Exploration for Komatiite-Associated Ni-Cu-(PGE) Mineralization in the Thompson Nickel Belt, Manitoba
Published:January 01, 2010
- PDF LinkChapter PDF
D. Layton-Matthews, C. M. Lesher, O. M. Burnham, L. Hulbert, D. C. Peck, J. P. Golightly, R. R. Keays, 2010. "Exploration for Komatiite-Associated Ni-Cu-(PGE) Mineralization in the Thompson Nickel Belt, Manitoba", The Challenge of Finding New Mineral Resources: Global Metallogeny, Innovative Exploration, and New Discoveries, Richard J. Goldfarb, Erin E. Marsh, Thomas Monecke
Download citation file:
The Ni-Cu-(PGE) deposits of the Thompson nickel belt in the Circum-Superior boundary zone of northern Manitoba define the second largest Ni-Cu-(PGE) mining camp in Canada and one of the premiere Ni-Cu-(PGE) camps of the world. Despite a complex deformation and metamorphic history, the deposits in the Thompson nickel belt exhibit many fundamental characteristics similar to those of other major magmatic Ni-Cu-(PGE) districts: they are hosted by or associated with ultramafic intrusions that appear to represent dynamic feeders, the ores occur at or near the bases of the intrusions, and there is evidence for incorporation of significant amounts of sulfur from the Ospwagan Group metasedimentary country rocks. However, they differ from most other deposits of this type in being metamorphosed to much higher grades, in being much more complexly deformed, and in being mobilized to much greater degrees into the country rocks. The ultramafic intrusions are generally lensoid in shape, reflecting the effects of superimposed deformation on the enclosing metasedimentary rocks, range in composition from komatiitic dunite to komatiitic pyroxenite, are variably serpentinized, and are interpreted to represent a series of sills and low-angle dikes that intruded and interacted with the Ospwagan Group metasedimentary rocks. High Fo contents in relict igneous olivine (as much as Fo92) indicate a low Mg komatiitic parental magma with 22 to 24 percent MgO. Mineralization occurs as type II disseminated sulfides within the ultramafic rocks (e.g., William Lake), as type V tectonically modified massive sulfides within or adjacent to the ultramafic bodies (e.g., Pipe and Birchtree), and as type IV magmatically and metamorphically mobilized sulfides within metasedimentary rocks of the Ospwagan Formation (e.g., Thompson). Intrusions occur at all almost all levels within the Ospwagan Group, but mineralized intrusions are localized exclusively within the lower and middle parts of the Pipe Formation, which contains abundant sulfide-facies iron formation. Density-driven magma emplacement models indicate that the Ospwagan metasedimentary rocks were likely partially lithified prior to magma emplacement and the absence of significant thermal aureoles suggests that they were being metamorphosed. Stratigraphic correlations between ultramafic intrusions, S-rich rocks of the Pipe Formation, and Ni-Cu-(PGE) sulfide mineralization, together with nonmantle δ34S values and S/Se ratios in the ores and nonmantle Th/Yb and Th/Nb ratios in the host rocks, collectively suggest that the mineralization formed by incorporation of S-rich sedimentary rocks by high-temperature komatiitic magmas. Postore deformation and metamorphism have significantly modified the primary characteristics of many of the Thompson nickel belt ore deposits, mobilizing Cu, Au, and Pt. The best exploration tools appear to be aeromagnetic surveys to identify serpentinized ultramafic bodies, which are the heat and metal sources; stratigraphic studies to recognize appropriate levels of the Pipe Member of the Ospwagan Group, which is the S source; lithogeochemical studies to identify the most magnesian and most contaminated host units; which provide the evidence of magma-sediment interaction; and recognition of areas of anomalous Cu, Au, and Pt dispersion halos.
The THompson Nickel Belt in northern Manitoba is the second largest Ni-Cu-(PGE) mining camp in Canada, after Sudbury, with a premining resource estimated at 150.3 million metric tons (Mt) at 2.32 percent Ni, 0.16 percent Cu, 0.046 percent Co, 0.10 g/t Pt, 0.54 g/t Pd, 0.046 g/t Rh, 0.072 g/t Ru, 0.033 g/t Ir, and 0.041 g/t Os (Naldrett, 2004). It is the second largest camp of komatiite-associated metal deposits in the world, after the Eastern Goldfields of Western Australia, containing ˜17 percent of the global resources associated with komatiites (Hoatson et al., 2006). The Ni-Cu-(PGE) ores in the Thompson nickel belt are hosted by or associated with komatiitic dunites and peridotites (Layton-Matthews et al., 2007) that intruded a sequence of pelitic rocks and sulfideoxide-silicate−facies iron formations (Zwanzig et al., 2007). The intrusions subsequently underwent multiple phases of deformation (Fueten and Robin, 1989; Bleeker, 1990a; Gapais et al., 2005) and upper amphibolites-facies metamorphism (Bleeker, 1990a; Couëslan et al., 2007; Layton-Matthews et al., 2007). The deposits are similar to many komatiite-associated Ni-Cu-(PGE) deposits because they are associated with the most magnesian and olivine-rich ultramafic units in the sequence and with S-rich metasedimentary rocks. However, they are also distinguished from other komatiite deposits because they are depleted in metals (PGE >> Cu > Ni), have relatively low magma/sulfide ratios, are much more strongly deformed and metamorphosed, and individual deposits are much larger on average than those in other areas (e.g., Lesher, 1989; Hoatson et al., 2006).
Inco Ltd. discovered the deposits of the Thompson nickel belt in the 1950s, using one of the first functional airborne electromagnetic systems. Mining began in the late 1950s at the Moak Lake deposit but was suspended in 1957 when the magnitude of the Thompson deposit discovered in 1956 was appreciated. Mining continues to present day and, to date, eight mined deposits, seven unmined deposits, and many subeconomic prospects have been identified (Figs. 1, 2), with a total production of ˜150 Mt ore averaging 2.3 wt percent Ni, 0.16 wt percent Cu, 0.046 wt percent Co, and 0.83 g/t PGE (Naldrett, 2004; Layton-Matthews et al., 2007). However, despite almost 60 years of exploration and the identification of hundreds of ultramafic intrusions at different stratigraphic intervals in the Thompson nickel belt, less than one dozen are known to contain or be associated with economic mineralization.
The purpose of this paper is to review recent research on the geology of the Thompson nickel belt that provides new insights into the genesis and localization of the ores and new constraints on exploration methods in the Thompson nickel belt and in similar environments worldwide.
The Thompson nickel belt is located within the Circum-Superior boundary zone, which, in the Thompson nickel belt separates 3.1 to 2.6 Ga autochthonous Archean Superior province rocks in the east from 1.92 to 1.83 Ga allochthonous Proterozoic tectonic domains of the Trans-Hudson orogen to the west (Fig. 1). The Circum-Superior boundary zone contains the remnants of a Neoarchean-Paleoproterozoic continental margin sequence and comprises (from southeast to northwest): (1) autochthonous Archean (3.1–2.6 Ga) Superior basement rocks, including Pikwitonei granulites; (2) parautochthonous Superior basement rocks in the Thompson nickel belt, including retrogressed Pikwitonei granulites; (3) parautochthonous cover units of the Thompson nickel belt of Paleoproterozoic age deposited on the Superior margin (Ospwagan and Winnipegosis Groups); (4) allochthonous Archean crust (3.6–3.1 Ga; Bohm et al., 2000); and (5) allochthonous Paleoproterozoic (˜1.8 Ga) rocks of the Reindeer zone of Trans-Hudson orogen. It is bordered to the northwest by the internal (Reindeer) zone of the Trans-Hudson orogen, a 400-km-wide collage of 1.92 to 1.83 Ga arc and oceanic volcanic rocks, plutons, and 1.87 to 1.83 Ga continental and turbidite deposits (Fig. 1). The Circum-Superior boundary zone in northern Manitoba is interpreted as a major promontory of the foreland margin of the Superior craton that is flanked by 100-km-scale re-entrants (e.g., Winnipegosis Komatiite belt and Fox River belt: Lucas et al., 1996).
The Thompson nickel belt contains a wide variety of metasedimentary, metavolcanic, mafic-ultramafic intrusive, and felsic intrusive rocks. The metamorphic grade varies, resulting in different mineral assemblages for the same lithologic units. Because the protoliths are well established (e.g., Peredery, 1979; Bleeker, 1990a; Zwanzig et al., 2007), sedimentary, volcanic, and igneous terminology will be used in order to simplify the lithologic descriptions.
The basement rocks in the Thompson nickel belt are upper amphibolite- to granulite-facies tonalitic gneisses that grade into the Archean Pikwitonei granulite across a southeasterly gradient of decreasing Proterozoic deformation and recrystallization (Bleeker,1990b, and references therein). An Archean age has been established by U-Pb zircon geochronology, but there is also a prominent Proterozoic granitoid component in the basement rocks (Machado et al., 1990).
