressive alteration from hanging wall gneiss to ore schist suggests that this is untenable. Further consideration of the whole rock data also suggests a contrasting setting for the Lumwana ores. First, using data from Nchanga as an example of typically mineralized Lower Roan it is evident that the

8

2.5

Si Cations

1.0

0.5

0.0 4

6

8

10

12

14

16

18

20

δ S (o/oo) 34

Fig. 13 Sulphur isotope data a, b show data from the ore schist at Chimiwungu and Lumwana. c Pyrite data from hanging wall gneiss at both localities

bulk rock assemblage of Nchanga is much less silica-rich (Zr-poor) and Cr-poor compared to the Lumwana rocks, and secondly the lack of any significant background mineralization or elevated values of Cu, Co, U and V away from the shear zones at Lumwana is particularly striking. Comparing hanging wall gneisses and ore schist within Grant (1986) diagrams (Fig. 9) shows enrichments and depletions which are consistent with the alteration of feldspar and generation of biotite. Applying densities of 2.8 g/cm3 for the

150

gneiss and 3.2 g/cm3 for the ore schist suggests that the ore schist shear zones at Malundwe and Chimiwungo are consistent with mass and volume losses between gneiss and schist of 32 % and 40 % for Malundwe and 40 % and 48 % for Chimiwungo. These mass and volume losses are in line with those reported from other ductile shear zones developed in granite protoliths, where high silica rocks are transformed to silica undersaturated schist when subject to strain and substantive fluid–rock ratios (Selverstone et al. 1991; Sturm and Streyrer 2003). Furthermore, the geochemistry and mineral assemblages at Lumwana are in keeping with the formation of kyanitebearing selvages around quartz veins during the Barrovian metamorphism of pelites in New England (Ague 1994) and during the development of ductile shear zones in granite rocks of the Ille de Yeu, France (Sassier et al. 2006). In each of these examples, kyanite-bearing rocks developed within shear zones which record a loss of feldspar and consequent relative depletions in whole rock CaO, Na2O and Sr values. In each case, the loss of feldspar is compensated for by an increased mode of biotite, whole rock MgO values and an increase in XMg of the phyllosilicates. Thus, the whole rock and mineral chemistry observed at Lumwana are entirely consistent with the development of shear zone-related schists within a granite gneiss. The close association of biotite and sulphides at Lumwana, with sulphides often developed along cleavage in the mica grains, suggests that sulphide precipitation may be linked in part to the presence of, or potential reduction capacity of biotite, as oxidized metal-bearing brines enter the shear zone. However, the biotite–sulphide relationship may simply reflect the post-mineralization metamorphic recrystallization of the sheared and mineralized rocks. Subsequent metamorphism of these refractory and aluminous horizons then leads to the development of kyanite during prograde metamorphism. Notably, a similar conclusion was reached for the uranium mineralization at Lumwana (Cosi et al. 1992). They suggested the U mineralization is most likely related to shear zones, which predate the onset of peak metamorphism and which are related to the development of a nappe pile within the Domes Region. The majority of the Zambia Copperbelt ores occur within the Lower Roan, with a broad coherence in their stratigraphic location (Selley et al. 2005). Nevertheless, recent studies suggest significant relationships between the mineralization and the tectonic evolution of the basin. For example, at Nchanga, sulphide mineralization cross cuts stratigraphy and shows a strong spatial location with inversion structures linked to faults within the underlying basement (McGowan et al. 2003, 2006; Roberts et al. 2009). At Nkana Cusulphide mineralization replaces early diagenetic disseminated framboidal pyrites with associated silicification, K-