The basement gneisses are unconformably overlain by medium- to high-grade metasedimentary and metavolcanic rocks of the 1,000- to 3,000-m-thick Paleoproterozoic (2.1–to 1.89 Ga) Ospwagan Group (Zwanzig et al., 2007). The Ospwagan Group is subdivided into five conformable formations (Table 1): (1) a 1- to 50-m-thick basal quartzite unit (Manasan Formation); 2) a 1- to 100-m-thick carbonate unit (Thompson Formation), interpreted to represent a platform rift-drift succession; (3) a 10- to 200-m-thick transitional unit containing several cycles of fine-grained pelite, semipelite, quartz wacke, iron formation, and chert (Pipe Formation), interpreted to represent a continental slope succession; 4) a 10- to 200-mthick upper turbidite unit (Setting Formation), interpreted to represent a younger active rift succession; and (5) an uppermost, possibly allocthonous, unit of submarine mafic volcanic rocks, the Bah Lake assemblage that was previously known as the Ospwagan Formation.
Each metasedimentary unit in the Ospwagan Group is unique in terms of its mineralogical and petrographic composition, texture, weathering color, and quality of surface caused by selective weathering. Not all lithostratigraphic units are necessarily present at a given locality, as some may be very thin or even pinch out, but the stratigraphic order of the remaining units remains intact.
Petrologic end members (e.g., orthoquartzite, dolomitic marble, chert) remain recognizable into the highest metamorphic grades (granulite facies) and are excellent regional marker units. The best markers are the contrast between the highly heterogeneous basement and relatively homogeneous M1 member of the Manasan Formation, the carbonate-rich T1 and T3 members of the Thompson Formation, an Al-rich layer within the P2 metapelite schist member of the Pipe Formation that is characterized by chiastolite porphyroblasts, the P2 sulfide facies iron formation member of the Pipe Formation, the P3 dolomite marble member of the Pipe Formation that is hosted by silicate-facies iron formation, the pink-buff and brown “concretions” in the S1 member of the Setting Formation, and the megacrystic picrite facies of the Bah Lake assemblage.
The mineralogical compositions, grain sizes, and textures of transitional rock types, such as impure sandstone and semipelite, change with metamorphic grade. In granulite facies, they become indistinguishable from the basement rocks and can only be recognized as “ghost successions” of the Ospwagan Group if interbedded with end-member marker units.
Rocks of the Ospwagan Group have been intruded by multiple igneous units, including 1890 to 1880 Ga granitoids (Zwanzig et al., 2007, and references therein), 1883 ± 2 Ma (Heaman et al., 1986) mafic-ultramafic dikes of the Molson swarm, 1881 ± 2 Ma (Hulbert et al., 2005; Scoates et al., 2010) mafic-ultramafic sills that host or are associated with most of the Ni-Cu-(PGE) mineralization in the Thompson nickel belt, and a 1860.9 ± 5 Ma large composite mafic dike on the northwestern shoulder of the South Pit deposit (Scoates et al., 2010). Mafic-ultramafic sills that are associated with the Ni-Cu-(PGE) mineralization in the Thompson nickel belt occur at multiple levels within the Ospwagan Group metasedimentary sequence and underlying Archean gneisses, but all of the mineralized intrusions are now interpreted to occur in the Pipe Formation (Macek et al., 2001; Burnham et al., 2003; Zwanzig et al. 2007). In particular, a reevaluation of the stratigraphy using the above criteria and markers by Macek (2001) revealed that the country rocks at the Bucko deposit, which were originally interpreted as Archean gneisses, are actually strongly metamorphosed parts of the Thompson Formation. This places all known mineralized intrusions in the middle of the P1 (e.g., Thompson and Soab) or P2 (Birchtree, Bucko, Manibridge, and Pipe) members of the Pipe Formation, which are also the locations of the thickest sulfide-facies iron formations (Fig. 3). This does not mean that mineralization may not occur at other levels but that these levels are particularly favorable exploration targets.
In order to further quantify the stratigraphic distributions of intrusions and mineralization in the Thompson nickel belt, Burnham et al. (2003) calculated a stratigraphic index for each ultramafic body in the northern, exposed part of the Thompson nickel belt, based on the relative proportions of each formation with which it was in contact (Figs. 3, 4). Although there are minor variations in absolute values from different areas of the exposed Thompson nickel belt, in general the ultramafic intrusions show a similar distribution of host rocks. By far the greatest numbers of intrusions appear to have been emplaced into rocks of either the Pipe Formation (59%) or the Thompson Formation (12%), or along their contact (17%).
Deformation and metamorphism
Fueten and Robin (1989), Bleeker (1990a), Gapais et al. (2005), and Layton-Matthews et al. (2007) have described the complex deformation history that has affected the country rocks, intrusions, and ores in the Thompson nickel belt (Table 2). The results of these studies and their implications for exploration are synthesized below:
The overall structural style is characterized by a steeply dipping north-northeast−south-southwest foliation associated with dome-shaped, doubly plunging, north-northeast−trending folds. The repetition of favorable stratigraphic horizons by isoclinal recumbent folding has greatly complicated the stratigraphy but has also increased the number of prospective horizons for exploration.
The stratigraphic sequences within major structures are chronologically continuous and coherent, indicating that they reflect folding, not thrusting as previously believed, and that the stratigraphy can be used to identify favorable horizons for intrusion of ultramafic magmas and the formation of Ni-Cu-(PGE) mineralization (see below).
Conjugate ductile shear zones are present at all scales, reflecting principal subvertical stretching, intermediate along-strike stretching, and subhorizontal east-southeast– west-northwest shortening. These and the kinematic indicators (e.g., northeast over southwest rotation of porphyroblasts, east-southeast-up shear bands with sinistral components) suggest a west-verging regional transpressional regime, not an east-verging thrust regime as previously believed (Bell, 1971).
The same strain field is observed from partial melting to greenschist-facies conditions over a 100-m.y. time interval (1850–1750 Ma), suggesting that deformation was a major, progressive, and protracted event, not a limited, late event as previously believed (Bell, 1971).
Deformation was diachronous. It appears to have stopped at ˜1799 Ma in the east, ˜1770 Ma at Thompson, and ˜1750 Ma at Pipe. Because of their greater ductility during deformation, ore zones record some of the youngest ages and the lowest (retrograde) metamorphic assemblages. This is a potentially useful exploration indicator.
On both regional and local scales, the mineralized ultramafic intrusions experienced all phases of deformation observed in the surrounding metasedimentary rocks, indicating that they were emplaced after sedimentation, but prior to the initial phases of deformation. They are less deformed than the enclosing metasedimentary rocks but have often been dislocated, producing a series of variably deformed boudins.
Deformation mobilized massive sulfides, as far as 100 m from their host ultramafic rocks, leading to the partial or complete separation from the host ultramafic intrusions, attenuation of ores along fold limbs, and thickening of ores in fold noses.
Metamorphism in the Thompson nickel belt has been studied by Russell (1981), Paktunç (1984), Bleeker (1990a, b), and Couëslan et al. (2007). The southern half of the exposed part of the Thompson nickel belt is characterized by widespread upper amphibolite- to granulite-facies metamorphism, whereas the northern half of the belt is characterized by a northeasttrending, trough-like metamorphic low oriented subparallel to the trend of the belt (Fig. 2). Temperatures appear to increase to the northwest and southeast, and thus away from this trough, reaching granulite grade along the contact of the Thompson nickel belt with the Kisseynew domain (Fig.1; Couëslan et al., 2007).
Estimated peak metamorphic pressures range between 3.5 and 5.0 kbars and estimated temperatures range from ˜550°C to greater than 800°C (Fig. 5). The majority of meta-pelitic rocks are migmatitic and contain abundant biotite, quartz, K-feldspar, sillimanite, and/or garnet, indicating upper amphibolite-facies metamorphism at temperatures of ˜650° to 750°C. However, metapelitic rocks in the Moak Lake-Mystery Lake area contain sillimanite and muscovite, indicating middle amphibolite-facies metamorphism at ˜580° to 650°C, and Pipe Formation metapelites at the north end of the city of Thompson contain staurolite, biotite, and muscovite, but no aluminosilicates, indicating temperatures of ˜575° to 600°C. Garnet- and cordierite-bearing metapelitic rocks, and biotite- and orthopyroxene-bearing semipelitic rocks, along the northwestern margin of the belt and in the Mel zone-Strong Lake area, indicate granulite-facies metamorphism at temperatures of ˜725° to 850°C. Semipelitic rocks in the Phillips Lake and Joey Lake areas contain orthopyroxene and biotite, suggesting temperatures exceeding 800°C (Couëslan et al., 2007).
Because of the high degree of serpentinization, the ultramafic intrusions in the northern exposed part of the Thompson nickel belt have experienced intense glacial erosion and form recessive topographic features that are now filled by lakes. During the initial discovery period in the Thompson nickel belt (1946–1961), Inco Ltd. spent $27 million (CDN) on airborne magnetic and electromagnetic surveys, which was followed by ground geophysical surveys to define drill targets. Exploration in the exposed northern Thompson nickel belt relied heavily on airborne electromagnetic and magnetic surveys to identify regional targets. Application of these techniques to the southern Thompson nickel belt has proven difficult. The southern Thompson nickel belt is covered by 100- to 500-m-thick Paleozoic carbonates containing saline aquifers that inhibit the successful use of airborne- and ground-based surveys (McRitchie, 1995). Drilling and geophysical surveys by Inco, Falconbridge, Cominco, and Hudson Bay Mining indicate that most of the ultramafic intrusions occur as discrete boudins. However, the geometry and internal structure of many of the intrusions is still poorly constrained and only one-dimensional information is available from diamond drill core. The following is summarized from the work of Burnham et al. (2003), who focused on intrusions that were best preserved and best understood geologically.