Miner Deposita (2013) 48:137–153

feldspar alteration, albitization and carbonatization typically at the contact between the Footwall Sandstone Formation and overlying organic-rich shales of the ore formation. However the high grade ores are generally restricted to the hinge zones of tight to isoclinal folds (Brems et al. 2009). At Kansanshi copper mineralization is hosted within a set of high-angle veins, which cross cut the prevailing fabric, show Re-Os ages of 511 and 503 Ma, and occur at stratigraphic levels which correlate with the Upper Roan (Broughton et al. 2002). These data suggest that mineralization occurred throughout much of the Lufilian fold and thrust belt during and after peak metamorphism. The recognition of multistage ore formation through the geological evolution of the Zambian Basin is a recurring theme: Selley et al. (2005), Hitzman et al. (2005) and Dewaele et al. (2006) all concluded that the Katangan Supergroup experienced a protracted history of mineralogical modification from earliest diagenesis to basin collapse which records multiple stages of ore formation and basinal brine migration. Recent modelling suggests that supersaturated brines, as recognised by McGowan et al. (2006), generated beneath a halite seal can develop convective hydrothermal plumes which penetrate through the Lower Roan and into the crystalline basement, despite its low permeability and regardless of the availability of cross-stratal conduits (Koziy et al. 2009). However, the exact timing of the mineralization at Lumwana remains an open question. Although petrographic evidence supports a pre-peak metamorphic origin, whether the initial mineralization is related to a rift or a post-rift inversion event is unclear. Sulphur isotope data The sulphur isotope data for Lumwana are comparable to many of the δ34S data for the Copperbelt, with the majority of deposits tending to show restricted values between 0‰ and 20‰, i.e. rarely showing the range of δ34S values reported from sedimentary copper deposits around the world. Following a comprehensive δ34S isotope study of the Copperbelt, Dechow and Jensen (1965) recognised that a simple biogenic origin for sulphur from most of the deposits studied was not supported by their data. They suggested that the sulphur in the deposits was originally biogenic, but subsequently modified and homogenized by later metamorphic and metasomatic events. However, data from Konkola (Sweeney et al. 1986), Kolwezi (Muchez et al. 2008; 2010) and Luiswishi and Kamoto (El Desouky et al. 2010) report low δ34S values for sulphides within ore horizons, especially within the Democratic Republic of Congo, which are consistent with biogenic sulphate reduction. Furthermore, in a detailed study at Nchanga, McGowan et al. (2003) recorded clear variations in the δ34S composition of the sulphides present in the host sediments and ore bodies. These

Miner Deposita (2013) 48:137–153

data confirmed that metamorphic homogenization had not taken place and that thermochemical sulphate reduction was the most likely mechanism responsible for the sulphur isotope ratios observed. In addition, at Nchanga the upper ore body, the Co mineralized ore, showed the largest range and greatest δ34S values. Interestingly, this is also the case at Lumwana where the greatest δ34S values are almost exclusively preserved within the Co-bearing Chimiwungu deposit. In contrast, Malundwe shows only a few δ34S values greater than 10 ‰. These data are best explained if either some form of close system fractionation is operating during the development of the ore deposits, which allows for the consumption of 32S in the generation of sulphide mineralization or alternatively, a sulphate source other than diagenetic anhydrite becomes available during the mineralization which has a different δ34S ratio than reported from much of the anhydrite in the Copperbelt, which typically shows a δ34S of ∼21‰ (Dechow and Jensen 1965). Other data from the domes region and basement are comparable to Lumwana with sulphur isotope data from the basement hosted mineralization of the Roan Extension (+2‰ to +10‰), Grays Quarry (+3‰ to +13‰) and Kafue Dome (+13‰ to +16‰), showing comparable values although a notable exception is Fimpimpa (−2.3‰ to −10.5‰; Dechow and Jensen 1965). Pre-Katangan basement mineralization of the Zambian Copperbelt Mineralization within basement to the Lower Roan ore deposits around the Kafue Anticline are reported but have been subject to little scientific analysis. Whilst commonly of low grade, when compared to the mineralization of the Zambian Copperbelt, higher grade mineralization has been identified associated with basement shear zones (Mendelsohn 1961; Pienaar 1961; Wakefield 1978). Mineralogical and textural similarities are apparent between the Lumwana deposits and the basement mineralization of the Kafue Anticline. For example, at Fimpimpa, veinhosted malachite, chalcopyrite and bornite mineralization is reported from within an 8-m-wide shear zone developed within a feldspathized quartz diorite (Pienaar 1961). The Samba Deposit of the Lufubu Metamorphic Complex (Wakefield 1978; Rainaud et al. 2005b) is a large concentration of Cu mineralization in the pre-Katangan basement of the Kafue Anticline. The sulphide mineralization of the Samba deposit shows some striking similarities to Lumwana, consisting of disseminations, stringers and veinlets of pyrite, chalcopyrite and bornite in a deformed quartz–sericite schist (Wakefield 1978). Textural and mineralogical similarities are apparent between the Lumwana and Samba deposits, with sulphides aligned with the foliation and bornite and pyrite show an antithetic relationship (Wakefield 1978). The deformed rocks at Samba are typically L-S tectonites that are also prevalent at

151

Lumwana. Wakefield (1978) also describes a pervasive sericitization of plagioclase at the Samba Deposit that increases in intensity towards the mineralized zone. The degree of sericitization correlates with the intensity of the D1 deformation. However, local zones of undeformed and totally sericitized rock indicate that alteration predates the D1 deformation. Although the Samba deposit has been subject to greenschist–facies conditions, the sericitization is highly analogous to the alteration observed at Lumwana. It is tempting to speculate the Samba deposit may be a much smaller and lower metamorphic grade equivalent of the Lumwana deposits.

Summary The ore schist at Lumwana developed within granite-gneiss sequences of the Mwombeshi Dome. The development of the ore schist, in terms of major and trace element chemistry, is entirely consistent with the formation of kyanite-bearing shear zones. Given the textural setting of the mineralization, it also seems reasonable that the mineralization was introduced prior to peak metamorphism and growth of kyanite, either within an initial extensional environment or early stages of north related thrusting. These new data provide compelling evidence that the Lumwana mineralization is a basement-hosted deposit similar to Samba and others. Such a conclusion significantly expands the potential sites of potential mineralization within the Zambian Copperbelt and Domes Region. Acknowledgements Robin Bernau was supported by a NERC studentship NER/S/A/2003/11857. Many geologists contributed their thoughts on the formation of the Lumwana deposits and we appreciate all their input, but in particular that of Greg Winch of Equinox Resources. Kate Davis is thanked for drafting some of the diagrams. We are grateful for two anonymous referees for robust reviews which greatly improved the manuscript.

References Ague JJ (1994) Mass-transfer during barrovian metamorphism of pelites, south-central Connecticut II. Channelized fluid-flow and the growth of staurolite and kyanite. Am J Sci 294:1061–1134 Annels AE (1989) Ore genesis in the Zambian Copperbelt, with particular reference to the northern sector of the Chambishi Basin. Geol Assoc Can Spec Pap 36:427–452 Annels AE, Simmonds JR (1984) Cobalt in the Zambian Copperbelt. Precambrian Res 25:75–98 Armstrong RA, Master S, Robb LJ (2005) Geochronology of the Nchanga Granite, and constraints on the maximum age of the Katanga Supergroup, Zambian Copperbelt. J Afr Earth Sci 42:32– 40 Benham DG, Greig DD, Vink BW (1976) Copper occurrences of Mwombezhi Dome area, northwestern Zambia. Econ Geol 71:433–442