Size and shape of intrusions
The ultramafic intrusions are generally lensoid to tabular in shape and are as large as ˜2 km in width and ˜10 km in length. They are always oriented subparallel to the regional foliation. There is no discernible correlation between the size of the ultramafic intrusions and their level of emplacement or amount of mineralization. Some of the largest intrusions occur in the basement (Moak Lake), Pipe Formation (North Manasan), and Setting Formation (North Mystery Lake), as well as at the Thompson and/or Pipe (Lower Ospwagan Lake West) and Pipe and/or Setting (Mystery Lake) contacts. However, the largest ultramafic intrusions are either unmineralized (e.g., North Mystery Lake, Upper Ospwagan Lake) or only low grade (e.g., Ospwagan Lake deposit), suggesting that mineralization is not related to the size of the intrusion.
Although the intrusions almost always have tectonized contacts with the country rocks, deformation is commonly restricted to the outer margins of the intrusions. The degree of preservation of magmatic fabrics increases inward toward the cores of the intrusions. Typically, pegmatites are found along the margins of the ultramafic intrusions that obscure the original intrusive contacts. In rare cases, however, original contacts have been observed. In these instances, the adjacent sedimentary rocks show evidence of thermal metamorphism that include silicification and hornfels as much as 5 to10 m from the contact (Burnham et al., 2003). The very narrow thermal aureoles (<10 m), in comparision to the moderate to large sizes of the ultramafic bodies (tens to hundreds of of meters in thickness) and the very high temperatures (1,400°C or more) of the ultramafic magma from which they formed (see below), suggest that the sedimentary rocks were lithified during contact metamorphism (e.g., Menard et al., 1996).
Composition and internal structure
The ultramafic bodies in the Thompson nickel belt vary widely in composition and degree of differentiation, including: (1) thick undifferentiated metadunites with thin lower and upper pyroxenitic margins, some of which are mineralized (e.g., Pipe intrusion) and some of which are unmineralized (e.g., Brostrom Lake, Hambone Lake East); (2) thick undifferentiated metaperidotites with thin lower and upper pyroxenitic margins, some of which are mineralized (e.g., Soab Lake) and some of which are unmineralized (e.g., Spur South, Rabbit Point), and thin undifferentiated amphibolites, all of which are unmineralized (e.g., Grass River Lineament); (3) moderately differentiated metadunites with thin lower pyroxenitic margins, thick lower dunitic zones, and thick upper pyroxenitic zones, some of which are mineralized (e.g., Pipe 1) and some of which are weakly to unmineralized (e.g., North Manasan), and (4) moderately differentiated metaperidotites with thin lower pyroxenitic margins, thick peridotitic zones, and thick upper gabbroic zones, all of which are unmineralized (e.g., Halfway Lake).
These variations can be interpreted to reflect variations in the degrees of olivine-(chromite) accumulation and differentiation after ponding (Lesher et al. 1984; Arndt et al., 2008). The large amount of olivine accumulation and low to moderate degree of differentiation suggest that most of the intrusions represent conduits formed in a relatively dynamic magmatic environment, and that the magmas ponded relatively late in their formation. These conditions are conducive for incorporation of sulfur from country rocks and efficient interaction between magma and sulfide (e.g., Lesher et al., 2001).
The differentiation in the least deformed and least metamorphosed intrusions (e.g., Pipe 1, Spur South, W-56 South body at William Lake; Fig. 2) is defined by systematic variations in cumulus olivine content and whole-rock MgO content (Fig. 6), which increase upward away from the lower margins and decrease upward through the cumulate zones. Similar asymmetric trends have been observed in mineralized bodies in less deformed areas (e.g., Lesher et al., 1984; Arndt et al., 2008; Lesher and Barnes, 2009). In areas of lower strain and lower metamorphic grade in the Thompson nickel belt, where younging directions are known for the enclosing metasedimentary rocks, they are consistent with those inferred from the petrological and geochemical variations within the intrusions, indicating that most of the intrusions were concordant to semiconcordant with the surrounding units during emplacement.
Age and source of intrusions
Scoates et al. (2010) obtained a U-Pb zircon age of 1880.8 ± 2.5 Ma for a metaperidotite adjacent to the Pipe II open pit, similar to the 1880 ± 5 Ma age obtained by Hulbert et al. (2005) for a mineralized ultramafic body in the Setting Lake area. Similar ages were obtained by Heaman et al. (1986) for the Molson dike swarm (Cross Lake dike: ; Cuthbert Lake dike: 1883 ± 2 Ma) and for the Fox River sill (; Fig. 1). Hulbert et al. (2005) also obtained an Re-Os isochon age of 1885 ± 49 Ma from 23 mineralized samples from seven deposits that span the length of the Thompson nickel belt, which is identical within error to the U-Pb age. The γOs values for individual samples range between 1.6 and 13.6, with no correlation with sulfide content or ultramafic or pelite host rock, consistent with derivation from undepleted mantle with variable degrees of contamination by continental crust and/or sedimentary rocks derived from continental crust. Great care must normally be placed in the use of Re-Os systematics in metamorphic systems, because Re and/or Os may be easily mobilized, and Os is a trace element in most magmas but an ultratrace element in most crustal rocks, so it can be easily reset in dynamic magmatic systems with high magma/sulfide ratios (Lesher and Stone, 1996; Lambert et al., 2000; Lesher and Burnham, 2001).
Mode of emplacement
The ultramafic intrusions in the Thompson nickel belt have been interpreted to represent a series of sills and low-angle dikes (Bleeker, 1990a; Layton-Matthews et al., 2007) based on the following evidence: (1) they occur at multiple stratigraphic levels in the Ospwagan Group; (2) they do not show any evidence for an extrusive origin, such as quench textures, volcaniclastic textures, pillow structures, or interflow sediments; (3) they are hundreds of meters thick and exhibit coarse textures consistent with slow cooling; (4) only thin chill zones occur at the top of the units and the upper contacts are not associated with any particular sedimentary rock type; and (5) in at least one location, they crosscut stratigraphy (e.g., on the eastern side of Ospwagan Lake).
The ability of a magma to rise up through the crust and form sills or flows is controlled by a number of separate, but interrelated factors, including magma density, which varies as a function of composition, temperature, and pressure; sedimentary rock density, which varies as a function of mineral composition, texture, and degree of consolidation and/or lithification; and sedimentary rock rheology, which will vary as a function of lithology and lithification. Although the precise location of the sills will be determined by rheology, the overall driving force behind magma ascent and emplacement is the buoyancy of the magma relative to the surrounding country rocks (e.g., Walker, 1989).
The emplacement of the ultramafic intrusions was modeled by Burnham et al. (2003), using the stratigraphy established by Bleeker (1990a) and published densities for consolidated sedimentary rocks and gneisses (Table 3). Magma densities were calculated as a function of temperature, pressure, and composition for three potential mafic and ultramafic magmas, using an empirical equation of state and constants for oxide components taken from Spera (2000). Two models were considered.
The first model (Fig. 7A) assumed that the Ospwagan Group sediments were unconsolidated and contained ˜10 percent trapped water. Despite the presence of denser iron formations in the sequence, the bulk densities of the sediments would have ranged between only 2.53 and 2.68 g cm–3, which is considerably less than would have been required for the emplacement of mafic or ultramafic magmas with densities of between 2.72 and 2.74 g cm–3, respectively. Because of the low densities of the sediments, the bulk density of the crust would have not exceeded that of the magmas for as much as 500 m into the basement, which was assumed to have a density of ˜ 2.85 g cm–3, at the upper end of the 2.52 to 2.81 g cm–3 range expected for granitic rocks (Clark, 1966), but within the 2.61 to 2.84 and 2.63 to 3.10 g cm–3 ranges expected for gneisses and granulite-facies rocks (Clarke, 1966). This model indicates that the ultramafic sills could not have been emplaced into the Ospwagan Group sediments if they still contained appreciable water and confirms the interpretations above that the sediments must have been lithified prior to sill emplacement.
The second model (Fig. 7B) assumed similar stratigraphic thicknesses as in the first model, but with only 0.5 to 1.0 percent pore water in the sediments. Although very dependent on the thicknesses and densities chosen for the different units, under such conditions, the densities of the magmas match that of the overlying bulk sediment at a point between the top of the Thompson Formation and the first iron formations of the Pipe Formation. This is consistent with the locations of the majority of the ultramafic bodies. Intrusion at this level would, therefore, be expected because of the pronounced decrease in bulk density that should occur at this contact. However, in areas where the high-density carbonate rocks of the Thompson Formation were thinner, the density of the magmas could have exceeded the bulk density of the sedimentary rocks at greater depths, resulting in deeper emplacement, possibly in the basement. Similarly, where the iron formations of the upper Pipe Formation were thicker, the bulk density of the sediments could have been greater and the magmas would have been emplaced higher in the sequence. This model suggests that prevalent emplacement of the sills into the lower, sulfide mineral-bearing units of the Pipe Formation may, therefore, be restricted to regions in which the iron formations in the upper units of the Pipe Formation are either thicker or more abundant.