152 Brems D, Muchez Ph, Sikazwe O, Mukumba W (2009) Metallogenesis of the Nkana copper-cobalt South Orebody, Zambia. J Afr Earth Sci 55:185–196 Broughton DW, Hitzman MW, Stephens AJ (2002) Exploration history and geology of the Kansanshi Cu(-Au) Deposit, Zambia. Spec Publ Soc Econ Geol 9:141–153 Cailteux JLH, Kampunzu AB, Lerouge C, Kaputo AK, Milesi JP (2005) Genesis of sediment-hosted stratiform copper-cobalt deposits, Central African Copperbelt. J Afr Earth Sci 42:134– 158 Chopin C (1984) Coesite and pure pyrope in high-grade blueschists of the western alps—a 1st record and some consequences. Contrib Mineral Petrol 86:107–118 Cosi M, Debonis A, Gosso G, Hunziker J, Martinotti G, Moratto S, Robert JP, Ruhlman F (1992) Late proterozoic thrust tectonics, high-pressure metamorphism and uranium mineralization in the domes area, Lufilian Arc, northwestern Zambia. Precambrian Res 58:215–240 Daly MC, Chakraborty SK, Kasolo P, Musiwa M, Mumba P, Naidu B, Namateba C, Ngambi O, Coward MP (1984) The Lufilian Arc and Irumide Belt of Zambia; results of a geotraverse across their intersection. J Afr Earth Sci 2:311–318 Dechow E, Jensen ML (1965) Sulfur isotopes of some central African sulfide deposits. Econ Geol 60:894–941 Dewaele S, Muchez P, Vets J, Fernandez-Alonzo M, Tack L (2006) Multiphase origin of the cu-co ore deposits in the western part of the lufilian fold-and-thrust belt, katanga (democratic republic of congo). J Afr Earth Sci 46(5):455–469 El DH, Muchez Ph, Boyce AJ, Schneider J, Cailteux JLH, Dewaele S, von Quadt A (2010) Genesis of sediemnt-hosted stratiform copper-cobalt mineralization at Luiswishi and Kamoto, Katanga Copperbelt (Democratic Republic of Congo). Miner Deposita 45:735–763 Equinox Minerals TSx Technical Reports 2008/2011. (http://www.equi noxminerals.com/files/files/409_TechnicalReport_16_Jun_08.pdf) Frimmel HE, Basei MS, Gaucher C (2011) Neoproterozoic geodynamic evolution of SW-Gondwana: a Southern African perspective. Int J Earth Sci 100:323–354 Grant JA (1986) The isocon diagram—a simple solution to Gresens equation for metasomatic alteration. Econ Geol 81:1976–1982 Hanson RE, Wardlaw MS, Wilson TJ, Mwale G (1993) U-Pb zircon ages from the Hook granite massif and Mwembeshi dislocation; constraints on Pan-African deformation, plutonism, and transcurrent shearing in central Zambia. Precambrian Res 63:189–209 Hitzman M, Kirkham R, Broughton D, Thorson J, Selley D (2005) The sediment-hosted stratiform copper ore system. In J. W. Hedenquist, J. F. H. Thompson, R. J. Goldfarb & J. P. Richards (Eds.), Economic geology; one hundredth anniversary volume, 1905–2005. Society of Economic Geologists, 609–642 John T, Schenk V, Haase K, Scherer E, Tembo F (2003) Evidence for a Neoproterozoic ocean in south-central Africa from mid-oceanicridge-type geochemical signatures and pressure-temperature estimates of Zambian eclogites. Geology 31:243–246 John T, Schenk V, Mezger K, Tembo F (2004) Timing and PT evolution of whiteschist metamorphism in the Lufilian Arc-Zambezi Belt Orogen (Zambia): implications for the assembly of Gondwana. J Geol 112:71–90 Kelley SP, Fallick AE (1990) High-precision spatially resolved analysis of δ34S in sulfides using a laser extraction technique. Geochim Cosmochim Acta 54:883–888 Koziy L, Bull S, Large R, Selley D (2009) Salt as a fluid driver, and basement as a metal source, for stratiform sediment-hosted copper deposits. Geology 37(12):1107–1110 Kranidiotis P, MacLean WH (1987) Systematics of chlorite alteration at the Phelps Dodge massive sulfide deposit, Matagami, Quebec. Econ Geol 82:1898–1911

Miner Deposita (2013) 48:137–153 Massonne HJ, Schreyer W (1989) Stability field of the high-pressure assemblage talc + phengite and 2 new phengite barometers. Eur J Mineral 1:391–410 McGowan RR, Roberts S, Foster RP, Boyce AJ, Coller D (2003) Origin of the copper-cobalt deposits of the Zambian Copperbelt; an epigenetic view from Nchanga. Geology 31:497–500 McGowan RR, Roberts S, Boyce AJ (2006) Origin of the Nchanga copper-cobalt deposits of the Zambian Copperbelt. Miner Deposita 40:617–638 McGregor JA (1965) The Lumwana copper prospect in Zambia. Unpub. PhD Thesis. Rhodes University Mendelsohn F (1961) The geology of the Northern Rhodesian Copperbelt. MacDonald, London, 523p Muchez P, Vanderhaeghen P, El Desouky H, Schneider J, Boyce A, Dewaele S, Cailteux J (2008) Anhydrite pseudomorphs and the origin of stratiform Cu-Co ores in the Katangan Copperbelt (Democratic Republic of Congo). Miner Deposita 43:575–589 Muchez Ph, Brems D, Clara E, De Cleyn A, Lammens L, Boyce A, De Muynck D, Mukumba W, Sikazwe O (2010) Evolution of Cu-Co mineralizing fluids at Nkana Mine, Central African Copperbelt, Zambia. J Afr Earth Sci 58:457–474 Nyambe IA, Kawamya M (2005) Approaches to sustainable minerals development in Zambia. Geol Soc Spec Publ 250:73–86 Pavoni (1992) Rifting of Africa and pattern of mantle convection beneath the African Plate. Tectonophysics 215(1–2):35–53 Pienaar PJ (1961) Mineralization in the basement. In: Mendelsohn F (ed) The geology of the Northern Rhodesian Copperbelt. MacDonald, London, pp 30–37 Porada H, Berhorst V (2000) Towards a new understanding of the Neoproterozoic-Early Palaezoic Lufilian and northern Zambezi Belts in Zambia and the Democratic Republic of Congo. J Afr Earth Sci 30:727–771 Rainaud C, Master S, Armstrong RA, Robb LJ (2005) Geochronology and nature of the Paleoproterozoic basement in the Central African Copperbelt (Zambia and the Democratic Republic of Congo), with regional implications. J Afr Earth Sci 42:1–31 Roberts S, Palmer MR, Cooper MJ, Buchaus P, Sargent D (2009) REE and Sr isotope characteristics of carbonate within the Cu-Co mineralized sedimentary sequence of the Nchanga Mine, Zambian Copperbelt. Mineralium Deposita 44:881–891 Sassier C, Boulvais P, Gapais D, Capdevila R, Diot H (2006) From granitoid to kyanite-bearing micaschist during fluid-assisted shearing (Ile d'Yeu, France). Int J Earth Sci 95:2–18 Selley D, Broughton D, Scott R, Hitzman M, Bull SW, Large RR, McGoldrick PJ, Croaker M, Pollington N, Barra F (2005) A new look at the geology of the Zambian Copperbelt. Economic Geology 100th Anniversary Volume, 965–1000 Selverstone J, Morteani G, Staude J-M (1991) Fluid channeling during ductile shearing: transformation of granodiorite into aluminous schist in the Tauern Window, Eastern Alps. J Metamorph Geol 9:419–431 Steven N, Armstrong R (2003) A metamorphosed proterozoic carbonaceous shale-hosted Co-Ni-Cu deposit at Kalumbila, Kabompo Dome: the Copperbelt ore shale in northwestern Zambia. Econ Geol 98:893–909 Sturm R (2003) SHEARCALC—a computer program for the calculation of volume change and mass transfer in a ductile shear zone. Comput Geosci 29:961–969 Sturm R, Steyrer HP (2003) Use of accessory zircon for the quantification of volume changes in ductile shear zones cutting plutonic rocks. Chemie de Erde 63:31–54 Sutton SJ, Maynard JB (2005) A fluid mixing model for copper mineralization at Konkola North, Zambian Copperbelt. J Afr Earth Sci 42:95–118 Sweeney M, Turner P, Vaughan DJ (1986) Stable isotope and geochemical studies in the role of early diagenesis in ore

Miner Deposita (2013) 48:137–153 formation, Konkola Basin, Zambian copper belt. Econ Geol 81:1838–1852 Wagner T, Boyce AJ, Fallick AE (2002) Laser combustion analysis of delta S-34 of sulfosalt minerals: determination of the fractionation systematics and some crystal-chemical considerations. Geochim Cosmochim Acta 66:2855–2863

153 Wakefield J (1978) Samba. A deformed porphyry-type copper-deposit in the basement of the Zambian Copperbelt. Trans Instit Min Metall Sect B-Appl Earth Sci 87:B43–B52 Zang W, Fyfe WS (1995) Chloritization of the hydrothermally altered bedrock at the Igarape-Bahia gold deposit, Carajas, Brazil. Miner Deposita 30:30–38