These models account for both the prevalence of ultramafic sills within rocks of the upper Thompson and/or Pipe Formations, the presence of gabbroic sills similar to those described in the upper Pipe or Setting Formations on Mystery Lake (Burnham et al., 2003), and the presence of a few large ultramafic sills in the upper parts of the Ospwagan Group stratigraphy. The models also predict that there could have been more mafic intrusive activity in the upper parts of the Ospwagan Group than is currently observed, particularly above some of the larger ultramafic sills.
The intrusion of sills within or at the upper contacts of denser layers of the stratigraphy has great significance for the genesis of the magmatic Ni-Cu-(PGE) ores because the majority of the sulfide-facies iron formations and sulfide mineral-rich layers of the Pipe Formation are closely associated with dense silicate-facies iron formations, which would have acted as density filters for ascending magmas. If the ultramafic intrusions were preferentially emplaced along these horizons, then they would have been predisposed to interaction with the sulfide-bearing units, leading to the interaction of the magmas with an external source of sulfur and the production of sulfide ores.
Mineralogy, petrography, and alteration
The ultramafic intrusions in the Thompson nickel belt have experienced deformation and metamorphism under upper amphibolite-facies conditions and serpentinization and potassic and carbonate alteration that resulted from the infiltration of volatile-rich fluids from metasedimentary country rocks (Layton-Matthews et al., 2007). Primary igneous textures and relict magmatic minerals are commonly preserved in the cores of the intrusions, but almost all igneous textures and minerals have been destroyed along the deformed margins of the bodies. Nevertheless, even the most altered rocks can be subdivided on the basis of metamorphic mineralogy and whole-rock geochemistry into three principal lithologic units: metadunite, metaperidotite (including harzburgite and rare wehrlite), and metapyroxenite (both orthopyroxenite and clinopyroxenite). Many of the ultramafic intrusions appear to have preserved a primary igneous zoning, where olivine modal abundances vary systematically through individual drill core intersections (Fig. 6).
Most of the metadunites in the Thompson nickel belt are serpentinized olivine adcumulate rocks with >90 percent fine- to medium-grained serpentine pseudomorphs after olivine in a fine-grained matrix of amphibole ± chlorite ± serpentine ± chromite. Despite pervasive alteration to mesh-textured serpentine, altered olivine grains commonly preserve elements of their original texture, including grain outlines and irregular fracture patterns. Olivine grains are generally equant to elongate, with either curvilinear grain boundaries that meet in triple point junctions and define a close-packed mosaic texture (Fig. 8A) or with ovoid, occasionally idiomorphic shapes (Fig. 8B). Olivine grain sizes vary between 0.2 and 10 mm but are generally ˜1 mm. Some samples exhibit a weak orientation or flattening of olivine grains. Magnetite defining the rims of altered olivine grains indicates that the dunites initially contained <10 percent intergranular material.
In addition to the fine-grained magnetite generated during serpentinization of olivine, the most altered dunites commonly contain stockworks of magnetite and/or carbonate veins. These veins indicate that several elements may have been locally mobile during alteration, including Fe, Mn, Ca, Sr, and possibly Mg. Although it is difficult to quantify because of the large amount of secondary magnetite formed during serpentinization of olivine, most of the dunites appear to have originally contained minor amounts of fine-grained euhedral intercumulus chromite (Fig. 8C,D).
Fresh dunite cores occur in the central parts of many ultramafic intrusions, including the unmineralized Brostrom Lake, Hambone Lake East, and Mystery Lake North intrusions, the weakly mineralized W-56 ultramafic body in the William Lake area (>300 m of unaltered dunite), the weakly mineralized Mid Lake and North Manasan intrusions, and the strongly mineralized Pipe 1 body, and a number of the larger intrusions in the T3 mine (Figs. 1, 2). In all of these intrusions, the dunites are composed of >95 percent olivine, with almost complete contact between the olivine grains and relatively little intergranular material, suggesting that the lack of serpentinization in the cores of the ultramafic intrusions can be attributed to the lower permeability of adcumulate rocks. Unaltered dunites commonly contain fine-grained disseminated chromite (1−2%, 60−100 μm), which may occur both as subhedral to anhedral inclusions within the olivine grains (Fig. 8C) and as euhedral intercumulus grains (Fig. 8D). The presence of chromite, both within and interstitial to olivine, indicates that olivine and chromite were both on the liquidus during the crystallization of the dunites, thus implicating a komatiitic basaltic to low Mg komatiitic magma (Murck and Campbell, 1986; Lesher and Stone, 1996). In addition to chromite inclusions, many relict olivine grains within dunite contain finely disseminated chromite microinclusions or inclusion trails (Fig. 8E) and 1 to 2 percent fine-grained intergranular blebs of predominantly disseminated pentlandite, pyrrhotite, and chalcopyrite (Fig. 8F).
Most of the metaperidotites in the Thompson nickel belt are composed of 40 to 90 percent fine- to medium-grained, variably serpentinized subhedral olivine grains in a matrix of fine-grained chlorite ± calcic amphibole ± serpentine, with trace amounts of fine-grained chromite (Fig. 8A). The chlorite ± amphibole ± serpentine matrix may represent altered komatiitic melt and/or intercumulus pyroxene, but the presence of calcic amphibole in these intergrowths suggests that the pyroxene was originally Ca-rich clinopyroxene, rather than Ca-poor orthopyroxene, and that the metaperidotites were, therefore, originally wehrlites, rather than harzburgites or lherzolites. Although komatiitic olivine crystallized in rapidly cooled extrusive environments may contain significant amounts of Ca (Lesher, 1989), the unaltered olivine in the ultramafic intrusions in the Thompson nickel belt has low Ca contents and appears to have crystallized in a slowly cooled intrusive environment. Because few unaltered peridotites were observed within the ultramafic intrusions, most were distinguished from metadunites by the presence of greater amounts of interstitial chlorite and amphibole and/or by lower whole-rock MgO contents and higher incompatible major element contents.
Although orthopyroxene-bearing olivine cumulate rocks have been reported in the ultramafic intrusions of the Thompson nickel belt by several previous workers, predominantly at the Thompson deposit (Peredery, 1982; Paktunc, 1984), few harzburgites or lherzolites were identified either petrographically or geochemically in the ultramafic intrusions that were studied by Burnham et al. (2003). Those that they found were generally poikilitic olivine mesocumulate rocks in which the orthopyroxene formed oikocrystic grains enclosing cumulus olivine. The paucity of orthopyroxene is understandable considering that our sampling was concentrated on unmineralized and subeconomically mineralized intrusions, as well as with the generally lower metamorphic grade of the ultramafic intrusions that were studied. The ultramafic intrusions at the Thompson deposit experienced upper amphibolite to granulite-facies metamorphism, during which the assemblage olivine ± talc ± anthophyllite or chlorite was unstable and reacted to form either orthopyroxene or orthopyroxene ± olivine ± spinel, respectively (Paktunc, 1984). The majority of the ultramafic intrusions sampled outside of the Thompson deposit area experienced only middle- to upper amphibolitefacies metamorphism, during which olivine ± talc, olivine ± anthophyllite, or chlorite are stable phases.
Partially serpentinized peridotites, in which the cores of olivine grains are partially preserved, were found in only few ultramafic intrusions (e.g., North Manasan and Spur South: Fig. 8B). These rocks are characterized by <90 percent fine-to medium-grained, partially serpentinized olivine in a matrix of >10 percent fine-grained chlorite ± amphibole ± serpentine. The greater abundance of chlorite compared to amphibole suggests that the matrix represents altered komattitic-trapped liquid rather than pyroxene. Relict igneous olivine is ovoid, between 0.1 and 3 mm in diameter, and contains opaque microinclusions similar to those in the dunites. However, the olivines in metaperidotite are less magnesian (Fo86) than those in the metadunites (Fo88-90; for additional data, contact first author).
Olivine metapyroxenites occur along the margins, and as discrete and semicontinuious layers, within the ultramafic intrusions. They are composed primarily of fine- to medium-grained olivine and/or pyroxene altered to either mesh-textured serpentine or porphyroblastic calcic amphibole in a matrix of acicular chlorite ± amphibole ± serpentine. Biotite is only present in the matrix in metapyroxenites near ultramafic– metasedimentary rock contacts and contains very fine grained inclusions of zircon, suggesting assimilation of sedimentary rock during intrusion emplacement and/or metamorphic mobilization of Zr into the pyroxenites. In most samples, the olivine is completely serpentinized. However, relict olivine fragments are preserved in olivine pyroxenites along the margins of the Spur South ultramafic body (Fig. 8C). The low magnesium contents (Fo81–84) of the olivine suggest that they crystallized from significantly less magnesian magmas than the olivine in the metaperidotite and dunite parts of the ultramafic intrusions.
Although pure orthopyroxenite and clinopyroxenite layers are preserved in a few of the ultramafic intrusions (e.g., Ospwagan Lake, Thompson mine, Spur South, and South Thompson nickel belt: Fig. 8D), such layers are apparently rare in the Thompson nickel belt. In most cases, the pyroxenites have been altered to an assemblage of either hornblende ± plagioclase (clinopyroxenite) or tremolite ± anthophyllite ± orthopyroxene (orthopyroxenites) and can only be easily recognized from their whole-rock chemical compositions.
Almost all of the ultramafic rocks in the Thompson nickel belt have been pervasively hydrated, carbonate altered, and/or K metasomatized, resulting in significant changes in their mineral assemblages (Table 4). Alteration and metasomatism occurred on both regional and local scales. Whereas regional alteration and metasomatism occurred in ultramafic bodies throughout the Thompson nickel belt, evidence for more local alteration and metasomatism is located adjacent to faults or the margins of pegmatite veins located within the intrusions or along their contacts.
Three styles of serpentinization have been identified from petrographic examination of the ultramafic bodies: 91) pseudomorphic pervasive serpentinization, (2) nonpseudomorphic pervasive serpentinization, and (3) fracture-controlled serpentinization. In many cases, all three styles occur in the same ultramafic body, with a transition from partial serpentinization, through complete pseudomorphic serpentinization, and to complete nonpseudomorphic serpentinization. Although the relative proportions of the different alteration phases vary with the original proportions of olivine, pyroxene, and trapped melt, the degree of serpentinization does not appear to exhibit any systematic correlation with the size, location, or major element composition of the ultramafic body (Layton-Matthews et al., 2007).
Potassic alteration is generally expressed as a modal increase in phlogopite in Mg-rich rocks or biotite in more Aland Fe-rich rocks, which may be present in the form of either subhedral porphyroblasts that cut existing serpentinization fabrics or fine (<5 μm) biotite-serpentine intergrowths in antigorite porphyroblasts. Phlogopite is present only in serpentinized rocks and is absent in rocks that contain little or no serpentine, indicating that it is not a relict igneous phase but is related to alkali metasomatism.
Carbonate alteration is the youngest alteration style in the bodies and, unlike potassic alteration, is unrelated to the degree of serpentinization. The dominant carbonate phase is magnesite, although minor amounts of calcite may also be present. Calcite and magnesite cannot coexist in an equilibrium assemblage (e.g., Anovitz and Essene, 1987) , so it is likely that calcite represents a later alteration phase. At the Pipe pit, late cream-colored magnesite veins crosscut all primary and secondary fabrics, including the late chrysotile veining. In unaltered to partially altered ultramafic lithologic units, magnesite commonly forms anhedral grains that crosscut primary olivine and secondary α-lizardite. Talc is commonly associated with magnesite and forms a feathery textured core to many of the anhedral magnesite grains.
Olivine mineral chemistry
In order to interpret the petrogenesis and sulfide saturation history of the intrusions, the compositions of relict olivine and chromite were determined by wavelength-dispersive X-ray emission spectrometry, using a CAMECA SX-50 electron probe microanalyzer at the Ontario Geoscience Laboratories in Sudbury or at the Geological Survey of Canada in Ottawa. Olivine grains adjacent to sulfide grains were avoided during analysis, because of the susceptibility of Fe-Ni exchange during postmagmatic cooling or metamorphism (e.g., Binns and Groves, 1976).
The compositions of relict olivine in the ultramafic intrusions exhibit a wide range of forsterite contents (Fo81.5−93.5: Fig. 9) and display a trimodal population with peaks at ˜Fo91, ˜Fo89.5, and ˜Fo83. The compositions of olivine within individual samples and thin sections generally show a more limited range of compositions (e.g., Fo90.5–92 in the W-56 ultramafic body at William Lake, Fo92–93 at Brostrom Lake, Fo81.5–84 at Spur South, and Fo88–90 at Pipe 1). The Mn and Ni contents of olivine are shown as a function of forsterite content in Figure 9 and individual analyses are presented in Appendix 1. Nickel concentrations range from 500 to 4,500 ppm for 282 samples from the southern Thompson nickel belt.
The compositions of most olivine samples in the unmineralized and subeconomic Mystery Lake and North Manasan ultramafic intrusions, the upper parts of two ultramafic intrusions at Thompson deposit, are consistent with those expected for the fractional crystallization of olivine from a sulfide-undersaturated magma with an initial major element composition similar to that of the most mafic Bah Lake Formation volcanic rocks. However, the olivine in the Brostrom Lake, Pipe 1, and Mid Lake ultramafic intrusions, one sample from the Mystery Lake ultramafic intrusion, and most samples from Soab, Birchtree, and the Thompson deposits are variably depleted in Ni, whereas those from the W-56 ultramafic intrusion at William Lake are significantly enriched in Ni relative to what would be expected for sulfide-undersaturated fractional crystallization of olivine. Because the Ni-depleted olivine grains all occur within sulfide-bearing cumulate rocks, the Ni depletion is inferred to reflect equilibration of olivine with sulfide during emplacement or crystallization from a magma that had previously equilibrated with sulfides. The anomalously high Ni contents of the olivine from William Lake occur in sulfide-poor cumulate rocks and require magma with an anomalously high Ni content (5,000 ppm), perhaps one that has assimilated small amounts of Ni from pre-existing sulfides prior to emplacement.
Whole-Rock Geochemistry of Intrusions
The whole-rock geochemistry and petrogenesis of the intrusions in the Thompson nickel belt have been studied by Lesher et al. (2001), Burnham et al. (2003), and Layton-Matthews et al. (2003). Whole-rock geochemical data used in the following sections can be obtained from the first author.
Layton-Matthews et al. (2003) have shown that Cs, Rb, Na, K, Ba, Sr, Ca, Cu, Pt, and Au were mobile, to varying degrees, during serpentinization, K metasomatism, and talc-carbonate alteration of the ultramafic rocks in the Thompson nickel belt, and that also La and possibly Ce were mobile during the most pervasive serpentinization. However, some major elements (Si, Al, Fe, Mg, Mn), many minor elements (Ti, Cr, V, Ni, Co, Pd, Rh, Ru, Ir), and most high field-strength elements (U, Th, Nb, Ta, MREE, HREE, Zr, Y) were relatively immobile. Therefore, sufficient geochemical information remains in the ultramafic rocks to decipher their magmatic and metallogenic histories.
The whole-rock major and selected minor element variations of 806 samples from mineralized ultramafic intrusions, 272 samples from weakly mineralized intrusions, 396 samples from unmineralized intrusions, and 61 samples from mafic dikes from all parts of the Thompson nickel belt were studied by Burnham et al. (2003) and are shown in Figure 10. Also plotted are the calculated compositions of magmas and cumulate rocks produced by fractional crystallization of a likely parental magma, with and without the assimilation of sedimentary rocks using the MELTS program (Ghiorso and Sack, 1995). The majority of the major element trends reflect fractionation crystallization and/or accumulation of olivine and, in the case of several pyroxenites from Ospwagan Lake and Thompson, pyroxene.
Olivine-rich samples may be subdivided, on the basis of their Cr contents, into Cr-rich cumulates, intermediate Cr cumulates, and Cr-poor cumulates.
Cr-rich cumulates: Cr-rich cumulates contain ˜1,800 to 10,680 ppm Cr and exhibit a positive correlation between Cr and Mg, defining a trend between ˜1,800 ppm at 16 percent MgO, which is the composition of the mafic rocks of the Bah Lake Formation, and 6,670 ppm Cr at 37 percent MgO (Fig 10). Similar rocks in other areas have been interpreted to represent fractionation and accumulation of olivine and magnesiochromite in roughly cotectic proportions of ˜50/1 (Lesher and Stone, 1996; Barnes and Brand, 1999) from a chromite-saturated komatiitic basalt magma.
Intermediate Cr cumulates: Intermediate Cr cumulates contain 3,000 to 6,000 ppm Cr and plot as a cluster between the trends of the Cr-rich and Cr-poor cumulate rocks (Fig 10). Barnes and Brand (1999) interpreted similar rocks as olivine-rich cumulates derived from chromite-undersaturated magma in which chromite had been physically added. Lesher and Stone (1996), however, interpreted them as cumulates derived from a magma that reached chromite saturation during accumulation and, therefore, as rocks that contained an early-crystallized component of olivine and a later crystallized component of olivine + chromite.
Cr-poor cumulates: Cr-poor cumulates contain 700 to 1,800 ppm Cr and exhibit a negative correlation between Cr and Mg, defining a trend between ˜2,500 to 3,000 ppm Cr at 20 to 22 wt percent MgO, which is the composition of the inferred parental magma, and 1,500 to 2,000 ppm Cr at 50 percent MgO, which is the measured concentration of Cr in olivine. Similar rocks in other areas have been interpreted to represent fractionation and accumulation of olivine from chromite-undersaturated komatiitic magmas (Lesher and Stone, 1996; Barnes and Brand, 1999).
Although there are considerable overlaps in the Cr contents of different ultramafic rocks in the Thompson nickel belt, Cr-rich cumulates derived from komatiitic basaltic magmas are more abundant in unmineralized or weakly mineralized intrusions, whereas Cr-poor cumulates derived from komatiitic magmas are more abundant in mineralized intrusions. Although the primary control on chromite solubility in most ultramafic magmas is interpreted to be the temperature and MgO content of the melt, with most magmas reaching chromite saturation at ˜1,400°C and ˜20 wt percent MgO (Lesher and Stone, 1996), variations in the olivine/ chromite ratios of ultramafic cumulate may also occur in response to changes in one or more of the different parameters that control the solubility of chromite in ultramafic magmas (Barnes and Roeder, 2001). Experimental studies (e.g., Murck and Campbell, 1986) have shown that redox conditions can influence chromite solubility such that high Mg komatiites may become saturated in chromite earlier under oxidizing conditions and low Mg komatiites may remain undersaturated in chromite later under reducing conditions.
The abundance of Cr in cumulate rocks in the central Thompson nickel belt correlates negatively with the average Fo contents of the cumulate olivine, implying a relationship between low Mg contents, chromite-saturation, and the absence of mineralization. However, the similarity between the measured and estimated compositions of the olivine from both mineralized (e.g., Pipe, Thompson, Birchtree, and Ospwagan Lake) and unmineralized ultramafic intrusions (e.g., Spur South, Mystery Lake North and East, and Hambone East) in the northern Thompson nickel belt indicates that the presence or absence of mineralization in that region is not related to the Mg content of the magma. In addition, the link between the behavior of chromite and the mineralization status of the intrusions may represent changes in the oxidation state of the magma, perhaps caused by crustal contamination processes that are unrelated to mineralization processes, such as assimilation of S-free rocks.
Nickel, cobalt, copper, and zinc
The orthopyroxenite and clinopyroxenite rocks at Ospwagan Lake and the Thompson deposit have lower than normal Ni and Co contents in their silicate components (Fig. 10), which resulted from the accumulation of Ni- and Co-poor orthopyroxene and clinopyroxene rather than Ni-rich olivine. The majority of the sulfide-poor (<0.3 wt %) mafic and ultramafic samples from the northern region of the Thompson nickel belt, however, plot within the range expected for mixtures between variably fractionated sulfide-undersaturated ultramafic magmas. These magmas have an initial composition similar to the most magnesian picrites and komatiitic basalts of the Bah Lake Formation and calculated equilibrium olivine compositions (Fig. 10). Such a variation indicates that most of the cumulate and mafic rocks formed from magmas that were not significantly depleted in Ni. Exceptions include many of the chromite-poor olivine adcumulate rocks collected from the Mid Lake and Mystery Lake ultramafic intrusions and most of the dunites, and analyzed olivines, in the Brostrom Lake ultramafic body. These samples plot significantly below the trend expected for sulfide-undersaturated olivine cumulates. Although this suggests that the magmas from which these rocks formed might have been depleted in chalcophile metals, most of these samples have very low sulfur contents, indicating that the magmas from which they formed were sulfide undersaturated at the time of emplacement. Possible explanations for this include the rocks may have formed from magmas that were previously depleted in Ni, some of the sulfides may have been dissolved, and/or Ni and S may have been mobile during metamorphism. There are presently insufficient data available to determine which process has been most important, although the presence of minor Ni-undepleted samples in the same intrusions suggests that the second or third explanations are more likely.
When plotted as a function of MgO content (Fig. 10), the Ni contents of unmineralized ultramafic intrusions define an array along a tie line between the inferred parental magma composition and the compositions of olivine calculated to be in equilibrium with it. This is also consistent with the accumulation of olivine from a sulfide-undersaturated ultramafic magma. However, the weakly mineralized intrusions located in the William Lake area are characterized by anomalously high Ni contents at similar MgO compared to the other intrusions, which mirror the high Ni contents determined in their constituent olivine. The exclusion of samples containing more than trace amounts of sulfide minerals indicates that the contrast between the two groups of intrusions cannot be accounted for by variations in the abundance of disseminated sulfides between the two groups of cumulates. Rather, it supports the interpretation that the magma from which these intrusions formed was either anomalously Ni rich or that Ni was more efficiently removed from the magma by the olivine that makes up the William Lake ultramafic intrusions. This may suggest that the southern Thompson nickel belt (STNB) magmas (i.e., William Lake intrusions) represent a more dynamic and primitive part of the system or that the magmas are derived from a unique source.
The reason for the greater depletion of many Thompson nickel belt samples in Co, especially many of those with high Mg contents, is not clear, but may reflect lower compatibility of Co in metamorphic serpentine-magnetite and therefore greater mobility. Liwanag (2000) showed that the Ni/Co ratios of pentlandites in disseminated sulfides in ultramafic rocks increased with increasing deformation and/or metamorphic grade, also suggesting that Co is more mobile than Ni during metamorphism and/or deformation.
The Cu contents of the ultramafic rocks scatter considerably, because Cu is soluble in most metamorphic-hydrothermal fluids and because it is concentrated almost entirely in late-magmatic base metal sulfides that are very susceptible to modification. However, the trend of variable Cu content in gabbros and amphibolites may also have variable degrees of depletion associated with sulfide saturation in magmas of that composition (see discussion by Lesher and Stone,1996). The Zn contents of the ultramafic rocks also scatter considerably for the same reason, but the Zn contents of the gabbros and amphibolites are less variable than Cu, suggesting that it was less mobile in those rocks and/or that Zn did not partition as strongly into sulfides as Cu. The overall trend of decreasing Zn with increasing Mg is consistent with its incompatibility in olivine.
Platinum group elements
Lesher et al. (2001) have shown that most of the ultramafic rocks in the Thompson nickel belt, regardless of whether mineralized or unmineralized, are variably enriched or depleted in PGE relative to the amounts expected from olivine fractionation or accumulation. As for Ni, the depletion in PGE in pyroxenites from Ospwagan Lake and Thompson can be attributed to accumulation of PGE-poor pyroxene, but the depletions and enrichments in the olivine-rich rocks are interpreted to reflect segregation and accumulation of sulfides, respectively. The lack of any differences between mineralized and unmineralized units can be attributed to the achievement of sulfide saturation even in unmineralized units and the very strong partitioning behavior of PGE into sulfide (e.g., Crocket, 2002). Although the PGE cannot be used to discriminate between mineralized and unmineralized intrusions within mineralized belts, in the Thompson nickel belt or elsewhere (Lesher et al., 2001), the PGE do provide a powerful tool in distinguishing between mineralized and unmineralized belts (Fiorentini et al., 2010).
Incompatible lithophile elements
Assimilation of continental crust, or sedimentary rocks derived from it, by an ultramafic magma may lead to only subtle changes in its major element composition and those of the cumulate rocks that form from it. However, the same contaminant may significantly affect the trace element chemistry of the magma, because there are more pronounced differences between the trace element signatures of the two components.
Many mantle-derived ultramafic magmas, including normal MORB and many plume-derived basalts, picrites, and komatiites, are strongly depleted in highly incompatible lithophile elements (HILE: U-Th-Nb-Ta-LREE) relative to the abundances of moderately incompatible lithophile elements (MILE: Zr-MREE-Ti-HREE). Most upper continental crustal rocks, on the other hand, are strongly enriched in U-Th-REE relative to Nb-Ta-(Ti), resulting in significantly different Nb/Th, Th/Zr, and Ti/Zr ratios (Lesher, 1989). Because the abundances of incompatible elements are considerably greater in the crustal component than in the magma, contamination of komatiitic magma by even small amounts of crustal material will introduce a recognizable crustal signature. It will be characterized by variable enrichment in HILE with negative Ta-Nb-(Ti) anomalies. Because it appears that La, and to a lesser extent Ce, have been mobile in the ultramafic rocks of the Thompson nickel belt under certain conditions (Layton-Matthews et al., 2003), Th, Nb, Yb, and Ti provide the best discrimination between mantle source enrichment, variable degrees of melt extraction, and crustal contamination.
Mantle-normalized Ce/Sm and Th/Nb ratios of the Thompson nickel belt ultramafic intrusions are plotted in Figure 11. Almost all of the ultramafic intrusions in the northern region of the Thompson nickel belt are highly enriched in LREE ([Ce/Sm]mn = 1.2–10) and depleted in Nb relative to Th ([Th/Nb]mn = 2–10), consistent with contamination of the parental magmas by Ospwagan Group sedimentary rocks. In contrast, the majority of the mafic and ultramafic volcanic rocks in the areas surrounding Bah Lake, Ospwagan Lake, and other locations within the northern region of the Thompson nickel belt exhibit trace element ratios that resemble those of the depleted mantle and plot in a distinctly different field.
The contrast between the two groups of rocks is highlighted when the Th/Nb and Ce/Sm ratios are plotted as a function of MgO (Fig. 11). With the exception of a small group of basalts in the Mystery Lake and Setting Lake areas, the basalts and picrites collected from drill cores in the Mystery Lake, Ospwagan Lake, and Pipe areas, and from surface exposures in the northern region of the Thompson nickel belt, are either depleted or only marginally enriched in Th and LREE relative to Nb and MREE. In contrast, the associated ultramafic rocks are almost all strongly enriched in highly incompatible trace elements and have negative Nb anomalies, irrespective of their mineralization status. In each case, the trace element ratios are similar for all MgO contents within the group and there are considerable overlaps in MgO contents between the two groups. Because the initial incompatible element contents of both the volcanic and intrusive mafic and ultramafic rocks may be expected to decrease with increasing MgO contents, the absence of any recognizable relationship between incompatible trace element enrichment and MgO content in either the mafic or ultramafic rocks is inferred not to have resulted from the greater susceptibility of the ultramafic rocks to alteration. Rather, it represents compositional differences that arose from different degrees of crustal assimilation during emplacement.
Although many of the mafic and ultramafic rocks in the central region of the Thompson nickel belt are enriched in incompatible elements relative to either the primitive or depleted mantle reservoirs (Fig. 11), the degree of enrichment is considerably less than that observed in the northern region of the Thompson nickel belt. It does not correlate with observed variations in either the major element chemistry or the mineralization status of the ultramafic intrusions.
Most of the mafic and ultramafic rocks from different areas of the Thompson nickel belt exhibit trends of increasing Th/Ti ratio with only a small change in Nb/Ti ratio (Fig. 12A-C) and increasing Th/Yb with increasing Nb/Yb (Fig. 12D, E). When rocks from different areas are compared, most of the samples from each area appear to be characterized by primary magmatic trends that originate from an N-MORB−OIB array (Pearce, 2008) and project toward the field of the Pipe iron formation (P2 member) and basement gneisses from the Thompson area. This is consistent with the assimilation of crustal material by discrete batches of magma derived from variably depleted or enriched mantle sources.
On this basis, it is possible to recognize at least four different groups of magmas in the Thompson nickel belt:
Moderately to strongly contaminated mafic and ultramafic magmas derived from a strongly depleted mantle source and contaminated with sedimentary rocks similar to those of the Ospwagan Group (e.g., Thompson, Birchtree, Halfway Lake, North Manasan, Pipe 1 and 2, and Spur South ultramafic intrusions, and rare basalts and picrites from Ospwagan Lake).
Slightly to moderately contaminated basalts and komatiitic basalts derived from an undepleted or slightly enriched mantle source and contaminated with sedimentary rocks similar to those of the Ospwagan Group (e.g., Ospwagan Lake and Pipe Pit).
Variably contaminated basalts and gabbros derived from an enriched source and contaminated with sedimentary rocks similar to those of either the Ospwagan or Sickle and/or Grass River Groups (mostly areas north and west of Setting Lake).
Highly enriched basaltic or pyroxenitic dikes (e.g., Ospwagan Lake, Mystery Lake).
Ore types and textures
The ores in the Thompson nickel belt can be subdivided into several types (Lesher and Keays, 2002). These include stratiform massive to semimassive sulfides at or near the base of the ultramafic bodies (type I), strata-bound disseminated and net-textured sulfides within the ultramafic bodies (type II), Ni-enriched sedimentary rock-hosted sulfides (type IV), and metamorphically and tectonically mobilized type I or IV sulfides (type V). No stratiform reef-style mineralization (type III) has been reported in the Thompson nickel belt.
Although layered sedimentary rock-hosted ores and breccia ores are more common in the Thompson nickel belt than in typical Ni-Cu-(PGE) districts, the textures of these types of ores are broadly similar to those observed in many other deposits with this ore type. The disseminated and net textures of the mineralization in the ultramafic rocks indicate that these ores are magmatic. The restriction of sedimentary rock-hosted mineralization to areas within 20 m of ultramafic rocks, together with their geochemical characteristics (see below), suggest that they formed by mobilization of metals from the intrusions into the sediments (Bleeker, 1990a). The suite of anomalous ore elements and the presence of similar mineralization in lower grade, less deformed environments at Kambalda, Western Australia (Lesher and Keays, 1984; Paterson et al., 1984) and Langmuir, Ontario (Green and Naldrett, 1981) suggest the involvement of magmatic diffusion rather than tectonic or metamorphic mobilization, which is typical of most type V ores (Lesher and Keays, 2002). The presence of inclusions of ultramafic rocks in many breccia ores indicates that they formed from ultramafic rock-hosted ores, whereas the presence of inclusions of only country rocks in other breccia ores indicates that they formed from sedimentary rock-hosted ores. Together, these observations suggest that most or all of the mineralization in the Thompson nickel belt was magmatic and variably mobilized into the country rocks during igneous emplacement, deformation, and metamorphism.
Although chalcopyrite is often concentrated along the contacts between massive sulfide horizons and metasedimentary country rocks during deformation, there does not appear to have been significant loss of a fractionated Cu liquid at the magmatic stage. This is indicated by the high Ni and low Cu contents of the original sulfide melt, and thus most of the Cu could have dissolved in MSS because of the moderate to high temperatures reached during deformation (e.g., Naldrett, 2004). However, as discussed below, there is geochemical evidence that Au, Cu, Pt, and As were mobile in metamorphic fluids during metamorphism and deformation, as suggested by Chen et al. (1993) from late, postmagmatic, native gold in association with gersdorffite and nickeline from the Thompson deposit.
The geochemistry of the ores in the Thompson nickel belt has been discussed by Campbell and Naldrett (1979), Bleeker (1990a), and Lesher and Keays (2002). Eckstrand et al. (1989) showed that S/Se ratios of Thompson ores (5,900–35,700) vary between those of the associated sulfide-facies iron formations This suggests that the ores were generated by melting of (12,800–35,700) and mantle sulfides (2,900–4,400). crustal S from the iron formations, similar to most other Ni-Cu-(PGE) deposits (Lesher and Keays, 2002).
The metal, semimetal, and PGE geochemistry of the ores in the Thompson nickel belt, compiled by Lesher and Keays (2002) and Layton-Matthews et al. (unpub. data), are summarized as follows:
The chalcophile element compositions of many of the ores are consistent with interaction of a sulfide melt generated by melting of sulfide-rich iron formations and a low Mg komatiitic magma of the composition inferred from the whole-rock geochemistry.
Magma/sulfide mass ratios (R factors) range from 20 to 30 for Pipe 2 massive ores, 30 to 200 for Birchtree massive ores, and 100 to 200 for Thompson massive ores, to 400 to 500 for Bucko disseminated ores, and 40 to 600 for Pipe 2 disseminated ores. Thus, the intrusions in the Thompson nickel belt had the capacity to generate higher tenor ores if the sulfides could have interacted with greater amounts of magma.
Sedimentary rock-hosted ores have Ni-Co-Cu-Rh-Pd-Au abundances that are broadly similar to those of the ultramafic rock-hosted ores but are strongly depleted in Cr, Os, Ir, Ru, and Pt. Although only useful over relatively short distances of no more than 20 m, such fractionation may provide a vector toward magmatic mineralization.
Copper, Au, and Pt have been mobilized for as far as 100 m from massive sulfides into adjacent country rocks during deformation and metamorphism. The majority of the metals appears to have remained within the current mining widths and are being recovered, but the greater mobilities of Cu, Au, and Pt may allow them to be used to vector toward mineralization.
Volcanic architecture and source characteristics of mafic and ultramafic rocks
Mass balance suggests that the ultramafic intrusions represent cumulates formed from larger volumes of ultramafic magmas that must have subsequently either erupted as volcanic rocks or were emplaced as subvolcanic sills higher in the sequence. The volume of the missing magma can be estimated from the volume of the ultramafic intrusions and their average compositions by mass balance:
where Vd = volume of evolved lava produced, Ve = volume of zone of enrichment (i.e., intrusion), Coi = concentration of element i in initial (unevolved) magma, Cdi = concentration of element i in evolved magma (missing), and Cei = concentration of element i in cumulates.
If this is applied to the MgO contents of the ultramafic bodies, then assuming an average olivine composition of Fo89 (CeMgO = 48.6 wt %), ˜21 wt percent MgO in the initial magma, and a minimum of ˜10 wt percent MgO in the evolved magma, Vd/Ve = 2.5. That is, the volume of the mafic material processed through the sills must have been at least two to three times that of the ultramafic bodies. If the evolved magmas were more magnesian, which is almost certainly the case for dunitic bodies containing few, if any, evolved lithologic units, then this value would be much higher. Although there are a few crustally contaminated and chalcophile element-depleted rocks in the Bah Lake Formation and a few uncontaminated and chalcophile element-depleted rocks in the ultramafic sills, the majority of the rocks in the Bah Lake Formation are uncontaminated and undepleted in chalcophile elements, indicating that they could not have been derived from the crustally contaminated and locally mineralized residues represented by the ultramafic sills (Lesher et al., 2001; Burnham et al., 2003; Layton-Matthews et al., 2007). Thus, the volume of magma missing from the sills has not been identified.
The degree of melting and the eruption temperature of an ultramafic magma significantly influence its chalcophile metal contents, S saturation state, both within the source region and during movement to the surface, and ability to incorporate external sulfur. The higher degrees of melting and higher temperatures inferred in this study for the Thompson nickel belt ultramafic magmas would have been important factors influencing the metallogenesis of the associated sulfide ores.
Because of the high degrees of melting, the Thompson nickel belt magmas would have dissolved all of the sulfide phase in the mantle source, leading to the enrichment of the magmas in both the moderately and highly chalcophile metals. These melts would also have been far from sulfide saturation during most of their transportation and emplacement in the crust, despite country-rock assimilation and/or fractionation of olivine. Because they may have been significantly more sulfide undersaturated than the magmas that erupted in other parts of the Circum-Superior boundary zone, they would have been less susceptible to sulfide loss and would have been better able to retain their metals until they intersected an external source of sulfur.
As a consequence of their higher temperatures, the ultramafic magmas of the Thompson nickel belt would have also had higher heat capacities and lower viscosities than less magnesian liquids formed by lower degrees of melting. Thermomechanical erosion of the country rocks is most effective when the magmas are flowing turbulently, either within lava channels (Huppert and Sparks, 1984, 1985b; Lesher et al., 1984; Williams et al., 1998), along magma conduits (Huppert and Sparks, 1988), or in channelized sills. This results in significant cooling of the magma (Huppert and Sparks, 1985a, b), such that the higher heat capacities and lower viscosities of the Thompson nickel belt ultramafic magmas would have significantly enhanced their capacity for bulk assimilation of sedimentary rocks of the Ospwagan Group, in particular the S-rich iron formations, leading to the production of greater quantities of immiscible sulfide liquid to form the ores. Whereas this would be an advantage in most environments where the ability to form a deposit is limited by the supply of sulfide, in the Thompson nickel belt this may have actually lead to the production of greater tonnage but lower grade deposits, owing to the dilution of the metals in a greater mass of sulfide liquid.
There is no clear evidence for the mode of emplacement (e.g., replacement vs. dilation) of the mineralized and unmineralized ultramafic intrusions in the Thompson nickel belt. Most transgressive contacts, synemplacement deformations of country rocks, or local contact metamorphic effects appear to have been destroyed by subsequent deformation and/or regional metamorphism. There is no direct physical record of any assimilation processes in the ultramafic intrusions in the Thompson nickel belt, in the form of xenoliths, xenocrysts, silicate xenomelts, or restites of partial melting. The Ni-Cu-(PGE) ores are interpreted to represent sulfide xenomelts, but they have not retained any textural or geochemical characteristics supporting this interpretation.
However, there is indirect chemical evidence of contamination and the composition of the contaminant is consistent with it being sedimentary rocks of the Ospwagan Group. Importantly, however, the trace element compositions of both mineralized and unmineralized intrusions can suggest as much as 40 percent contamination, indicating that contamination alone is insufficient to induce sulfide saturation and produce Ni-Cu-(PGE) mineralization in komatiitic magmas, as discussed by Lesher and Stone (1996) and Lesher and Keays (2002).
MgO contents of intrusions and olivine composition
Mineralized intrusions of the Thompson nickel belt are generally more magnesian than unmineralized intrusions. This is manifested mainly in higher olivine contents, rather than by the more forsteritic olivines. This means that the association between mineralization and higher Mg contents is related mainly to the degree of olivine accumulation and not the temperature of the magma. Because magmas containing more than ˜50 percent phenocrysts are too dense and/or too viscous to be erupted (Marsh, 1989; Costa, 2005), much, if not all, of the olivine crystallized in situ and the higher olivine contents reflect more dynamic systems that may have been able to more effectively melt country rocks to form molten sulfides. Thus, the duration and volume of magma flow is more important than the parental magma composition, as observed in many examples of deposits hosted by more differentiated intrusions (e.g., Noril’sk, Pechenga).
Many mineralized intrusions (e.g., Pipe, Mystery Lake) contain olivine that is depleted in Ni, but some unmineralized intrusions contain olivine that is depleted in Ni (e.g., Mid Lake) and some mineralized intrusions contain olivine that is not depleted in Ni (e.g., Thompson, Pipe). The presence of both Ni-depleted and normal Ni olivine in mineralized ultramafic intrusions indicates Ni depletion and, therefore, sulfide segregation must have occurred locally. If all of the olivines were depleted, then Ni may have been held back in sulfides in the source and/or sulfides may have segregated during transport or in a deeper staging chamber.
Trace element compositions of ultramafic intrusions
Mineralized ultramafic units in the Thompson nickel belt are enriched in HILE relative to MILE and are depleted in Nb-Ta-(Ti) relative to HILE of similar compatibility. Therefore, they exhibit higher Th/Yb and Th/Nb ratios than unmineralized ultramafic units. However, there is no systematic correlation between the grade of mineralization and amount contamination in deposits in the Thompson nickel belt or elsewhere (e.g., Barnes et al., 1985; Lesher and Arndt, 1995; Perring et al., 1996; Lesher et al., 2001). The reason for this is that the amount of contamination associated with the mineralizing process depends on several factors (Lesher et al., 2001), including the following: (1) the stratigraphic architecture of the system, defined by thickness and physical accessibility of the contaminant; (2) the fluid dynamics and thermodynamics of the lava and/or magma; (3) the physical, chemical, and thermal characteristics of the contaminant; (4) the amount of contaminant melted and incorporated; (5) the S and metal content of the contaminant; (6) the initial sulfide saturation state of the magma; (7) the assimilation/crystallization ratio; (8) the amount of lava replenishment; and (9) the effective magma/sulfide ratio (R factor) of the system. Because these processes vary independently from deposit to deposit, from area to area within a deposit, and within a single area with time, there are many opportunities to decouple mineralization from contamination (Lesher and Arndt, 1995; Lesher and Stone, 1996; Lesher et al., 2001).
Resource Implications and Exploration Guides
Hoatson et al. (2005) noted that individual deposits in the Thompson nickel belt are generally larger than those in the Eastern Goldfields and Southern Cross provinces of Western Australia or the Abitibi belt of Ontario. They suggested that intense deformation may have generated larger and fewer deposits by mobilization of the smaller deposits in the Thompson nickel belt during tectonic events, but there is no evidence that deformation has agglomerated smaller deposits. Hoatson et al. (2005) also suggested that the greater sizes may reflect the difference between deposits formed in intrusive (Thompson) versus volcanic (Eastern Goldfields) environments, which is possible given that intrusive systems will lose heat less rapidly and may, with all else equal, incorporate more S from country rocks. However, it may simply reflect the much greater abundance of S in the iron formations in the Thompson nickel belt compared to the S available in the country rocks at most other deposits of this type.
The guides to exploring for Ni-Cu-(PGE) mineralization in the Thompson nickel belt can be summarized as follows: recognition of the P1 and P2 members of the Pipe Formation, which represent the locus of intrusion of ultramafic sills, which are the heat and metal source, and sulfide-facies iron formation, which is the S source; (2) close association with ultramafic sills, which represent both the principal metal source and a source of heat for generating sulfide melt from the country rocks: (3) more magnesian and more contaminated dunites, which represent areas of greater magma flux, where there is more heat for melting of country rock sulfides and higher R factors; (4) dunites exhibiting evidence for local depletion of Ni in olivine; (5) fold noses where ductile Fe-Ni-Cu sulfides may be concentrated by deformation; (6) late and retrograde shear zones, where deformation may have persisted because of the greater ductility of massive Fe-Ni-Cu sulfides; if primary, they may also represent loci for magma emplacement and subsequent deformation; and (7) metal mobilization haloes with metal fractionation in the order Au > Cu >> Pt >> Ni >> Pd-Rh-Ru-Co-Ir >> Cr.
Our research on the Thompson nickel belt has been supported by grants from Canadian Mining Industry Research Organization (94-EO4 and 97E-02) and the Natural Sciences and Engineering Research Council of Canada (IRC 663-001-97, DG 203171-98, DG 203171-02, DG 203171-07, DG 342440-07). This study would not have been possible without the support of Inco Limited, Falconbridge Limited, Teck-Cominco Limited, Hudson Bay Exploration and Development Company Limited, BHP-Billiton Limited, and WMC Limited, who provided access to their mining properties, drill core, and data. We are very grateful to J. Pearson of Cominco; J. Robertson, P. Tirschmann, and K. Wells of Falconbridge; L. Lutz, H. Mahoney, S. Mooney, D. Owens, R. Sommerville, G. Sorensen, and R. Stewart of Inco; and J. Pickell of Hudson Bay for assistance with the field work and for many insightful discussions on the geology of the Thompson nickel belt. We are very grateful to our collaborators, J. Liwanag, N. Halden, and D. Michalak of the University of Manitoba; D. Gapais of Université de Rennes; N. Machado (deceased) and A. Potrel of Université du Québec à Montreal; L. Heaman and K. Toope of the University of Alberta; K. Ansdell of the University of Saskatchewan; C. Chandler, E. Ducharme, and C. Freund of Brandon University; T. Corkery, P. Lenton, M. Pacey, P. Theyer, and H. Zwanzig of the Manitoba Geological Survey; and J. Krause and W. Bleeker of the Geological Survey of Canada for their contributions to the CAMIRO Thompson nickel belt project and for many beneficial discussions. The manuscript greatly benefited from the constructive comments of R. Sproule, E. Marsh, and R. Goldfarb.
Figures & Tables
The Challenge of Finding New Mineral Resources: Global Metallogeny, Innovative Exploration, and New Discoveries
VOLCANIC-ASSOCIATED and sedimentary-exhalative massive sulfide deposits on land account for more than one-half of the world's total past production and current reserves of zinc and lead, 7 percent of the copper, 18 percent of the silver, and a significant amount of gold and other by-product metals (Singer, 1995). A new source of these metals is now being considered for exploitation from deep-sea massive sulfide deposits. Because the oceans cover more than 70 percent of the Earth's surface, many expect the ocean floor to host a proportionately large number of these deposits. However, there have been few attempts to estimate the global mineral potential. Significant accumulations of metals from hydrothermal vents have been documented at some locations (e.g., 91.7 Mt of 2.06% Zn, 0.46% Cu, 58.5 g/t Co, 40.95 g/t Ag, and 0.51 g/t Au in the Atlantis II Deep of the Red Sea: Mustafa et al., 1984; Nawab, 1984; Guney et al., 1988). Even more metal is contained in deep-sea manganese nodules. Current estimates in the U.S. Geological Survey (USGS) mineral commodities summaries indicate a global resource of copper in deep-sea nodules of about 700 Mt. In the Pacific "high-grade" area, an estimated 34,000 Mt of nodules contain 7,500 Mt of Mn, 340 Mt of Ni, 265 Mt of Cu, and 78 Mt of Co (Morgan, 2000; Rona, 2003). A number of countries, including China, Japan, Korea, Russia, France, and Germany, are actively exploring this area.